Lecture Notes in Physics Editorial Board R. Beig, Wien, Austria W. Beiglböck, Heidelberg, Germany W. Domcke, Garching, ...

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Lecture Notes in Physics Editorial Board R. Beig, Wien, Austria W. Beiglböck, Heidelberg, Germany W. Domcke, Garching, Germany B.-G. Englert, Singapore U. Frisch, Nice, France P. Hänggi, Augsburg, Germany G. Hasinger, Garching, Germany K. Hepp, Zürich, Switzerland W. Hillebrandt, Garching, Germany D. Imboden, Zürich, Switzerland R. L. Jaffe, Cambridge, MA, USA R. Lipowsky, Golm, Germany H. v. Löhneysen, Karlsruhe, Germany I. Ojima, Kyoto, Japan D. Sornette, Nice, France, and Los Angeles, CA, USA S. Theisen, Golm, Germany W. Weise, Garching, Germany J. Wess, München, Germany J. Zittartz, Köln, Germany

The Lecture Notes in Physics The series Lecture Notes in Physics (LNP), founded in 1969, reports new developments in physics research and teaching – quickly and informally, but with a high quality and the explicit aim to summarize and communicate current knowledge in an accessible way. Books published in this series are conceived as bridging material between advanced graduate textbooks and the forefront of research to serve the following purposes: • to be a compact and modern up-to-date source of reference on a well-deﬁned topic; • to serve as an accessible introduction to the ﬁeld to postgraduate students and nonspecialist researchers from related areas; • to be a source of advanced teaching material for specialized seminars, courses and schools. Both monographs and multi-author volumes will be considered for publication. Edited volumes should, however, consist of a very limited number of contributions only. Proceedings will not be considered for LNP. Volumes published in LNP are disseminated both in print and in electronic formats, the electronic archive is available at springerlink.com. The series content is indexed, abstracted and referenced by many abstracting and information services, bibliographic networks, subscription agencies, library networks, and consortia. Proposals should be sent to a member of the Editorial Board, or directly to the managing editor at Springer: Dr. Christian Caron Springer Heidelberg Physics Editorial Department I Tiergartenstrasse 17 69121 Heidelberg/Germany [email protected]

James W. LaBelle Rudolf A. Treumann (Eds.)

Geospace Electromagnetic Waves and Radiation

ABC

Editors Professor James W. LaBelle Department of Physics Wilder Laboratory 6127 Dartmouth College Hanover, NH 03755 USA E-mail: [email protected]

Professor Rudolf A. Treumann Universität München Sektion Geophysik Theresienstraße 41 80333 München Germany E-mail: [email protected]

J.W. LaBelle and R.A. Treumann, Geospace Electromagnetic Waves and Radiation, Lect. Notes Phys. 687 (Springer, Berlin Heidelberg 2006), DOI 10.1007/b11580119

Library of Congress Control Number: 2005937155 ISSN 0075-8450 ISBN-10 3-540-30050-3 Springer Berlin Heidelberg New York ISBN-13 978-3-540-30050-2 Springer Berlin Heidelberg New York This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, speciﬁcally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microﬁlm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable for prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media springer.com c Springer-Verlag Berlin Heidelberg 2006 Printed in The Netherlands The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a speciﬁc statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Typesetting: by the authors and TechBooks using a Springer LATEX macro package Printed on acid-free paper

SPIN: 11580119

54/TechBooks

543210

Preface

The “Ringberg Workshop on High Frequency Waves in Geospace” convened at Ringberg Castle, Bavaria, from July 11 to 14, 2004. Approximately 30 attendees from 11 countries gathered at the castle for a program of invited talks and posters focussed on outstanding problems in high-frequency waves, deﬁned broadly as waves exceeding a few kHz in frequency. Thirteen invited presentations comprise the contents of this volume. These articles provide introductions to current problems in geospace electromagnetic radiation, guides to the associated literature, and tutorial reviews of the relevant space physics. As such, this volume should be of value to students and researchers in electromagnetic wave propagation in the environment of the Earth at altitudes above the neutral atmosphere, extending from the ionosphere into outer space. The contributions are broadly grouped into three parts. Part I, entitled “High-Frequency Radiation” focusses on radiation processes in near-Earth plasmas. Benson et al. present a tutorial review of Z-mode emissions, which so far have received relatively little attention and are the subject of few such reviews despite their abundant presence in geospace. Hashimoto et al. continue with another tutorial review on the terrestrial continuum radiation, the relatively weak radio emissions that ﬁll the entire outer magnetosphere and provide information about the magnetospheric plasma boundaries and the state of the magnetospheric plasma density. Louarn reviews the ideas relevant to the generation of Auroral Kilometric Radiation (AKR), by far the most powerful and signiﬁcant of the high-frequency radiations in the magnetosphere. Fleishman introduces the topic of diﬀusive synchrotron radiation, a mechanism not widely appreciated by geophysicists, but which may play a role in several magnetospheric, heliospheric, and even astrophysical settings. Pottelette and Treumann end this chapter with a discussion of the latest ideas about the relationship between auroral acceleration processes and radiation processes such as AKR, a subject which has been transformed in the last decade due to observations with the FAST and CLUSTER satellites. Part II of this monograph, entitled “High-frequency waves,” focusses on wave physics. Sonwalkar presents a lengthy and comprehensive review of

VI

Preface

whistler-mode propagation in the presence of density irregularities. James’ paper deals with recent results from the OEDIPUS-C sounding rocket, combined with recent innovations in antenna theory, which lead to the provocative but signiﬁcant conclusion that ﬁeld strengths measured by many previous observations of auroral hiss using dipole antennas may need to be revised downward. Lee et al. present a novel theoretical method for analyzing modecoupling and mode-conversion of high-frequency waves, with applications to geophysical plasmas. Yoon et al. treat the subject of mode-conversion radiations, which are replete in the Earth’s environment, both in the ionosphere, magnetosphere, and solar wind. Vaivads et al. conclude this part with a review of high-frequency waves related to magnetic reconnection as the generator region of high-frequency waves and radiation in geospace, a very important and hot topic, especially in the light of recent CLUSTER satellite observations. Part III of the monograph is devoted to new analysis techniques and instrumentation transforming research on high-frequency waves. P´ecseli and Trulsen discuss novel ideas on the forefront of linking wave observations to theoretical models. Santol´ık and Parrot apply sophisticated wave propagation analysis tools to the study of AKR. Finally, Kletzing and Muschietti discuss wave particle correlators, describing the physics that can be investigated with them and including results from a recent state-of-the-art wave-particle correlator ﬂown in the Earth’s auroral ionosphere. This monograph would not have been possible without the assistance of the many referees. Special thanks are due to M. Andr´e, R.E. Ergun, J.R. Johnson, E.V. Mishin, R. Pottelette, O. Santol´ık, V.S. Sonwalkar, and A.T. Weatherwax. We thank Dr. Axel H¨ ormann and his team for creating the gracious, welcoming environment at Ringberg Castle, which allowed a creative workshop to take place and thereby inspired this volume. We also thank the International Space Science Institute Bern for support. Finally, the editors at Springer, especially Dr. Christian Caron, deserve thanks for supporting the timely publication of this work and helping to assure its high quality.

Hanover, New Hampshire, and Munich June 2005

James LaBelle Rudolf Treumann

Contents

Part I

High-Frequency Radiation

1 Active Wave Experiments in Space Plasmas: The Z Mode R.F. Benson, P.A. Webb, J.L. Green, D.L. Carpenter, V.S. Sonwalkar, H.G. James, B.W. Reinisch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Plasma Wave Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Sounder-Stimulated Z-Mode Waves in the Topside Ionosphere . . . . . 1.3.1 Single Spacecraft: Vertical and Oblique Propagation . . . . . . . 1.3.2 Single Spacecraft: Ducted Propagation . . . . . . . . . . . . . . . . . . . 1.3.3 Single Spacecraft: Wave Scattering . . . . . . . . . . . . . . . . . . . . . . 1.3.4 Dual Payloads: Slow Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.5 Dual Payloads: Fast Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.6 Possible Role of Z-Mode Waves in Sounder/Plasma Interactions . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Sounder-Stimulated Z-Mode Waves in the Magnetosphere . . . . . . . . 1.4.1 Remote O-Z-O-Mode Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2 Local Z-Mode Echoes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.3 Z-Mode Refractive-Index Cavities . . . . . . . . . . . . . . . . . . . . . . . 1.4.4 Whistler- and Z-Mode Echoes . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Active/Passive Investigation of Z-Mode Waves of Magnetospheric Origin . . . . . . . . . . . . . . . . . . . . . 1.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Review of Kilometric Continuum K. Hashimoto, J.L. Green, R.R. Anderson, H. Matsumoto . . . . . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Trapped and Escaping NTC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Continuum Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Generation Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 4 5 10 10 12 15 16 18 19 22 22 22 24 28 29 31 31 37 37 39 40 41

VIII

Contents

2.4 Kilometric Continuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Geomagnetic Activity Dependence of Kilometric Continuum . . . . . . 2.6 Simultaneous Wave Observations by Geotail and IMAGE . . . . . . . . 2.7 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Generation of Auroral Kilometric Radiation in Bounded Source Regions P. Louarn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Spacecraft Observations in the Sources of the Auroral Kilometric Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Structure of Sources and Wave Properties . . . . . . . . . . . . . . . . 3.2.2 AKR Sources as Regions of Particle Accelerations . . . . . . . . 3.2.3 AKR Sources as Plasma Cavities . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4 Free Energy for the Maser Process . . . . . . . . . . . . . . . . . . . . . . 3.2.5 FAST Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Cyclotron Maser in Finite Geometry Sources . . . . . . . . . . . . . . . . . . . 3.3.1 A Simple Model of AKR Sources . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Generation in Plasma Cavities: A Simple Approach . . . . . . . . 3.3.3 Observations of Finite Geometry Eﬀects . . . . . . . . . . . . . . . . . 3.4 The Cyclotron Maser in Finite Geometry . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 Mathematical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 Solutions of the Dispersion Relation . . . . . . . . . . . . . . . . . . . . . 3.4.3 The Problem of Energy Escape . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.1 O and X Mode Production in Narrow Sources . . . . . . . . . . . . 3.5.2 Fine Structure of Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.3 Conclusion and Pending Questions . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Generation of Emissions by Fast Particles in Stochastic Media G.D. Fleishman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Statistical Methods in the Theory of Electromagnetic Emission . . . 4.2.1 Spectral Treatment of Random Fields . . . . . . . . . . . . . . . . . . . . 4.2.2 Emission from a Particle Moving along a Stochastic Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3 Kinetic Equation in the Presence of Random Fields . . . . . . . . 4.2.4 Solution of Kinetic Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Emission from Relativistic Particles in the Presence of Random Magnetic Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 General Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

42 46 49 50 51

55 55 57 57 59 59 61 63 64 65 65 66 68 74 74 78 79 82 82 82 83 83

87 87 88 88 89 90 92 94 94

Contents

IX

4.3.2 Special Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.3.3 Emission from an Ensemble of Particles . . . . . . . . . . . . . . . . . . 97 4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 5 Auroral Acceleration and Radiation R. Pottelette, R.A. Treumann . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 5.2 Morphology of the Auroral Acceleration Region . . . . . . . . . . . . . . . . . 107 5.2.1 Upward Current Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 5.2.2 Downward Current Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 5.3 Parallel Electric Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 5.3.1 Electrostatic Double Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 5.3.2 Microscopic Structures – Phase Space Holes . . . . . . . . . . . . . . 120 5.3.3 Relation between Holes and Double Layers . . . . . . . . . . . . . . . 125 5.3.4 Holes in the Upward Current Region . . . . . . . . . . . . . . . . . . . . 128 5.4 Auroral Kilometric Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 5.5 The Tail Acceleration Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 5.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 Part II

High-Frequency Waves

6 The Inﬂuence of Plasma Density Irregularities on Whistler-Mode Wave Propagation V.S. Sonwalkar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 6.2 Magnetospheric Plasma Distribution: Field Aligned Irregularities . . 143 6.3 Propagation of Whistler-Mode Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 147 6.3.1 Propagation of Plane Whistler-Mode Waves in Uniform Magnetoplasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 6.3.2 Propagation of Whistler-Mode Waves in the Magnetosphere 151 6.4 Observations and Interpretations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 6.4.1 W-Mode Observations When the Source is Below the Ionosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 6.4.2 W-Mode Observations When the Source is in the Magnetosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 6.5 Discussion and Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 7 Dipole Measurements of Waves in the Ionosphere H.G. James . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 7.2 Summary of OEDIPUS-C Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 7.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

X

Contents

7.2.2 Whistler-Mode Propagation Near the Resonance Cone . . . . . 197 Retrospective on Past Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 7.3.1 Dipole Eﬀective Length for EM Propagation . . . . . . . . . . . . . . 199 7.3.2 Intensity of Auroral Hiss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 7.3.3 Intersatellite Whistler-Mode Propagation Near a Resonance Cone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 7.3.4 Intersatellite Slow-Z-Mode Propagation Near a Resonance Cone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209 7.3

8 Mode Conversion Radiation in the Terrestrial Ionosphere and Magnetosphere P.H. Yoon, J. LaBelle, A.T. Weatherwax, M. Samara . . . . . . . . . . . . . . . . 211 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 8.2 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 8.2.1 Auroral-Zone Mode-Conversion Radiation . . . . . . . . . . . . . . . . 213 8.2.2 Magnetospheric Mode-Conversion Radiation . . . . . . . . . . . . . . 218 8.3 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 8.3.1 Nonlocal Theory of Electrostatic Waves in Density Structures222 8.3.2 Trapped Langmuir Waves with Discrete Frequency Spectrum225 8.3.3 Discrete Upper-Hybrid Waves in Density Structures . . . . . . . 228 8.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 9 Theoretical Studies of Plasma Wave Coupling: A New Approach D.-H. Lee, K. Kim, E.-H. Kim, K.-S. Kim . . . . . . . . . . . . . . . . . . . . . . . . . . 235 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 9.2 Numerical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 9.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 9.4 Invariant Embedding Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 9.5 Application of IEM to the Mode-Conversion of O and X Waves . . . 243 9.6 Discussion and Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 10 Plasma Waves Near Reconnection Sites A. Vaivads, Yu. Khotyaintsev, M. Andr´e, R.A. Treumann . . . . . . . . . . . . . 251 10.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 10.1.1 Observations of Diﬀerent Wave Modes . . . . . . . . . . . . . . . . . . . 253 10.2 Numerical Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 10.3 Lower Hybrid Drift Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 10.3.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 10.3.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 10.3.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 10.4 Solitary Waves and Langmuir/Upper Hybrid Waves . . . . . . . . . . . . . 259

Contents

XI

10.4.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 10.4.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 10.4.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 10.5 Whistlers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.5.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.5.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.5.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.6 Electron Cyclotron Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 10.6.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 10.6.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 10.6.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 10.7 Free Space Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 10.7.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 10.7.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 10.7.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 10.8 Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 Part III High-Frequency Analysis Techniques and Wave Instrumentation 11 Tests of Time Evolutions in Deterministic Models, by Random Sampling of Space Plasma Phenomena H.L. P´ecseli, J. Trulsen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 11.2 Model Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 11.2.1 Spatial Sampling with One Probe Available . . . . . . . . . . . . . . 277 11.2.2 Spatial Sampling with Two Probes Available . . . . . . . . . . . . . 282 11.2.3 Temporal Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 11.3 Nonlinear Lower-Hybrid Wave Models . . . . . . . . . . . . . . . . . . . . . . . . . 283 11.4 Probability Densities for Observables . . . . . . . . . . . . . . . . . . . . . . . . . . 286 11.5 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 11.6 Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292 11.6.1 Length Scales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 11.6.2 Time Scales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294 11.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 12 Propagation Analysis of Electromagnetic Waves: Application to Auroral Kilometric Radiation O. Santol´ık, M. Parrot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 12.2 Analysis Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 12.2.1 Plane Wave Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 12.2.2 Wave Distribution Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304

XII

Contents

12.3 Analysis of Auroral Kilometric Radiation . . . . . . . . . . . . . . . . . . . . . . . 305 12.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 13 Phase Correlation of Electrons and Langmuir Waves C.A. Kletzing, L. Muschietti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 13.2 Finite-Size Wave Packet in a Vlasov Plasma . . . . . . . . . . . . . . . . . . . . 316 13.2.1 Linear Perturbation of the Electrons . . . . . . . . . . . . . . . . . . . . . 317 13.2.2 Case of a Gaussian Packet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 13.3 Extended Wave Packet: A BGK Analysis . . . . . . . . . . . . . . . . . . . . . . . 322 13.3.1 Passing Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 13.3.2 Trapped Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 13.4 Electron Phase Sorting Measurements . . . . . . . . . . . . . . . . . . . . . . . . . 324 13.4.1 Measurements of the Resistive Component . . . . . . . . . . . . . . . 325 13.4.2 Measurements of the Reactive Component . . . . . . . . . . . . . . . . 328 13.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 13.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

List of Contributors

Allan T. Weatherwax Siena College, Dept. Physics 515 Loudon Rd. Loudonville, NY 12211-1462, USA [email protected] Andris Vaivads Swedish Institute of Space Physics Box 537, Uppsala, SE 751 21, Sweden [email protected] Bodo W. Reinisch University of Massachusetts Lowell, MA 01854-0000, USA [email protected] Craig Kletzing University of Iowa/Phys. & Astronomy Iowa City, IA 52242-0000, USA [email protected] Donald L. Carpenter Stanford University Star Lab. Electr. Engng. Dept. Stanford, CA 94305-9515 [email protected] D.-H. Lee Kyung Hee University Dept. Astron. Space Sci. Yongin, Kyunggi 449-701, Korea [email protected]

E.-H. Kim Kyung Hee University Dept. Astron. Space Sci. Yongin, Kyunggi 449-701, Korea [email protected] Gregory D. Fleishman 26 Polytekhnicheskaya St. Petersburg 194021 Russian Federation [email protected] Hans L. Pecseli University of Oslo/Inst. Physics Box 1048, Blindern, N-0316 Oslo, Norway [email protected] Hiroshi Matsumoto Res. Inst. Sustainable Humanosphere Kyoto University Uji, Kyoto 611-0011, Japan [email protected] H. Gordon James Communications Research Center Canada Ottawa, Ontario KH2 882, Canada [email protected]

XIV

List of Contributors

James LaBelle Dartmouth College Dept. Phys. Astron. Wilder Laboratory Hanover, New Hampshire 03755 USA [email protected] James L. Green NASA/Goddard Space Flight Center MC 630, Bldg. 26, Greenbelt, MD 20771-1000, USA [email protected] Jan Trulsen University of Oslo Institute of Theoretical Astrophysics Box 1029 Blindern N-0315 Oslo, Norway [email protected] Kozo Hashimoto Res. Inst. Sustainable Humanosphere Kyoto University Uji, Kyoto 611-0011, Japan [email protected] K. Kim Ajou University Dept. Molecular Sci. Techn. Suwon, Kyunggi 443-749, Korea [email protected] K.-S. Kim Kyung Hee University Dept. Astron. Space Sci. Yongin, Kyunggi 449-701, Korea [email protected] Laurent Muschietti University of California Space Sciences Laboratory Berkeley, CA 94720-7450, USA [email protected]

Marilia Samara Dartmouth College Dept. Phys. Astron., Wilder Laboratory Hanover, NH 03755, USA [email protected] Mats Andr´ e Swedish Institute of Space Physics Box 537, Uppsala, SE 751 21 [email protected] Michel Parrot LPCE/CNRS 3a av. de Recherche Science 45071 Orleans, Cedex 02, France [email protected] Ond˘ rej Santol`ık Charles Univ. Prague Faculty of Math. & Phys. Prague 8, CZ-18000 Czech Republic [email protected] Peter H. Yoon University of Maryland Inst. Physical Sci. & Techn. College Park, MD 20742, USA [email protected] Philippe Louarn CNRS/CESR 9 av. Colonel Roche Toulouse, 31329 France [email protected] Phillip A. Webb NASA/GSFC Code 612.3 Greenbelt, MD 20771-0000, USA [email protected] Raymond Pottelette CNRS/CETP 4 av. de Neptune St. Maur des Foss´es Cedex, France [email protected]

List of Contributors

Robert F. Benson NASA/Goddard Space Flight Center MC 692, Bldg. 21, Room 252 Greenbelt, MD 20771-0001, USA [email protected] Roger R. Anderson University of Iowa Dept. Phys. & Astron. Iowa City, IA 52242-1479, USA [email protected] Rudolf A. Treumann Ludwig-Maximilians Universit¨ at M¨ unchen, Sektion Geophysik Theresienstr. 41, 80333 Munich Germany [email protected]

XV

Vikas S. Sonwalkar Univ. Alaska Fairbanks Dept. Electr. Engng. 306 Tanana Dr. Room 229 Duckering Fairbanks, AK 99775-0000, USA [email protected] Yuri V. Khotyaintsev Swedish Inst. Space Physics Box 537, Uppsala 75121, Sweden [email protected]

Part I

High-Frequency Radiation

1 Active Wave Experiments in Space Plasmas: The Z Mode R.F. Benson1 , P.A. Webb1 , J.L. Green1 , D.L. Carpenter2 , V.S. Sonwalkar3 , H.G. James4 , and B.W. Reinisch5 1

2 3 4 5

NASA/Goddard Space Flight Center, Greenbelt, MD USA [email protected] Stanford University, California USA University of Alaska Fairbanks, Alaska USA Communications Research Centre, Toronto, Canada University of Massachusetts Lowell, Massachusetts USA

Abstract. The term Z mode is space physics notation for the low-frequency branch of the extraordinary (X) mode. It is an internal, or trapped, mode of the plasma conﬁned in frequency between the cutoﬀ frequency fz and the upper-hybrid frequency fuh which is related to the electron plasma frequency fpe and the electron 2 2 2 = fpe + fce ; fz is a function of fpe cyclotron frequency fce by the expression fuh and fce . These characteristic frequencies are directly related to the electron number density Ne and the magnetic ﬁeld strength |B|, i.e., fpe (kHz)2 ≈ 80.6Ne (cm−3 ) and fce (kHz)2 ≈ 0.028|B|(nT). The Z mode is further classiﬁed as slow or fast depending on whether the phase velocity is lower or higher than the speed of light in vacuum. The Z mode provides a link between the short wavelength λ (large wave number k = 2π/λ ) electrostatic (es) domain and the long λ (small k) electromagnetic (em) domain. An understanding of the generation, propagation and reception of Z-mode waves in space plasma leads to fundamental information on wave/particle interactions, Ne , and ﬁeld-aligned Ne irregularities (FAI) in both active and passive wave experiments. Here we review Z-mode observations and their interpretations from both radio sounders on rockets and satellites and from plasma-wave receivers on satellites. The emphasis will be on the scattering and ducting of sounder-generated Z-mode waves by FAI and on the passive reception of Z-mode waves generated by natural processes such as Cherenkov and cyclotron emission. The diagnostic applications of the observations to understanding ionospheric and magnetospheric plasma processes and structures beneﬁt from the complementary nature of passive and active plasma-wave experiments.

Key words: Auroral kilometric radiation, Z-mode, free space radiation, wave transformation, radiation escape, cavity modes, active experiments

R.F. Benson et al.: Active Wave Experiments in Space Plasmas: The Z Mode, Lect. Notes Phys. 687, 3–35 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

4

R.F. Benson et al.

1.1 Introduction According to cold plasma theory, at high frequencies there are two characteristic electromagnetic (em) waves, or modes, that can propagate in a magnetoplasma. They are often referred to as the free-space ordinary (O) and extraordinary (X) modes because waves propagating in these modes can smoothly connect to free space. The X mode has two branches. In addition to the freespace mode, it has a mode called the slow branch. This name is used because it is restricted to propagation velocities less than the vacuum speed of light c. Since this mode only exists within a plasma, there was considerable interest in explaining observations indicating that it was responsible for a unique signature on early ground-based radars designed to probe the ionosphere. In their most common application these radars, called ionosondes, operate by transmitting a radio pulse of short time duration at a particular frequency and receiving, at the same frequency, for a time interval suﬃcient to receive an echo from the ionosphere overhead. This process is repeated over a range of frequencies likely to produce reﬂections. The resulting record is called an ionogram. Normally, there are two ionospheric reﬂections, one due to the O mode and one due to the X mode. “On rare occasions”, as ﬁrst reported by Eckersley [23], there is a third reﬂection with the same polarization as the O mode. This third reﬂection trace, corresponding to the slow X-mode branch, was dubbed the Z mode in ionospheric research; a designation commonly used in space physics. In order to explain the presence of the Z mode at ground level, i.e., far below the ionospheric plasma, and the polarization (same as the O mode), a Z-O mode coupling process involving obliquely-propagating Omode waves was introduced by Ellis [24] as discussed in Sect. 13.5 of Ratcliﬀe [57]. An ionogram showing this triple splitting of the ionospheric reﬂection is schematically illustrated in Fig. 1.1. Here the apparent height h (or apparent, or virtual, range) corresponds to ct/2, where t is the round trip echo delay time, and the frequency f is the sounding frequency. For a description of the sounding technique, and the inversion from h (f ) to Ne (h), where Ne is the electron number density and h is the true altitude, see Thomas [64] and Reinisch [58] and references therein. In order to understand how the Z mode is related to the free-space O and X modes it is necessary to discuss plasma-wave dispersion. This topic will be addressed in Sect. 1.2. Since the Z mode is an internal (or trapped) mode of the plasma, the emphasis in this paper will be on the reception of the Z mode by space-borne receivers during active and passive experiments. Sections 1.3 and 1.4 will deal with sounder-stimulated Z-mode waves in the ionosphere and the magnetosphere, respectively. Particular attention will be given to the information that the sounder-stimulated Z-mode waves provide concerning magnetic-ﬁeld aligned Ne irregularities (FAI). FAI are irregularities in Ne transverse to the direction of the background magnetic ﬁeld B that are maintained for long distances along B. They eﬃciently scatter and

1 Active Wave Experiments in Space Plasmas: The Z Mode

Apparant Height (h')

Z

O

5

X

fH Sounding Frequency (f)

Fig. 1.1. Ground-based ionogram schematic illustrating Z-, O-, and X-mode reﬂection traces. Here the ionospheric notation for the electron cyclotron frequency, fH , is used [adapted from 57]

duct sounder-stimulated Z-mode waves. Section 1.5 discusses a combined active/passive investigation of Z-mode waves generated by natural processes. A summary is presented in Sect. 1.6. There have been many spacecraft that have generated Z-mode waves in the ionosphere and magnetosphere using radio sounders. Similarly, there have been many satellites that have detected Z-mode waves of magnetospheric origin using plasma wave receivers [LaBelle and Treumann, 43, included a review of auroral Z-mode observations and theory]. Our goal is not to review the Z-mode observations from all of these missions. Rather, it is to select speciﬁc examples that illustrate the range of Z-mode phenomena observed in active space wave-injection experiments and to demonstrate their diagnostic capability. In the case of the ionosphere, we will mainly use data from two missions, namely, (1) the ISIS (International Satellites for Ionospheric Studies) satellites [Jackson and Warren, 33] and (2) the OEDIPUS sounding rocket double payloads (Observations by Electric-ﬁeld Determinations in the Ionospheric Plasma-A Unique Strategy) [see, e.g., 30, 36]. In the case of the magnetosphere, data from the Radio Plasma Imager (RPI) [Reinisch et al., 59] on the IMAGE (Imager for Magnetopause-to-Aurora Global Exploration) satellite [Burch, 15] will be used.

1.2 Plasma Wave Dispersion Waves in a cold plasma are described by a dispersion relation, i.e., the scalar relation expressing the angular frequency ω = 2πf in terms of the propagation vector k, which is related to the refractive index n by n = kc/ω where k = |k| = (2π/λ) and λ is the wavelength.This description has been given in a number of books and review papers [see, e.g., 1, 19, 27, 39, 57, 63]. Figure 1.2 presents dispersion curves for waves propagating in a homogeneous cold

6

R.F. Benson et al. ο

ο ο θ = 0 90 0

R-X

ο

θ=0

(a)

c

(b)

c

θ = 90 θ=0

ο

fx fuh fce

CMA 4

Z (X) Z (R-X) L-O

ο

θ=0

Whistler (R)

θ = 90 ο

CMA 3

θ =0

CMA 6a CMA 7

fpe

ο

ο

θ =0

)

fce

ο

CMA 3 Z (X)

L-X )

fz

ο

0

Z (L-X

L-O

Z(

f

ο

R-X

fx fuh fpe

90

fz

Whistler (R)

k

k

fpe > fce

fpe < fce

Fig. 1.2. Schematic dispersion diagrams. (a) Example when fpe /fce > 1. (b) Example when fpe /fce < 1 [adapted from 27, 62] (Reprinted with permission of the American Geophysical Union)

plasma, where ion motions are neglected, with k making an angle θ relative to B. The ﬁgure shows the dispersion curves for θ = 0 and θ = π/2 cases for a range of frequency and wave number. The region between these two limiting cases is shown by various shades of gray – indicating various modes – where the propagation at oblique wave normal angles is permitted. The waves are labelled based on their polarization for parallel or perpendicular propagation, i.e., R or L for right- or left-hand polarization (with respect to the direction of B) when θ = 0, and X or O for extraordinary and ordinary mode polarization when θ = π/2. In some regions, only one letter is used indicating that propagation is not possible for both θ = 0 and θ = π/2. Thus Z(X) indicates that the Z mode does not include the condition θ = 0 in the region indicated based on the cold-plasma approximation. The Z-mode regions in Fig. 1.2 are also labelled with the CMA designation using the notation of Stix [63]. Thus Z(X) occurs in CMA region 3 where k, and thus n = |n|, can become large leading to a condition (n = ∞) known as resonance; the condition k = 0 (or n = 0) is known as a cutoﬀ. The plasma resonances and cutoﬀs in Fig. 1.2 are given by the following expressions: fce (kHz) =

|e| |B| ≈ 0.028 |B(nT)| 2πme )

(1.1)

1 Active Wave Experiments in Space Plasmas: The Z Mode

7

1

2 1 1 e2 fpe (kHz) = Ne2 ≈ 80.6 Ne (cm−3 ) 2 (4π 2 0 me )

fuh

=

fx

=

fz

1

2 2 2 (fpe + fce )

12 2 fpe 1+4 2 fce

fce 1+ 2 12 2 fpe fce ≡ fx − fce −1 + 1 + 4 2 = 2 fce

(1.2) (1.3) (1.4)

(1.5)

where e is the electron charge, me is the electron mass and 0 is the permittivity of free space. For θ < π/2, the resonance condition that replaces (3) above, in CMA 3 of Fig. 1.2, is known as the Z-inﬁnity and is given by 12 1 1 2 4 2 2 fuh + (fuh − 4fce fpe cos2 θ) 2 fZI = √ 2

(1.6)

The Z inﬁnity is also referred to as the upper oblique resonance [see, e.g., Beghin et al., 4]. The above cutoﬀs and resonances are described using diﬀerent notations in Sects. 6.4 and 6.5 of Ratcliﬀe [57] and Sects. 1–5 of Stix [63]. Figure 1.2 is often presented in the form of ω vs. k. In this presentation the magnitudes of the phase and group velocities, ω (1.7) |vp | = k and

∂ω |vg | = , ∂k

(1.8)

respectively, correspond to the slope of the line from the origin to a particular point on a dispersion curve, and to the slope of a line tangent to the dispersion curve at that point, respectively. In Fig. 1.2 there is a slanting line labelled c to indicate that it corresponds to free-space propagation. The curves to the left of this line (labelled R-X, L-O and L-X) have vp ≥ c and those to the right have vp < c. Accordingly, the Z-mode waves labelled L-X are called fast Z (CMA 4 in Fig. 1.2a and CMA 7 in Fig. 1.2b) and those labelled Z(X) are called slow Z (CMA 3 in Figs. 1.2a and 2b). Both fast and slow Z-mode waves can be found in the region labelled R-X (CMA 6a in Fig. 1.2b). The ionosphere and magnetosphere, contrary to the conditions appropriate to Fig. 1.2, are neither homogeneous nor cold and the ions are not motionless. Yet the dispersion properties derived by using these assumptions, and illustrated by Fig. 1.2, have proved very successful in describing many phenomena. The standard approach is to consider the ionosphere as a horizontallystratiﬁed medium with Ne varying only in the vertical direction. Then the wave is considered to behave as if it were in a homogeneous medium at each

8

R.F. Benson et al.

ionospheric level. Both diagrams in Fig. 1.2 correspond to a speciﬁc value of fpe /fce . The curves change shape as fpe /fce changes. For example, in Fig. 1.2a, the √ band of no propagation between fce and fZ only appears when fpe /fce > 2. Also, the Z(X) region, i.e., CMA-region 3 in Figs. 1.2a and 1.2b, is maximum for the condition fpe = fce . The progress of a radio wave through the non-homogeneous ionosphere can be modelled by considering the change in the shape of the refractive-index surface as fpe /fce changes. This process is illustrated in Fig. 1.3 to illustrate how Z-mode signals at a frequency f from a high-latitude source of natural origin could propagate over great distances in the horizontal direction. The conditions correspond to a source location where f ≈ fpe < fce . The left side of the diagram shows the evolution of the refractive-index surfaces from low to high altitudes corresponding to plasma conditions changing from CMA regions 7 to 6a to 3 in Fig. 1.2b. Gurnett et al. [28] used a construction technique introduced by Poeverlein [54], based on Snell’s law, to argue that a wave at frequency f originating in the region where f < fce will be refracted at the f = fce level and will be able to propagate long distances in the horizontal direction. In Fig. 1.3, the arrows originating at the intersections of the vertical dashed line with these refractive index surfaces indicate the direction of vg in (benson-eq8). Note the change from a closed refractive index surface to an open surface as CMA region 3 is encountered.

RESONANCE

f = f uh

Bo CMA 3

f = f ce

INDEX OF REFRACTION SURFACE CMA 6a

Vg

f = f pe

SOURCE

CMA 7

f = fz Z-MODE CUTOFF

Fig. 1.3. Illustration of the large change in the shape of the refractive-index surface as CMA region 3 is encountered and the ability for long-range horizontal propagation in the Z mode in the polar regions where B is nearly vertical [adapted from 28] (Reprinted with permission of the American Geophysical Union)

1 Active Wave Experiments in Space Plasmas: The Z Mode

9

NORMALIZED FREQUENCY f/fce

6

5 4 3

2

1 0 10 -4

10 -3

10 -2

10 -1

1

10

NOR M ALIZE D WAV E NUMBER kR

Fig. 1.4. Normalized calculated dispersion diagram for a wide range of wavelengths with electron thermal motions included where f is normalized by fce and k is normalized by 1/R where R is the electron cyclotron radius for the case fpe /fce = 1.6 [adapted from Oya, 53] (Reprinted with permission of the American Geophysical Union)

Figure 1.4 shows the result of numerical solutions of the dispersion relations, for a particular case of fpe /fce > 1, for all θ, and for θ ranging from 0 to π/2, when hot-plasma eﬀects are included. In these solutions, electron thermal motions are included and a Maxwellian electron velocity distribution function is assumed but collisions are neglected and the ions are still considered to provide an immobile neutralizing background [Oya, 53]. The right portion of the diagram corresponds to the electrostatic (es) domain and the left portion corresponds to the electromagnetic (em) domain. New wave modes, known as the Bernstein modes [Bernstein, 13], now appear between the nfce harmonics in the es-domain. The Bernstein modes correspond to undamped modes with θ = π/2. Damping rapidly increases for these modes as θ departs from π/2. These es-modes are coupled to the em-domain, with negligible damping, through the Z-mode when θ ≈ π/2, corresponding to frequencies close to fuh , each, in turn, as fuh increases for increasing fpe /fce values. Thus the Z-mode near θ = π/2, i.e., near fuh , is of prime importance in the coupling of energy resulting from wave/particle interactions in the es-domain into the em-domain where the information can be transmitted out of the plasma.

10

R.F. Benson et al.

1.3 Sounder-Stimulated Z-Mode Waves in the Topside Ionosphere 1.3.1 Single Spacecraft: Vertical and Oblique Propagation In a space plasma such as the ionosphere the Z mode can be directly detected by ionospheric topside sounders. Figure 1.5a presents an example of a midlatitude ionogram where ionospheric reﬂections form a clear Z mode trace in addition to O- and X-mode traces. Also seen in Fig. 1.5a are sounderstimulated plasma resonances at fce = fH , fpe = fN , fuh = fT and 2fce = 2fH and an oblique Z trace labelled Z’. The resonances can be used to accurately determine the ambient |B| and Ne values from (1.1)–(1.3) as, e.g., given in the reviews of Muldrew [50] and Benson [7]. The presently accepted interpretation for these principal plasma resonances stimulated by ionospheric topside sounders is based on the investigation by Calvert [16] of the Z trace. He showed that this trace, which lies between fpe and fuh , is the result of ionospheric reﬂections of obliquelypropagating Z-mode waves. These oblique reﬂections result from the shape of the refractive index surfaces, in particular from the change in the shape to form a sphere plus a line parallel to B when the downward-propagating Z-mode wave encounters the level where f = fpe (see the left side of Fig. 1.3). The Z trace is caused by waves reﬂected at this level. This condition was called a “spitze” by Poeverlein [54] and is discussed in some detail in Budden [14]. The Z trace could be explained by using ray tracing with the cold plasma theory. Calvert [16] did not restrict his calculations to the cold plasma approximation, however, and found ray paths that could return to the spacecraft that included propagation beyond the resonance-cone angle limit of cold plasma theory. These echoes were electrostatic in nature and had echo times much greater than the observed Z echoes so the solutions were discarded. McAfee [46], in his investigation of the plasma resonance observed at fpe , found that these hot-plasma solutions had echo delay times comparable to the delays observed on topside ionograms when frequencies very close to fpe were considered. This oblique-echo model was later extended to the plasma resonances observed at fuh [47] and at nfce [50]. Thus, the investigation of oblique Zmode propagation by Calvert [16] provided the fundamental ﬁrst step toward our understanding of sounder-stimulated plasma resonances. Even though the Z-mode is deﬁned as the slow branch of the X mode, it is important to note that the Z and X modes have comparable group velocities near their respective cutoﬀs. This behavior is clearly illustrated in Fig. 1.5b. Though Z-mode waves cannot travel as far from the spacecraft as the freespace O- and X-mode waves, they are useful for determining the vertical Ne distribution out to a few hundred km from the spacecraft as seen in Fig. 1.5c. The good agreement between the Ne values obtained in Fig. 1.5c by inverting the Z-, O- and X-mode ionospheric-reﬂection traces of Fig. 1.5a provides

1 Active Wave Experiments in Space Plasmas: The Z Mode

11

Fig. 1.5. (a) Alouette-2 mid-latitude ionogram, in negative format with signal reception in white on a black background, showing Z, O and X traces. Here the ionospheric notation of H, N , and T is used to represent the subscripts ce, pe, and uh, respectively, in the present work. Also, “S” is used to designate the cutoﬀ frequencies at the satellite. The fzI label and arrow below the ionogram, and near the 0.9 MHz frequency marker, identiﬁes the Z inﬁnity condition given by (6). (b) Corresponding group velocities (Ne = 6800 cm−3 corresponds to fpe = 0.74 MHz from (2), B = 0.254 = 0.254 × 10−4 T = 0.254 × 105 nT corresponds to fce = 0.7112 MHz from (1) and fuh = 1.026 MHz from (3)). (c) Calculated Ne values from each of the traces [adapted from Jackson, 32]

12

R.F. Benson et al.

conﬁdence that the two main assumptions used in the inversion process, namely, vertical propagation and a horizontally-stratiﬁed ionosphere, were justiﬁed since this process is independent for each trace [Jackson, 32]. For vertical propagation the propagation angle relative to B is 90◦ -dip angle, or φ = 51.4◦ using Jackson’s notation of Fig. 1.5b. Using this φ value in (6), with the other values from Fig. 1.5b where fpe /fce = 1.04, yields fZI = 0.968 MHz in agreement with the observed narrow vertical modulated feature observed from zero to approximately 3,000 km apparent range; it is labelled fzI just below the ionogram. This feature can be observed because the wide receiver bandwidth (3 dB bandwidth of 37 kHz) allows long-duration signals returning from the previous pulse (diﬀering by only a few kHz) to be observed at the start of the receiving interval, i.e., it corresponding to a wrap-around of the apparent-range scale for the asymptotic Z-mode echo. Figure 1.6 shows examples of Z-mode echoes that clearly illustrate the limiting behavior of the Z-mode cold-plasma dispersion curves in Figs. 1.2 and 1.4 for conditions of nearly parallel and nearly perpendicular propagation relative to B. In Fig. 1.6a, corresponding to high latitude and thus nearly parallel propagation, the Z- and O-mode traces touch one another near 700 km apparent range and 0.9 MHz suggesting coupling like in the dispersion diagrams in Figs. 1.2 and 1.4 for θ = 0. Also, the Z trace has a large apparent range at this frequency, coinciding with the combined fpe and fce plasma resonances, as would be expected from (1.6) with θ = 0, i.e., fZI = fpe = fce . From the observed plasma resonances and wave cutoﬀs in Fig. 1.6a, and equations (1.3)–(1.5), fpe /fce = 0.92/0.935 = 0.98 for this ionogram. In Fig. 1.6b, corresponding to low latitude and thus perpendicular propagation, the Z-mode trace becomes asymptotic to fuh , again, as expected from (1.6) (now with θ = π/2). In this case, fpe /fce = 1.89/0.565 = 3.35. When nearly parallel or nearly perpendicular propagation is involved in the presence of FAI, dramatic ionogram signatures can be produced due to ducting and scattering, respectively, of the sounder-generated Z-mode signals. 1.3.2 Single Spacecraft: Ducted Propagation When ionospheric topside sounders encounter equatorial plasma bubbles, dramatic ﬂoating X-mode echoes are observed that resemble epsilons [Dyson and Benson, 22]. They are called ﬂoating because they are not tied to the zerotime baseline at the top of the ﬁgure. These traces are the result of soundergenerated signals that echo in both the local and conjugate hemispheres (relative to the location of the satellite) due to ducted propagation in FAI that are maintained from one hemisphere to the other. The bottom portion of Fig. 1.7 illustrates the top segment of such an X-mode epsilon in the frequency range above about 2.9 MHz and at apparent ranges beyond 2200 km. The X-mode echo just above this echo signature, i.e., corresponding to virtual ranges less than 2200 km in the bottom portion of Fig. 1.7, is due to ducted propagation in the local hemisphere. In the top portion of Fig. 1.7 the distances to the

1 Active Wave Experiments in Space Plasmas: The Z Mode

13

Fig. 1.6. (a) ISIS-2 high-latitude ionogram showing Z, O and X traces under conditions of nearly parallel propagation (Resolute Bay digital ionogram obtained from http://nssdc.gsfc.nasa.gov/space/isis/isis-status.html corresponding to 1500:28 UT on day 126 of 1973; 62.8◦ latitude, −101.8◦ longitude, 1394 km altitude, 83◦ dip). (b) Low-latitude Alouette-2 ionogram showing Z, O and X traces under conditions of nearly perpendicular propagation [adapted from Benson, 6]. As in Fig. 1.5, the ionospheric notation of N , and T is used to represent the subscripts pe, and uh, respectively, and numerals are used to identify the nfH = nfce resonances. The 3fce resonance contains two spurs on the low-frequency side (see Sect. 1.3.6) with delay times near 4 and 7 ms. The insert shows ﬁve selected receiver amplitude vs. time traces corresponding to these spur observations. Each trace represents a vertical scan line on the ionogram display, the amplitude modulation on the insert traces corresponding to the intensity modulation on the ionogram scan lines. The initial systematic positive and negative spikes on each of the insert traces are calibration and sync pulses [see Fig. 31 of Franklin and Mclean, 25]; the time-delay zero point was taken as the left side of the dashed line segment on the insert trace labelled (a). The nfce resonances indicated that fce = 0.565 MHz corresponding to τp = 3.25 ms (Reprinted with permission of American Geophysical Union)

14

R.F. Benson et al.

Fig. 1.7. ISIS-1 ionograms recorded 1000 km above the dip equator showing Zand X-mode echoes from within an equatorial plasma bubble [9]. The ionospheric notation of H, and N is used to represent the subscripts ce and pe, respectively; numerals are used to identify the nfH = nfce resonances (Reprinted with permission of American Geophysical Union)

reﬂection levels in each hemisphere were the same, as the two traces merge and have the same virtual ranges from about 2200 to 2400 km. Z-mode echoes from waves that are ducted along FAI can form truncated versions of these ﬂoating X-mode epsilons. In both portions of Fig. 1.7 the ducted Z-mode waves are conﬁned to a narrow frequency range below the label Z at the top of the ﬁgure. As in the case of the ducted X-mode signals, the Z-mode traces tied to the zero virtual-range scale correspond to signals ducted within the local hemisphere and those beyond about 2000 km in virtual range correspond to signals that experience ducted propagation into both

Virtual Range (km)

1 Active Wave Experiments in Space Plasmas: The Z Mode

0

H

2

15

1112 13 3 4 5 6 7 8 9 10 ZNX 14

1000 2000 3000 4000 0.1 0.2

0.5 0.55

0.9

1.25 1.5 1.6

2.5 2.0

4.5

6.5 7.0

8.5 10.5

Frequency (MHz) Fig. 1.8. A portion of an Alouette-2 ionogram showing Z-mode ﬂoating echoes, attributed to wave ducting in FAI, from fZ , to a maximum frequency prior to fpe labelled Z and N at the ﬁgure, respectively. The label notation is the same as in Fig. 1.7 [adapted from Benson, 9] (Reprinted with permission of the American Geophysical Union)

hemispheres. Note that the ducted Z-mode traces terminate before they reach the local fpe , designated by N at the top of the ﬁgure. Similar frequency restrictions of the reception of sounder-stimulated Z-mode waves attributed to wave ducting were commonly observed in an investigation of FAI near 500 km based on Alouette-2 perigee observations [9]; an example is shown in Fig. 1.8. In a comparison of wave ducting in diﬀerent wave modes, assuming small propagation angles inside a duct produced by a small increase in refractive index, Calvert [17] showed that under conditions of high Ne , similar to the conditions of Fig. 1.8, Z-mode ducting should be stronger than X- and O-mode ducting. The strongest Z-mode ducting was found to occur in the frequency range from fZ to midway between fZ and fpe where a transition from trough ducting to crest ducting occurs due to a curvature reversal in the refractiveindex surface. Calvert [17] argues that ducting cannot be maintained across the curvature reversal in agreement with the upper-frequency truncation of the ﬂoating Z-mode signals attributed to FAI wave ducting in Fig. 1.7 and Fig. 1.8. 1.3.3 Single Spacecraft: Wave Scattering Sounder-generated Z-mode waves that are scattered by FAI lead to strong signal returns in the frequency region between the greater of fce and fpe and less than fuh , i.e., in CMA region 3 in Figs. 1.2 and 1.4. Z-mode ray-tracing calculations indicate that the ray becomes horizontal in the ionosphere as the refractive index tends to inﬁnity [Lockwood, 45] leading to a condition Muldrew [49] termed “wave trapping”. The resulting signature is labelled as a noise band in Fig. 1.9. Denisenko et al. [21] used such signatures in the

16

R.F. Benson et al.

fN

fH

fT

APPARENT RANGE (Km)

0

1000

2000

3000 NOISE BAND

4000

0.2

0.5

0.9

1.25

1.5 1.6

2.0

3.5

FREQUENCY (MHz) o

o

29 DECEMBER , 1967 2257/14 GMT (99 W, 59 N) SATELLITW HEIGHT 1550Km

Fig. 1.9. An Alouette-2 ionogram showing sounder-generated Z-mode signal returns attributed to wave scattering in FAI (labelled “noise band”) [Muldrew, 49]. The label notation is the same as in Fig. 1.5a (Copyright 1969 by IEEE; reprinted with permission)

COSMOS-1809 topside-sounder data to investigate the global distribution of small-scale (∼10–100 m) FAI in the 940 to 980 km altitude range. 1.3.4 Dual Payloads: Slow Z The above discussion pertained to observations where the receiver and transmitter were on the same spacecraft and they shared a common antenna. Thus the received Z-mode signals correspond to echo returns either from short distances, due to scattering from FAI in the vicinity of the satellite (Fig. 1.9), or from long distances (∼100’s km) due to vertical propagation (Z trace in Fig. 1.5a), oblique propagation (Z trace in Fig. 1.5a), or to nearly parallel propagation that is ducted along FAI (Figs. 7 and 8). The Z mode has also been investigated during space experiments involving wave propagation between receivers and transmitters on diﬀerent payloads. James [34, 35] performed such experiments during a high-latitude rendezvous between ISIS 1 and ISIS 2. The operating frequency was in the range fpe < fce < f < fuh , i.e., corresponding to “slow Z-mode” propagation in CMA-region 3 of Fig. 1.2b. These waves were observed to propagate over distances of several hundred km. It was found [James, 35] that the observed transmission-reception signal delay times could be explained by ray optics but that the observed distortion of the received pulse relative to the transmitted pulse indicated the importance of signal scattering by FAI. The OEDIPUS-A dual-payload rocket was launched from the Andøya Rocket range in Norway on January 30, 1989. It was dedicated to such

1 Active Wave Experiments in Space Plasmas: The Z Mode

17

two-point measurements and provided an additional opportunity to investigate slow Z-mode propagation [James, 36]. In this case, fce < fpe < f < fuh , i.e., corresponding to slow Z-mode propagation in CMA-region 3 of Fig. 1.2a. The transmitting and receiving payloads were separated by nearly 1 km and the separation direction diﬀered from the B direction by only a few degrees. The observed delay times for the Z-mode were too large to be explained by free-space propagation and the full em solution of the hot plasma dispersion equation, based on the work of Lewis and Keller [44] and Muldrew and Estabrooks [51], was used to investigate the problem. Using this hot-plasma approach, James [36] constructed refractive-index surfaces appropriate to the problem (see Fig. 1.10). Note how diﬀerent these hot-plasma Z-mode refractive-index surfaces are from the cold plasma surfaces illustrated on the left side of Fig. 1.3 in the CMA-region 3. James [36] found that direct ray paths connecting the two payloads had |vg | values too small to explain the observed delays in the large n region assuming a smooth horizontally-stratiﬁed medium. Thus waves corresponding to such large-n dispersion solutions would arrive well after the OEDIPUS A ionogram display time limit (by approximately a factor of 10) and would not be detected. When the region of smaller n was investigated, labelled “electromagnetic and quasi-electrostatic domain” in Fig. 1.10, and the payloads were assumed to be within an Ne depletion duct (with cross-B dimension ∼ 100 m), ducted ray paths could be found that were consistent with the observations. James [36] suggested that such ducting may be common in the auroral ionosphere and that it should be considered when trying to interpret natural Z-mode emissions. The OEDIPUS-C rocket dual-payload was launched from the Poker Flat rocket range in Alaska on November 7, 1995. Again, the sounder-transmitter was on one payload and the sounder-receiver was on the other, and the separation direction between the payloads, now separated by more than a 1 km, differed from the B direction by only a few degrees. In this case, James [37] investigated the slow Z-mode propagation corresponding to fpe < fce < f < fuh , i.e., to the CMA-region 3 of Fig. 1.2b. He found, using hot-plasma dispersion theory, that the calculated propagation times for rays directly connecting the observed payloads were typically more than a factor of three greater than the observed time delays between signal transmission and reception. Thus waves corresponding to such solutions would arrive well beyond the observing time base and would not be detected. In all other cases, there were no solutions corresponding to the desired direction. The received signals could be explained, however, in terms of incoherent Cherenkov and cyclotron radiation from sounder-accelerated electrons (SAE). Particle detectors on both payloads detected SAE following sounder transmissions from the transmitting payload. James [37] could reproduce the observed signal delay times and could predict (within an order of magnitude) the observed signal intensities.

18

R.F. Benson et al.

Fig. 1.10. Hot-plasma Z-mode refractive-index surfaces for f = 2.534 MHz, fce = 1.2 Hz, Te = 2000 K and fpe , labelled fp in the ﬁgure, in the range 2.233 = fpe = 2.533 MHz [James, 36] (Reprinted with permission of the American Geophysical Union)

1.3.5 Dual Payloads: Fast Z Horita and James [29, 30] investigated fast Z-mode propagation using OEDIPUS-C data corresponding to frequencies below fpe and above the greater of fZ or fce , i.e., to CMA-region 4 in Fig. 1.2a where there are no competing cold-plasma wave modes to complicate the interpretation. The Z

1 Active Wave Experiments in Space Plasmas: The Z Mode

19

mode was found to be stronger than almost all the other cold-plasma modes and to be strongest at frequencies just below fpe for fpe /fce near and greater than 1 but strongest just above fZ for higher fpe /fce values (between 2 and 3). Using cold-plasma dispersion theory, the Balmain [2] antenna-impedance theory and the Kuehl [40] dipole radiation theory they found that the observed and calculated signal intensities were generally in good agreement. They attribute the strength of the Z-mode signals, relative to the other free-space modes, to antenna-impedance values that permit eﬃcient coupling between the antenna and the transmitter and receiver. 1.3.6 Possible Role of Z-Mode Waves in Sounder/Plasma Interactions Among the plasma instability and nonlinear phenomena stimulated by ionospheric topside sounders investigated by Benson [8] was a diﬀuse feature observed in the frequency range above the greater of fce and fpe and below fuh , i.e., in the CMA 3 slow Z-mode regions of Fig. 1.2. It was designated as the DNT resonance because of its generally diﬀuse appearance on topside ionograms and its location between fN and fT (ionospheric notation for fpe and fuh ). A weak short-duration (< 1 ms) example of this resonance is shown in Fig. 1.6a. It is often observed for up to about 5 ms. It is not observed over the entire listening range, however, and thus is distinguished from the noise band in this frequency range. This noise band, attributed to wave scattering (see Sect. 1.3.3), is evident in Fig. 1.6a and, more prominently, in Fig. 1.9. No theoretical interpretation has been oﬀered that explains the frequency and time-duration characteristics of this resonance. It has been attributed by Pulinets [55], Pulinets et al. [56] to the scattering of sounder-generated Z-mode waves and used as a diagnostic tool for the investigation of the distribution of small-scale FAI in the topside ionosphere. James [38] attempted to explain the DNT resonance as observed by the ISIS-II sounder in terms of radiation from SAE in analogy with the successful explanation of Z-mode signals observed by the OEDIPUS-C sounder receiver as described in Sect. 1.3.4. While this explanation was not found to explain the ISIS-II observations, he concluded that SAE may still play a role because SAE pulses that persisted for milliseconds were observed when the OEDIPUS-C sounder transmitter was tuned to the DNT frequencies. Features have been observed on topside ionograms that imply that ion motions must be considered for a proper interpretation. They appear either as prominent protrusions (called spurs) on the electron resonances (most often from the low-frequency side) or as narrow (in time delay, i.e., apparent range) emissions between the resonances. In either case, they appear with delay times that correspond to multiples of the proton gyroperiod 1/fcp . One of these phenomena, the proton spurs on the nfce resonances, appears to be strongly inﬂuenced by Z-mode transmissions [Benson, 6]. The spurs are

20

R.F. Benson et al.

greatly enhanced when fZ , from (1.5), is near, but slightly less than, nfce for n = 2, 3, 4, . . . and the largest spurs are observed for large n. Figure 1.6 illustrates the spurs observed on the 3fce resonance when fZ ≈ 3fce . It was suggested that the Z mode may be more eﬃcient at coupling energy into the plasma under these conditions. Note that this frequency region just above fZ corresponds to the fast Z region where Horita and James [29, 30] found the strongest Z-mode signals which they attributed to optimum antenna-impedance values (see Sect. 1.3.5). Their larger fpe /fce values, which produce the strongest signals just above fZ , would correspond to larger fZ values which, in turn, would correspond to higher nfce resonances satisfying the condition fZ ≈ nfce ; the largest proton spurs were observed under just such conditions [Benson, 6]. Unique Z-mode topside-ionogram signatures have been observed in low latitudes that suggest topside sounders are capable of stimulating, or enhancing, FAI when they encounter the plasma conditions fpe /fce ≈ n where n is an integer larger than 3 [Benson, 10]. An illustration for the case of n ≈ 5 is presented in Fig. 1.11. Note that well-deﬁned Z, O and X traces are clearly seen for the cases fpe /fce < 5 (top panel) and fpe /fce > 5 (bottom panel) but that the Z trace is masked by a long-duration diﬀuse signal that extends from fZ to part way to fpe . It was argued that the frequent occurrence of such signatures made it unlikely that the spacecraft was just encountering FAI when the ambient conditions fpe /fce ≈ n were satisﬁed. Thus the sounder-generated Z-mode waves were considered to be ducted in FAI stimulated, or enhanced, by the sounder on a short time scale (1 s). The possibility that this sensitive diagnostic role of the Z-mode waves could be due to eﬃcient scattering when fpe /fce ≈ n was investigated by Zabotin et al. [69]. They did not ﬁnd any sensitivity in the scattering of Zmode waves by FAI near the magnetic equator to the conditions fpe /fce ≈ n and concluded that the above examples were either due to sounder stimulation, or enhancement, as proposed or to ducting conditions sensitive to these conditions. No study of the sensitivity of Z-mode ducting by FAI to the conditions fpe /fce ≈ n has been made. As pointed out in Sect. 1.3.2, however, Calvert [17] found that under high Ne conditions (as indicated in Fig. 1.11) Zmode ducting in the frequency range between fZ and midway to fpe should be stronger than O- or X-mode ducting, a prediction consistent with the observations in the middle panel of Fig. 1.11. Thus the sensitivity to the fpe /fce ≈ n condition is likely in the generation, or enhancement of existing, FAI by the sounder. Benson [10] gave other examples of sounder-stimulated plasma phenomena when fpe /fce ≈ n and suggested that more energy is deposited into the plasma under these conditions and, particularly, when fpe /fce ≥ 4. Osherovich et al. [52] investigated large-amplitude cylindrical electron oscillations appropriate to FAI, with initial conditions chosen so as to favor Z-mode stimulation, and found that the resulting frequency spectrum was very sensitive to the fpe /fce value, with larger amplitudes, and more nonlinear frequency components, observed when fpe /fce ≈ n, and that the eﬀect

1 Active Wave Experiments in Space Plasmas: The Z Mode

21

Fig. 1.11. Consecutive Alouette-2 low-latitude ionograms revealing long-duration Z-mode echoes only when fpe /fce ≈ 5 at 05:15 UT on 17 June 1969 [Benson, 10]. The label notation is the same as in Fig. 1.7 (Reprinted with permission from Elsevier)

was observed to increase with increasing n. Kuo et al. [41] investigated the creation of FAI by a Z-mode pump wave under the above resonant conditions. They added the constraint that the Z-mode pump wave frequency fo ≈ fpe at a short distance from the satellite where a four-wave coupling process takes place. One of the products in this coupling process corresponds to short-scale (meter size) FAI. They propose that the observed Z-mode diﬀuse signals, like those in the middle panel of Fig. 1.11, are caused by scattering from these FAI but stress the need for additional research to identify mechanisms that could generate (or enhance) large-scale FAI (>100 m) capable of supporting wave ducting.

22

R.F. Benson et al.

1.4 Sounder-Stimulated Z-Mode Waves in the Magnetosphere 1.4.1 Remote O-Z-O-Mode Coupling Radio sounding in the magnetosphere is challenging because the distances are so large and Ne is so low. These constraints motivated the RPI design for the IMAGE mission. As a result, the IMAGE satellite contains the largest structures ever placed on a spinning satellite, namely, the RPI spin-plane crossed dipole antennas (originally 500 m tip-to-tip length for each). Soon after the IMAGE launch on 25 March 2000 into an elliptical polar orbit, with an apogee of 8 Earth radii (RE ) geocentric distance and a perigee altitude of about 1000 km, the RPI detected discrete long-range echoes outside the plasmasphere in the north polar region. Reinisch et al. [60] and Carpenter et al. [19] attributed them to signals propagating in the X mode in FAI to the polar ionosphere where they were reﬂected; they were clearly distinguished from the diﬀuse shorter-range echoes from the nearby highly-irregular plasmapause boundary. Later, echo signatures indicating inter-hemisphere propagation, similar to the ionospheric example shown in Fig. 1.7, were identiﬁed by Fung et al. [26]; an example where the Z mode played a prominent role in the interpretation is shown in Fig. 1.12 from Reinisch et al. [61]. This record is called a plasmagram and is the magnetospheric analog of the ionospheric topside-sounder ionogram examples shown in Figs. 1.5–1.9 and 1.11. The virtual range vs. frequency curves through the data points corresponding to X-mode reﬂections from the local and conjugate hemispheres (labelled SX and N X, respectively) were used to derive the hemisphere-tohemisphere Ne distribution along the magnetic ﬁeld line through the satellite [31]. This ﬁeld-aligned Ne distribution was then used to calculate the reﬂections expected for transmitted O-mode signals that coupled to Z-mode signals, which reﬂected at the distances where the transmitted frequencies were equal to the fZ cutoﬀ frequencies given by (1.5), and then coupled back to the O-mode signals that were received at the satellite. These O-Z-O traces are labelled N Z and SZ for the echoes of this type from the northern and southern hemispheres, respectively, in Fig. 1.12. This mode-coupling interpretation of Reinisch et al. [61], involving the Z-mode to explain the weaker companion echoes to the inter-hemisphere magnetospheric RPI X-mode echoes, differs from the interpretation of Muldrew [48] of inter-hemisphere ionospheric echoes observed by Alouette 1 in that Muldrew [48] did not invoke O-Z-O mode coupling. 1.4.2 Local Z-Mode Echoes Echoes of Z-mode signals are often directly observed by RPI in the magnetosphere. Figure 1.13 shows an example when IMAGE was near the plasmapause and a strong Ne gradient could be determined from multiple plasma

1 Active Wave Experiments in Space Plasmas: The Z Mode

23

RPI Plasmagram 17:50 October 24, 2000 7.0

SX+NX

Virtual Range (RE)

6.0

NZ

5.0

NP

4.0

NX

SP SZ

3.0 2.0

SX

ce

2

4

1.0 30

100

300

1000

Frequency (kHz) Fig. 1.12. An RPI plasmagram showing echoes from both hemispheres in the frequency region above about 300 kHz, with interpretive traces through the data points, and short-duration plasma resonances at fce , 2fce and 4fce (lower left). The labels S, N, X, and Z denote southern and northern hemispheres and X and O-Z-O traces respectively. The N X trace has been extrapolated to higher frequencies. The insert shows the orbit, location of IMAGE (x) and the L = 4 dipole ﬁeld lines; NP and SP indicate the magnetic poles [adapted from Reinisch et al., 61] (Reprinted with permission of the American Geophysical Union)

resonances and wave cutoﬀs. The frequency range of this high-resolution plasmagram fortuitously included the Z- and X-mode cutoﬀs (fZ and fX , respectively) and several plasma resonances. (The linear frequency step size between transmissions of single 3.2 ms pulses was equal to the RPI bandwidth of 300 Hz.) The plasma conditions corresponded to Fig. 1.2a; the Z mode from fZ to fpe is in CMA region 4 and is the only cold-plasma wave mode. The diﬀuse nature is attributed to scatter returns from FAI. These scatter returns become more prominent in the CMA region 3 between fpe and fuh (see Fig. 1.2a). This enhanced scatter forms a noise band analogous to the noise band in the ionospheric example of Fig. 1.9 (which, however, corresponds to the CMA 3 region of Fig. 1.2b). Using the values scaled from Fig. 1.13 in (1.3)– (1.5) reveals that consistent solutions cannot be obtained for these equations with constant fce and fpe values over the 41-s time interval required to record this plasmagram. Three independent fpe determinations can be made, however, if the Tsy 96–1 model magnetic-ﬁeld values [Tsyganenko, 65, 66, 67],

24

R.F. Benson et al.

Fig. 1.13. An RPI plasmagram recorded during an outbound plasmapause crossing revealing Z- and X-mode cutoﬀs and plasma resonances at fpe , fuh , and 4fce labelled at the top by z, x, pe, uh, and 4, respectively. Also labelled, as Q3, at the top is one of the resonances associated with the Bernstein modes discussed in Sect. 1.2 in connection with Fig. 1.4 [adapted from Benson et al. 11] (Reprinted with permission of the American Geophysical Union)

with a percentage oﬀset correction based on the observed 4fce plasma resonance in Fig. 1.13, are used corresponding to the spacecraft locations at the times of the recording of fZ , fuh , and fX . The plasma resonance observed at fpe in Fig. 1.13 provides a fourth measurement of fpe and it is independent of the fce value. While the deduced Ne gradient is large (an order-of-magnitude decrease in a change in L value of approximately 1) [Benson et al., 11], it is only about 1/10 the gradient of a well-developed plasmapause [Carpenter and Anderson, 18]. 1.4.3 Z-Mode Refractive-Index Cavities Among the most spectacular echo signatures observed on RPI plasmagrams are those corresponding to the direct transmission and reception of Z-mode waves that are ducted along FAI within refractive-index cavities [Carpenter et al., 20]. Examples are shown in Fig. 1.14. They were obtained when RPI was operating with a linear frequency step size of 900 Hz. From the observed fZ values in Fig. 1.14 (corresponding to the onset of the Z-mode traces), and the model values for fce (oﬀ scale to the right in both plasmagrams), it is deduced from (1.5) that fpe is also oﬀ scale to the right in both plasmagrams. In both cases, fpe < fce , i.e., propagation corresponding to CMA region 7 in Fig. 1.2b is involved.

1 Active Wave Experiments in Space Plasmas: The Z Mode

25

4.0 (a)

3.0 2.0

50 40

fZ

0

fZ

4.0

30 20

3.0

AmpX (dBnV/m)

VIRTUAL RANGE (RE )

60

1.0

10 W

2.0

0

1.0 (b)

0 180

200 220 FREQUENCY (kHz)

240

Fig. 1.14. RPI plasmagrams showing multicomponent Z-mode echoes recorded near perigee in the region of Ne gradients between the southern aurora zone and the plasmasphere. (a) L ≈ 3.2, altitude ∼3800 km, fZ = 194 kHz, fce (model) = 382 kHz implies fpe ≈ 334 kHz or fpe /fce ≈ 0.9; 0824 UT on 26 July 2001. (b) L ≈ 2.9, altitude ∼ 2700 km, fZ = 216 kHz, fce (model) = 469 kHz implies fpe ≈ 384 kHz or fpe /fce ≈ 0.8; 0245 UT on 12 July 2001 (whistler-mode echoes, due to reﬂections from the bottom side of the ionosphere [62], are marked “W”) [adapted from Carpenter et al. 20] (Reprinted with permission of the American Geophysical Union)

The virtual ranges of the observed echoes starting near 1.5 RE , which appear as upward slanting epsilons in Figs. 1.14a and 1.14b, are too short to be explained in the same manner as used for the echoes shown in Fig. 1.12, i.e., they are too short to be attributed to echoes from the conjugate hemisphere. They can be explained, however, in terms of ducted echoes returned from within a refractive-index cavity in the hemisphere containing the IMAGE satellite. The interpretation presented by Carpenter et al. [20] is illustrated in Fig. 1.15 for the case when the IMAGE satellite is assumed to be located below a relative minimum in the proﬁle of fZ along B, as deduced from (1.5) with fce and fpe values based on models and RPI observations. The most prominent features in Fig. 1.14a are reproduced in Fig. 1.15a with labels corresponding to the ray paths deﬁned in Fig. 1.15b which displays

26

R.F. Benson et al.

Fig. 1.15. (a) Reproduction of the most prominent echoes in Fig. 1.14a. (b) Schematic fZ proﬁle limiting ray paths A and B for representative frequencies fi , fj , and fk , relative to the location of the IMAGE satellite. These ray paths correspond to the labels used for the echo traces in (a) [Carpenter et al., 20] (Reprinted with permission of American Geophysical Union)

an idealized fZ proﬁle along B. When the sounder frequency reaches the frequency fi , corresponding to the condition fi = fZ at the satellite level as illustrated in Fig. 1.15b, a wave can propagate upward along path B into the region where fi > fZ and be reﬂected at a higher altitude where the condition fi = fZ is again satisﬁed. This returning wave is responsible for the echo with a virtual range of ≈ 1.5 RE , i.e., the nose of the ﬁrst upward slanting epsilon signature in Fig. 1.15a. This wave is reﬂected again at the fi = fZ condition at the satellite and the process is repeated. Two such repetitions are evident in the data of Fig. 1.14a and the reproduction in Fig. 1.15a. Higher sounder frequencies, such as fj in Fig. 1.15b, correspond to the condition fj > fZ at the satellite level, and waves can now propagate both upward along path B and downward along path A, within the region where fj > fZ , and be reﬂected at both higher and lower altitudes where the condition fj = fZ

1 Active Wave Experiments in Space Plasmas: The Z Mode

27

Fig. 1.16. Ne along the L = 3.2 magnetic-ﬁeld line above the IMAGE satellite derived from inverting the “B” Z trace of Fig. 1.15a compared with an RPI-derived empirical model of Huang et al. [31] for a diﬀerent day (8 June 2001) and a scaling of that model by a factor of 0.7. The portion of the lowest curve corresponding to magnetic latitude values less than 13◦ is an extrapolation [Carpenter et al. 20] (Reprinted with permission of the American Geophysical Union)

is satisﬁed. Multiple combinations of these echoes produced the elements of the epsilon signatures seen in Figs. 1.14a and 1.15a. Carpenter et al. [20] also presented examples of a special form of echo signature on RPI plasmagrams that corresponds to ducted Z-mode propagation along FAI within Z-mode refractive-index cavities when IMAGE is assumed to be located above the relative minimum in the proﬁle of fZ along B. Carpenter et al. [20] introduced an inversion method to determine the Ne distribution along B from the upward propagating signals within Z-mode refractive-index cavities of the type discussed above. The results of applying this method to the trace corresponding to B in Fig. 1.15a are presented in Fig. 1.16. Also presented in Fig. 1.16 are predicted values from an empirical Ne model based on the inversion of RPI X-mode echoes from signals that propagated on multiple ﬁeld-aligned paths on a diﬀerent day, namely, 8 June 2001 [Huang et al., 31]. The lower Ne values derived from the Z-mode data were attributed to the movement of IMAGE through a region of plasmapause Ne gradients at the time of the measurements. Since these Z-mode echoes from waves trapped in Z-mode refractive-index cavities are often the only prominent echoes observed on a single plasmagram, this inversion method provides a valuable diagnostic tool for determining the Ne distribution along B.

28

R.F. Benson et al.

1.4.4 Whistler- and Z-Mode Echoes In an investigation of IMAGE/RPI data in the inner plasmasphere and at moderate to low altitudes over the polar regions, Sonwalkar et al. [62] found diﬀuse Z-mode echoes often accompanied whistler (W)-mode echoes. An example from their study is shown in Fig. 1.17. The W-mode echo in this ﬁgure, with narrowly deﬁned time delay as a function of frequency, is an example of a discrete echo. The Z-mode echo, with a time-delay spread that increases with frequency, is an example of a diﬀuse echo. As discussed by Carpenter et al. [20], this Z-mode pattern is characteristic of the low altitude polar region and the plasma condition fpe /fce < 1. The abrupt high-frequency cutoﬀ of this Z-mode echo, and the long-duration sounder-stimulated plasma resonance at ∼787 kHz in Fig. 1.17, provides a measure of fuh [Benson et al., 14]; the gap, or decrease in echo spreading at ∼685 kHz, provides a measure of fce [Carpenter et al., 20]. From these values fpe is calculated to be ∼387 kHz from (1.3). Whistler-mode echoes with a much broader range of time delays with frequency than those shown in Fig. 1.17 are also observed on IMAGE. They are called diﬀuse W-mode echoes [Sonwalkar et al., 62]. In regions poleward of the plasmasphere, diﬀuse Z-mode echoes of the kind illustrated in Fig. 1.17 were found to accompany both discrete and diﬀuse W-mode echoes 90% of the time, and were also present during 90% of the soundings when no W-mode echoes were detected. Based on comparisons of ray tracing simulations with the observed dispersion of W- and Z-mode echoes, Sonwalkar et al. [62] proposed that: (1) the observed discrete W-mode echoes are due to RPI signal reﬂections from the

Fig. 1.17. RPI plasmagram displaying both W-mode echoes (frequencies below ∼300 kHz) and diﬀuse Z-mode echoes (frequencies above ∼300 kHz) labelled WM and ZM, respectively. The minimum observable time delay is 13 ms due to the 3.2ms minimum transmitted pulse length and additional time needed for the receiver to recover from the high voltage generated during the transmitter pulses. The amplitude scale is coded from 10 to 50 dB nV/m [adapted from Sonwalkar et al., 62] (Reprinted with permission of the American Geophysical Union)

1 Active Wave Experiments in Space Plasmas: The Z Mode

29

lower boundary of the ionosphere, (2) the diﬀuse W-mode echoes are due to scattering of RPI signals by FAI located within 2000 km earthward of IMAGE and in directions close to that of the ﬁeld line passing through IMAGE, and (3) the diﬀuse Z-mode echoes are due to scattering of RPI signals from FAI within 3000 km of IMAGE, particularly to signals propagating in the generally cross-B direction. This interpretation suggests that Z-mode echoes occur most frequently (∼ 90%), both in the presence and absence of whistler-mode echoes, because the Z-mode waves capable of returning to the sounder can propagate long distances in all directions, i.e., not only close to the ﬁeld lines as in the case of the whistler mode waves that are capable of returning to the sounder. Thus there is a much larger probability of encountering plasma irregularities which may lead to Z-mode echoes. These RPI results are consistent with previous investigations in that they indicate that the high-latitude magnetosphere is highly structured with FAI that exist over cross-B scales ranging from 10 m to 100 km and that these FAI profoundly eﬀect W- and Z-mode propagation.

1.5 Active/Passive Investigation of Z-Mode Waves of Magnetospheric Origin The ﬁrst observation of enhanced Z-mode radio emissions of natural origin, corresponding to CMA-region 3 in Fig. 1.2, were made during the radioastronomy rocket experiment of Walsh et al. [68]. They ruled out a thermal source due to the large signal intensities and suggested Cherenkov radiation as a likely source mechanism because of the large refractive index (and hence low wave phase velocities) in this frequency domain. Bauer and Stone [3], using satellite observations, were the ﬁrst to show that the observed frequency limits of this CMA region 3 Z-mode radiation could be used to determine the magnetospheric Ne . Several later experiments have investigated these emissions attributed to CMA region 3 Z-mode radiation by comparing the observed frequencies with fpe values determined by active techniques. Beghin et al. [4] used the AUREOL/ARCADE-3 mutual impedance probe in the 400–2000 km altitude region, Kurth et al. [42] used a sounder during a brief (5 min) period of the single pass through the terrestrial magnetosphere by Cassini, and Benson et al. [12] used active soundings by the RPI on four passes of IMAGE in the vicinity of the plasmapause region. In each of these investigations, it was concluded that the upper and lower frequency boundaries of an observed intense upper-hybrid band corresponded to fuh and fpe , respectively, in the region where fpe > fce . In the study by Benson et al. [12] these frequency identiﬁcations were found to hold to an accuracy of a few per cent in fpe by interpolating between active soundings to the intervening passive dynamic spectra. Figure 1.18 shows the results of superimposing the plasmagram-determined fce , fpe and fuh values from active

30

R.F. Benson et al. 200 100

dBV/ Hz

50

10

-90

5 200

a 25 March 2003

100

fuh

50 20

fce

10 fpe

4 2 0 -2 -4

-120 -130 -140

02

-150

00

0 -2 -4 -6 -8 X (RE)

5 0000

-100 -110

Z (RE)

Frequency (kHz)

20

b 0030

0100

0130

0200

UT Fig. 1.18. (a) Passive RPI dynamic spectrum. (b) Same as (a) except with superimposed fce , fpe and fuh values determined from active RPI plasmagrams [adapted from Benson et al., 12] (Reprinted with permission of the American Geophysical Union)

sounding on the passive RPI dynamic spectrum corresponding to the same time interval. Comparing Figs. 1.18a and 1.18b illustrates the beneﬁt of having active soundings to conﬁdently determine fpe , particularly when fpe < fce . The sounder-derived fpe values (white triangles) follow the upper edge of an intense, presumably whistler mode, emission. They deviate from this upper edge, however, at a location that would be diﬃcult to determine without the active soundings. Also, in this fpe < fce frequency domain, the upper-hybrid band enhancement, in this case between fce and fuh , is often not very well deﬁned as fuh approaches fce . Beghin et al. [4] never observed an enhancement when fpe < fce ; they attribute this ﬁnding to a lack of instability growth of Z-mode waves in the upper-hybrid band under these conditions. The upperhybrid band is relatively broad (extending from ∼50 to 60 kHz) near 00:30 UT where fpe ∼ fce , and it narrows in bandwidth as time progresses. The scaled fpe and fuh frequencies in the region beyond about 00:30 UT, i.e., in the region

1 Active Wave Experiments in Space Plasmas: The Z Mode

31

where fpe > fce , allow the boundaries of the upper-hybrid band to be identiﬁed and distinguished from the slanting ﬁnger-like higher-frequency emissions which are attributed to Bernstein-mode emissions discussed in connection with Fig. 1.4. Figure 1.18 suggests that there are two diﬀerent sources of the observed W-mode emissions, one more intense than the other. The most intense one extends out to slightly beyond 01:00 UT and is limited by the minimum in fpe near 10 kHz as determined from the active sounding. The weaker one extends out to about 01:45 UT and is limited by fce from this point backward in time to 00:30 where it is limited by fpe . At earlier times, the weak emissions in the frequency domain from fpe to fuh could be either L-O or CMA region 6 a Z-mode emissions (see Fig. 1.2b). Conﬁrming identiﬁcations of passive dynamic-spectral features by nearlysimultaneous active determinations of fpe , such as illustrated in Fig. 1.18, provides conﬁdence in the interpretation of the passive dynamic spectra when supporting active measurements are not available. It also provides conﬁdence in the use of the passive dynamic spectra to help interpret plasmagrams when the spectrum of sounder-stimulated resonances is complex [Benson et al., 11].

1.6 Summary Even though the Z mode is an internal, or trapped, mode of the plasma it has valuable diagnostic applications in space plasmas in both active and passive wave experiments. In active experiments discrete Z-mode echo traces can be inverted to provide Ne proﬁles, diﬀuse traces provide information about FAI, and two-point propagation studies provide information concerning wave propagation, wave ducting and wave/particle interactions. In passive experiments, intense Z-mode signals of magnetospheric origin provide valuable ambient Ne information.

Acknowledgements We are grateful to the reviewer for many helpful comments on the manuscript. The work at the University of Alaska Fairbanks was supported by NASA under contract NNG04GI67G. Support for B. W. R. was provided by NASA under subcontract 83822 from SwRI.

References [1] Allis, W.P., S.J. Buchsbaum, and A. Bers: Waves in anisotropic plasmas, MIT Press, Cambridge, 1963.

32

R.F. Benson et al.

[2] Balmain, K.G.: Dipole admittance for magnetoplasma diagnostics, IEEE Trans. Antennas Propag. 17, 389–392, 1969. [3] Bauer, S.J. and R.G. Stone: Satellite observations of radio noise in the magnetosphere, Nature 218, 1145–1147, 1968. [4] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and HF auroral hiss generated at low altitude, J. Geophys. Res., 94, 1359–1379, 1989. [5] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and HF auroral hiss generated at low altitude, J. Geophys. Res. 94, 1359–1379, 1989. [6] Benson, R.F.: Ion eﬀects on ionospheric electron resonance phenomena, Radio Sci. 10, 173–185, 1975. [7] Benson, R.F.: Stimulated plasma waves in the ionosphere, Radio Sci. 12, 861– 878, 1977. [8] Benson, R.F.: Stimulated plasma instability and nonlinear phenomena in the ionosphere, Radio Sci. 17, 1637–1659, 1982. [9] Benson, R.F.: Field-aligned electron density irregularities near 500 km – Equator to polar cap topside sounder Z mode observations, Radio Sci. 20, 477–485, 1985. [10] Benson, R.F.: Evidence for the stimulation of ﬁeld-aligned electron density irregularities on a short time scale by ionospheric topside sounders, J. Atm. and Solar-Terr. Phys. 59, 2281–2293, 1997. [11] Benson, R.F., V.A. Osherovich, J. Fainberg, and B.W. Reinisch: Classiﬁcation of IMAGE/RPI-stimulated plasma resonances for the accurate determination of magnetospheric electron-density and magnetic ﬁeld values, J. Geophys. Res. 108, 1207, doi:10.1029/2002JA009589, 2003. [12] Benson, R.F., P.A. Webb, J.L. Green, L. Garcia, and B.W. Reinisch: Magnetospheric electron densities inferred from upper-hybrid band emissions, Geophys. Res. Lett. 31, L20803,doi:1029/2004GL020847, 2004. [13] Bernstein, I.B.: Waves in a plasma in a magnetic ﬁeld, Phys. Rev. 109, 10–21, 1958. [14] Budden, K.G.: The propagation of radio waves, the theory of radio waves of low power in the ionosphere and magnetosphere, 669 pp., Cambridge University Press, New York, 1985. [15] Burch, J.L.: The ﬁrst two years of IMAGE, Space Sci. Rev. 109, 1–24, 2003. [16] Calvert, W., Oblique z-mode echoes in the topside ionosphere, J. Geophys. Res. 71, 5579–5583, 1966. [17] Calvert, W.: Wave ducting in diﬀerent wave modes, J. Geophys. Res., 100 (A9), 17,491–17,497, 1995. [18] Carpenter, D.L. and R.R. Anderson: An ISEE/whistler model of equatorial electron density in the magnetosphere, J. Geophys. Res. 97, 1097–1108, 1992. [19] Carpenter, D.L., M.A. Spasojevic, T.F. Bell, U.S. Inan, B.W. Reinisch, I.A. Galkin, R.F. Benson, J.L. Green, S.F. Fung, and S.A. Boardsen: Small-scale ﬁeld-aligned plasmaspheric density structures inferred from RPI on IMAGE, J. Geophys. Res., 107(A9), 1258, doi:10.1029/2001JA009199, 2002. [20] Carpenter, D.L., T.F. Bell, U.S. Inan, R.F. Benson, V.S. Sonwalkar, B.W. Reinisch, and D.L. Gallagher: Z-mode sounding within propagation “cavities” and other inner magnetospheric regions by the RPI instrument on the IMAGE satellite, J. Geophys. Res. 108, 1421, doi:10.1029/2003JA010025, 2003.

1 Active Wave Experiments in Space Plasmas: The Z Mode

33

[21] Denisenko, P.F., N.A. Zabotin, D.S. Bratsun, and S.A. Pulinets: Detection and mapping of small-scale irregularities by topside sounder data, Ann. Geophysicae 11, 595–600, 1993. [22] Dyson, P.L. and R.F. Benson: Topside sounder observations of equatorial bubbles, Geophys. Res. Lett. 5, 795–798, 1978. [23] Eckersley, T.L.: Discussion of the ionosphere, Proc. Roy. Soc. A141, 708–715, 1933. [24] Ellis, G.R.: The Z propagation hole in the ionosphere, J. Atmosph. Terr. Phys. 8, 43–54, 1956. [25] Franklin, C.A. and M.A. Maclean: The design of swept-frequency topside sounders, Proc. IEEE 57, 897–929, 1969. [26] Fung, S.F., R.F. Benson, D.L. Carpenter, J.L. Green, V. Jayanti, I.A. Galkan, and B.W. Reinisch: Guided Echoes in the Magnetosphere: Observations by Radio Plasma Imager on IMAGE, Geophys. Res. Lett. 3, 1589, doi:10.1029/2002GL016531, 2003. [27] Goertz, C.K. and R.J. Strangeway: Plasma waves, in: Introduction to Space Physics, edited by M.G. Kivelson, and C.T. Russell, pp. 356–399, Cambridge University Press, New York, 1995. [28] Gurnett, D.A., S.D. Shawhan, and R.R. Shaw: Auroral hiss, Z mode radiation, and auroral kilometric radiation in the polar magnetosphere: DE 1 observations, J. Geophys. Res. 88, 329–340, 1983. [29] Horita, R.E. and H.G. James: Enhanced Z-mode radiation from a dipole, Adv. Space Res. 29, 1375–1378, 2002. [30] Horita, R.E. and H.G. James: Two-point studies of fast Z-mode waves with dipoles in the ionosphere, Radio Sci., 39, RS4001, doi:10.1029/2003RS002994, 2004. [31] Huang, X., B.W. Reinisch, P. Song, P. Nsumei, J.L. Green, and D.L. Gallagher: Developing an empirical density model of the plasmasphere using IMAGE/RPI observations, Adv. Space Res. 33, 829–832, 2004. [32] Jackson, J.E.: The reduction of topside ionograms to electron-density proﬁles, Proc. IEEE 57, 960–976, 1969. [33] Jackson, J.E. and E.S. Warren: Objectives, history, and principal achievments of the topside sounder and ISIS programs, Proc. IEEE 57, 861–865, 1969. [34] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites, 1, General results, Radio Sci. 13, 531–542, 1978. [35] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites 3. Z mode pulses, J. Geophys. Res. 84, 499–506, 1979. [36] James, H.G.: Guided Z mode propagation observed in the OEDIPUS A tethered rocket experiment, J. Geophys. Res. 96, 17,865–17,878, 1991. [37] James, H.G.: Slow Z-mode radiation from sounder-accelerated electrons, J. Atmos. Solar-Terr. Phys. 66, 1755–1765, 2004. [38] James, H.G., Radiation from sounder-accelerated electrons, Adv. Space Res., in press, 2005. [39] Kelso, J.M.: Radio ray propagation in the ionosphere, 408 pp., McGraw Hill, New York, 1964. [40] Kuehl, H.H.: Electromagnetic radiation from an electric dipole in a cold anisotropic plasma, Phys. Fluids 5, 1095–1103, 1962.

34

R.F. Benson et al.

[41] Kuo, S.P., M.C. Lee, and P. Kossey: Excitation of short-scale ﬁeld-aligned electron density irregularities by ionospheric topside sounders, J. Geophys. Res. 104, 19,889–19,894, 1999. [42] Kurth, W.S., G.B. Hospodarsky, D.A. Gurnett, M.L. Kaiser, J.-E. Wahlund, A. Roux, P. Canu, P. Zarka, and Y. Tokarev: An overview of observations by the Cassini radio and plasma wave investigation at Earth, J. Geophys. Res. 106, 30239–30252, 2001. [43] LaBelle, J. and R.A. Treumann: Auroral radio emissions, 1. hisses, roars, and bursts, Space Sci. Rev. 101, 295–440, 2002. [44] Lewis, R.M. and J.B. Keller: Conductivity tensor and dispersion equation for a plasma, Phys. Fluids, 5, 1248–1264, 1962. [45] Lockwood, G.E.K.: A ray-tracing investigation of ionospheric Z-mode propagation, Can. J. Phys. 40, 1840–1843, 1962. [46] McAfee, J.R.: Ray trajectories in an anisotropic plasma near plasma resonance, J. Geophys. Res. 73, 5577–5583, 1968. [47] McAfee, J.R.: Topside ray trajectories near the upper hybrid resonance, J. Geophys. Res. 74, 6403–6408, 1969. [48] Muldrew, D.B.: Radio propagation along magnetic ﬁeld-aligned sheets of ionization observed by the Alouette topside sounder, J. Geophys. Res. 68, 5355–5370, 1963. [49] Muldrew, D.B.: Nonvertical propagation and delayed-echo generation observed by the topside sounders, Proc. IEEE 57, 1097–1107, 1969. [50] Muldrew, D.B.: Electron resonances observed with topside sounders, Radio Sci. 7, 779–789, 1972. [51] Muldrew, D.B. and M.F. Estabrooks: Computation of dispersion curves for a hot magnetoplasma with application to the upper-hybrid and cyclotron frequencies, Radio Sci. 7, 579–586, 1972. [52] Osherovich, V.A., J. Fainberg, R.F. Benson, and R.G. Stone: Theoretical analysis of resonance conditions in magnetized plasmas when the plasma/gyro frequency ratio is close to an integer, J. Atm. and Solar-Terr. Phys. 59, 2361–2366, 1997. [53] Oya, H.: Conversion of electrostatic plasma waves into electromagnetic waves: numerical calculation of the dispersion relation for all wavelengths, Radio Sci. 6, 1131–1141, 1971. [54] Poeverlein, H.: Strahlwege von Radiowellen in der Ionosph¨ are, Z. Angew. Phys. 1, 517, 1949. [55] Pulinets, S.A.: Prospects of topside sounding, in WITS handbook N2, edited by C.H. Liu, pp. 99–127, SCOSTEP Publishing, Urbana, Illinois, 1989. [56] Pulinets, S.A., P.F. Denisenko, N.A. Zabotin, and T.A. Klimanova: New method for small-scale irregularities diagnostics from topside sounder data, in: SUNDIAL Workshop, McLean, Virginia, 1989. [57] Ratcliﬀe, J.A.: The Magneto-Ionic Theory and its Applications to the Ionosphere, 206 pp., Cambridge University Press, New York, 1959. [58] Reinisch, B.W.: Modern Ionosondes, in: Modern Ionospheric Science, edited by H. Kohl, R. Ruster, and K. Schlegel, pp. 440–458, European Geophysical Society, Katlenburg-Lindau, Germany, 1996. [59] Reinisch, B.W., D.M. Haines, K. Bibl, G. Cheney, I.A. Gulkin, X. Huang, S.H. Myers, G.S. Sales, R.F. Benson, S.F. Fung, J.L. Green, W.W.L. Taylor, J.-L. Bougeret, R. Manning, N. Meyer-Vernet, M. Moncuquet, D.L. Carpenter, D.L.

1 Active Wave Experiments in Space Plasmas: The Z Mode

[60]

[61]

[62]

[63] [64] [65] [66]

[67]

[68]

[69]

35

Gallagher, and P. Reiﬀ: The radio plasma imager investigation on the IMAGE spacecraft, Space Sci. Rev. 91, 319–359, 2000. Reinisch, B.W., X. Huang, D.M. Haines, I.A. Galkin, J.L. Green, R.F. Benson, S.F. Fung, W.W.L. Taylor, P.H. Reiﬀ, D.L. Gallagher, J.-L. Bougeret, R. Manning, D.L. Carpenter, and S.A. Boardsen: First results from the radio plasma imager on IMAGE, Geophys. Res. Lett. 28, 1167–1170, 2001a. Reinisch, B.W., X. Huang, P. Song, G.S. Sales, S.F. Fung, J.L. Green, D.L. Gallagher, and V.M. Vasyliunas: Plasma density distribution along the magnetospheric ﬁeld: RPI observations from IMAGE, Geophys. Res. Lett. 28, 4521– 4524, 2001b. Sonwalkar, V.S., D.L. Carpenter, T.F. Bell, M.A. Spasojevic, U.S. Inan, J. Li, X. Chen, A. Venkatasubramanian, J. Harikumar, R.F. Benson, W.W.L. Taylor, and B.W. Reinisch: Diagnostics of magnetospheric electron density and irregularities at altitudes 5. There were almost no Geotail kilometric continuum radiation observations when Kp > 5 in 1996. Figure 2.9 provides dramatic evidence for the enhancement of KC at solar maximum (right panel) over solar minimum (left panel). The cause of this solar cycle diﬀerence is unknown. Both studies show that kilometric continuum radiation is observed even if Kp = 0. In order to check the Kp dependence, an example when kilometric continuum radiation was observed and Kp = 0 is shown in Fig. 2.10. The kilometric continuum is observed from 06–18 UT on Geotail on December 20, 1996, as shown in the top dynamic spectra. The magnetic latitudes of the satellite are shown for one day in the center of the upper right panel next to the spectrogram and the circles indicate the observations of kilometric continuum. Kp indices are shown above the panel for the day in the second line and for the previous day in the ﬁrst line. Kp had been less or equal to 1 for more than 24 hours. This demonstrates that kilometric continuum occurs even in such a very quiet time. Dst had been almost zero as seen from the lower right panel. These observations pose a problem for the energy source of the emissions.

48

K. Hashimoto et al.

Fig. 2.10. Geotail observations for December 20, 1996, when Kp was always less than 1. Kilometric continuum radiation is observed from 06 UT to 18 UT

Fig. 2.11. Geotail observations April 8, 2001, when Kp ranged from 2 to 7. Kilometric continuum radiation was not observed until after 22 UT even though Geotail was within 10 degrees of the equator since 09 UT

Figure 2.11 is an example of occurrence of large Kp without kilometric continuum. The continuum is observed only after 22 UT although the satellite was within 10 degrees of the equator since 09 UT, Kp was large for a long time, and AKR was often observed. It should be noted that the kilometric continuum radiation from about 600 kHz to about 800 kHz was quite strong from 22 UT to 05 UT the following day.

2 Review of Kilometric Continuum

49

2.6 Simultaneous Wave Observations by Geotail and IMAGE Data from the extreme ultraviolet (EUV) imager [47] of the IMAGE satellite demonstrated the existence of notches, where the electron densities are low near the equatorial plasmapause, and the notches are the source of kilometric continuum radiation. The radio plasma imager (RPI) of the satellite indicated that the electromagnetic waves were not only trapped in the low density region but also escape radially out or are generated outside as kilometric continuum. The IMAGE satellite observed the notch near 4 UT in the dayside as shown in Fig. 2.12 [10]. The orbit is the black curve and the plasmapause is shown as the light curve. Near 190◦ in the magnetic longitude, the density is decreased and the plasmapause position was inside the normal one. Kilometric continuum is observed inside the notch region.

Fig. 2.12. Notch observation by IMAGE on April 8, 2001 [10] (Reprinted with permission of the American Geophysical Union)

On the same day, Geotail did not observe kilometric continuum under very high Kp as shown in Fig. 2.11 until 18 hours later. The notch was observed around local noon, but Geotail was in the nightside. This fact indicates that kilometric continuum is radiated during high Kp, but Geotail was not able to observe it. The satellite position could be the cause of the low occurrence probability at high Kp. Simultaneous observations of kilometric continuum with IMAGE RPI and Geotail PWI are displayed in a frequency range of 300 − 800 kHz in the top and the bottom of Fig. 2.13, respectively [Hashimoto et al., 20]. Intense kilometric continuum was received during the disturbed time, especially for Kp > 7 from 20 UT to 03 UT. It should be noted that both kilometric continuum spectra show quite good similarity including the ﬁne structures from 21 UT to 06 UT. IMAGE moved from the southern hemisphere to 30◦ N. On the other hand, Geotail moved in the equatorial region from 4.4◦ N to 12.3◦ N at 01 UT and

50

K. Hashimoto et al.

Fig. 2.13. IMAGE and Geotail observations and their orbits on May 29–30, 2003. Note the similarity of the spectra [20] (Reprinted with permission of the American Geophysical Union)

then back down to 2.4◦ N as shown in the right hand panel of Fig. 2.13. Both satellites observed almost the same spectra in a wide latitude range of more than 30◦ . Their longitudes are close within 10◦ . IMAGE RPI observed the emission in a wide latitudinal range diﬀerent from general trends reported by Hashimoto et al. [19] and Green et al. [10]. The vertical line at 0420 UT is a type III burst. The intensity observed by IMAGE is weaker around 400 kHz after 0430 UT when the satellite is at latitudes higher than 25◦ . It would be diﬃcult to explain these quite similar spectra by multiple narrow beam sources. Rather, this can be explained if the sources radiate uniformly in wide emission cones in latitude and both satellites receive the emissions from the same sources, which is contrary to the beaming theory.

2.7 Summary and Conclusions Non-thermal continuum radiation is one of the fundamental electromagnetic emissions in planetary magnetospheres [cf. the review by Kaiser, 27]. It has been observed in every planetary magnetosphere visited by spacecraft armed with wave instruments and has even been found to be generated in the magnetosphere of the Galilean moon Ganymede [Kurth et al., 33]. Although

2 Review of Kilometric Continuum

51

this emission has been observed and studied for more than 35 years, there are still several unveriﬁed theories on how this emission is generated. There is also much more which we do not know about this emission and its relationship to the dynamics of the plasmasphere. Many of the characteristics of the lower frequency portion of the nonthermal continuum (trapped component) have been diﬃcult to determine due to the multiple reﬂections of the emission from the magnetopause and plasmapause. Recently there is renewed interest in studying the high frequency extension of this emission (the escaping component), especially the extension into the kilometric frequency range. Kilometric continuum has been reported to be observed by Polar and Cluster [Menietti et al., 39, 40] and INTERBALL-1 [Kuril’chik et al., 34, 35] in addition to Geotail, IMAGE, and CRRES [10, 11, 19, 20, and the present paper]. It has been conﬁrmed that the kilometric continuum is generated at steep density gradients at density irregularities in the equatorial region. These irregularities do not only exist at the plasmapause, but also inside the plasmapause and in notches. Although the observations are consistent with the mode conversion mechanism at the plasma frequency, they are not consistent with the beaming model of Jones [22, 23]. The simultaneous observations given in Fig. 2.13 provide striking evidence against the latter although this is discussed in more detail by Hashimoto et al. [20]. The relations to solar and geomagnetic activities are also interesting topics. Several new features of the high frequency escaping kilometric continuum, such as the narrow latitudinal beam structure and relationship to plasmaspheric notch or notch structures, provide a new opportunity to observe the triggering of this emission and its relationship to plasmaspheric dynamics. The insight gained in performing multi-spacecraft correlative measurements should provide key measurements on separating spatial from temporal eﬀects that are essential to verifying existing theories. Observing the radiation while the instability has been initiated and grows, and examining the dynamics of the large-scale plasmasphere should lead to signiﬁcant advances in delineating the best theory for the generation of this emission.

References [1] Anderson, R.R., D.A. Gurnett, and D.L. Odem: CRRES plasma wave experiment, J. Spacecraft Rockets 29, 570, 1992. [2] Barbosa, D.D.: Low-level VLF and LR radio emissions observed at earth and Jupiter, Rev. Geophys. Space Phys. 20, 316, 1982. [3] Carpenter, D.L. and R.R. Anderson: An ISEE/whistler model of equatorial electron density in the magnetosphere, J. Geophys. Res. 97, 1097, 1992. [4] Carpenter, D.L., R.R. Anderson, W. Calvert, and M.B. Moldwin: CRRES observations of density cavities inside the plasmasphere, J. Geophys. Res. 105, 23323, 2000.

52

K. Hashimoto et al.

[5] D´ecr´eau, P.M.E. et al.: Observation of continuum radiations from the Cluster ﬂeet: First results from direction ﬁnding, Ann. Geophys. 22, 2607, 2004. [6] Filbert, P.C. and P.J. Kellogg: Observations of low-frequency radio emissions in the earth’s magnetosphere, J. Geophys. Res. 94, 8867, 1989. [7] Frankel, M.S.: LF radio noise form the Earth’s magnetosphere, Radio Sci. 8, 991, 1973. [8] Fung, S.F. and K. Papadopoulos: The emission of narrow-band Jovian kilometric radiation, J. Geophys. Res. 92, 8579, 1987. [9] Green, J.L. and S.A. Boardsen: Conﬁnement of nonthermal continuum radiation to low latitudes, J. Geophys. Res. 104, 10307, 1999. [10] Green, J.L., B.R. Sandel, S.F. Fung, D.L. Gallagher, and B.W. Reinisch: On the origin of kilometric continuum, J. Geophys. Res. 107, doi: 10.1029/ 2001JA000193, 2002. [11] Green, J.L. et al.: Association of kilometric continuum radiation with plasmaspheric structures, J. Geophys. Res. 109, A03203, doi: 10.1029/2003JA010093, 2004. [12] Green, J.L. and S.F. Fung: Advances in Inner Magnetospheric Passive and Active Wave Research, In: AGU Monograph on Physics and Modeling of the Inner Magnetosphere, AGU, Washington D.C., in press, 2005. [13] Gough, M.P.: Nonthermal continuum emissions associated with electron injections: Remote plasmapause sounding, Planet. Space Sci. 30, 657, 1982. [14] Gurnett, D.A. and R.R. Shaw: Electromagnetic radiation trapped in the magnetosphere above the plasma frequency, J. Geophys. Res. 78, 8136, 1973. [15] Gurnett, D.A.: The Earth as a radio source: The nonthermal continuum, J. Geophys. Res. 80, 2751, 1975. [16] Gurnett, D.A. and L.A. Frank: Continuum radiation associated with low-energy electrons in the outer radiation zone, J. Geophys. Res. 81, 3875, 1976. [17] Gurnett, D.A., W. Calvert, R.L. Huﬀ, D. Jones, and M. Sugiura: The polarization of escaping terrestrial continuum radiation, J. Geophys. Res. 93, 12817, 1988. [18] Hashimoto, K., K. Yamaashi, and I. Kimura: Three-dimensional ray tracing of electrostatic cyclotron harmonic waves and Z mode electromagnetic waves in the magnetosphere, Radio Science 22, 579, 1987. [19] Hashimoto, K., W. Calvert, and H. Matsumoto: Kilometric continuum detected by Geotail, J. Geophys. Res. 104, 28645, 1999. [20] Hashimoto, K., R.R. Anderson, J.L. Green, and H. Matsumoto: Source and Propagation Characteristics of kilometric continuum observed with multiple satellite, J. Geophys. Res., 110, A09229, doi: 10.1029/2004JA010729, 2005. [21] Jones, D.: Source of terrestrial nonthermal radiation, Nature 260, 686, 1976. [22] Jones, D.: Latitudinal beaming of planetary radio emissions, Nature 288, 225, 1980. [23] Jones, D.: Beaming of terrestrial myriametric radiation, Adv. Space Res. 1, 373, 1981. [24] Jones, D.: Terrestrial myriametric radiation from the Earth’s plasmapause, Planet. Space Sci. 30, 399, 1982. [25] Jones, D., W. Calvert, D. A. Gurnett, and R. L. Huﬀ: Observed beaming of terrestrial myriametric radiation, Nature 328, 391, 1987.

2 Review of Kilometric Continuum

53

[26] Jones, D.: Planetary radio emissions from low magnetic latitudes – Observations and theories, In: Planetary Radio Emissions II (H.O. Rucker, S.J. Bauer, and B.M. Pedersen. Eds., Austrian Acad. Sci., Vienna) p. 245, 1988. [27] Kaiser, M.L.: Observations of non-thermal radiation from planets, In: Plasma Waves and Instabilities at Comets and in Magnetospheres, (B. Tsurutani and H. Oya, Eds., AGU, Washington) p. 221, 1989. [28] Kasaba, Y., H. Matsumoto, K. Hashimoto, R.R. Anderson, J.-L. Bougeret, M.L. Kaiser, X.Y. Wu, and I. Nagano: Remote sensing of the plasmapause during substorms: GEOTAIL observation of nonthermal continuum enhancement, J. Geophys. Res. 107, 20389, 1998. [29] Kurth, W.S., J.D. Craven, L.A. Frank, and D.A. Gurnett: Intense electrostatic waves near the upper hybrid resonance frequency, J. Geophys. Res. 84, 4145, 1979. [30] Kurth, W.S., D.A. Gurnett, and R.R. Anderson: Escaping nonthermal continuum radiation, J. Geophys. Res. 86, 5519, 1981. [31] Kurth, W.S.: Detailed observations of the source of terrestrial narrowband electromagnetic radiation, Geophys. Res. Lett. 9, 1341, 1982. [32] Kurth, W.S.: Continuum radiation in planetary magnetospheres, In: Planetary Radio Emissions III (H. O. Rucker, S. J. Bauer, and M. L. Kaiser, Eds., Austrian Acad. Sci., Vienna) p. 329, 1992. [33] Kurth, W.S., D.A. Gurnett, A. Roux, and S.J. Bolton: Ganymede: A new radio source, Geophys. Res. Lett. 24, 2167, 1997. [34] Kuril’chik, V.N., M.Y. Boudjada, and H.O. Rucker: Interball-1 observations of the plasmaspheric emissions related to terrestrial “continuum” radio emissions, In: Planetary Radio Emissions V (H.O. Rucker, M.L. Kaiser and Y. Leblanc, Eds., Austrian Acad. Sci., Vienna) p. 325, 2001. [35] Kuril’chik, V.N., I.F. Kopaeva, and S.V. Mironov: INTERBALL-1 observations of the kilometric “continuum” of the Earth’s magnetosphere, Cosmic Research 42, 1, 2004. [36] Lee, L.C.: Theories of non-thermal radiations from planets, In: Plasma Waves and Instabilities at Comets and in Magnetospheres (B. Tsurutani and H. Oya, Eds., AGU, Washington) p. 239, 1989. [37] Matsumoto, H., I. Nagano, R.R. Anderson, H. Kojima, K. Hashimoto, M. Tsutsui, T. Okada, I. Kimura, Y. Omura, and M. Okada: Plasma wave observations with GEOTAIL spacecraft, J. Geomagn. Geoelectr. 46, 59, 1994. [38] Melrose, D.B.: A theory for the nonthermal radio continuum in the terrestrial and Jovian magnetospheres, J. Geophys. Res. 86, 30, 1981. [39] Menietti, J.D., R.R. Anderson, J.S. Pickett, D.A. Gurnett, and H. Matsumoto: Near-source and remote observations of kilometric continuum radiation from multi-spacecraft observations, J. Geophys. Res. 108, 1393, doi: 10.1029/2003JA009826, 2003. [40] Menietti, J.D., O. Santolik, J.S. Pickett, and D. A. Gurnett: High resolution observations of continuum radiation, Planet. Space Sci. 53, 283, 2005. [41] Morgan, D.D. and D.A. Gurnett: The source location and beaming of terrestrial continuum radiation, J. Geophys. Res. 96, 9595, 1991. [42] Okuda, H., M. Ashour-Abdalla, M.S. Chance, and W.S. Kurth: Generation of nonthermal continuum radiation in the magnetosphere, J. Geophys. Res. 87, 10457, 1982.

54

K. Hashimoto et al.

[43] Oya, H., M. Iizima, and A. Morioka: Plasma turbulence disc circulating the equatorial region of the plasmasphere identiﬁed by the plasma wave detector (PWS) onboard the Akebono (EXOS-D) satellite, Geophys. Res. Lett. 18, 329, 1991.) [44] Reinisch, B.W. et al.: The Radio Plasma Imager investigation on the IMAGE spacecraft, Space Sci. Rev. 91, 319, 2000. [45] R¨ onnmark, K.: Emission of myriametric radiation by coalescence of upper hybrid waves with low frequency waves, Ann. Geophys. 1, 187, 1983. [46] R¨ onnmark, K.: Conversion of upper hybrid waves into magnetospheric radiation, In: Planetary Radio Emissions III (H.O. Rucker, S.J. Bauer, and M.L. Kaiser, Eds., Austrian Acad. Sci., Vienna) p. 405, 1992. [47] Sandel, B.R., R.A. King, W.A. King, W.T. Forrester, D.L. Gallagher, A.L. Broadfoot, and C.C. Curtis: Initial results from the IMAGE Extreme Ultraviolet Imager, Geophys. Res. Letts. 28, 1439, 2001. [48] Vesecky, J.F. and M.S. Frankel: Observations of a low-frequency cutoﬀ in magnetospheric radio noise received on IMP 6, J. Geophys. Res. 80, 2771, 1975. [49] Yamaashi, K., K. Hashimoto, and I. Kimura: 3-D electrostatic and electromagnetic ray tracing in the magnetosphere, Mem. Natl. Inst. Polar Res., Spec. Issue 47, 192, 1987.

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions P. Louarn Centre d’Etudes Spatiales du Rayonnement, Toulouse, France [email protected]

Abstract. Owing to the complete and precise wave/particle measurements performed by the Viking and FAST spacecraft in the sources of the terrestrial Auroral Kilometric Radiation, the conditions of generation of this coherent radio emission are now well documented. It has been demonstrated that the production of radio waves occurs in thin laminar regions that exactly correspond to the regions of auroral particle acceleration. We present and discuss observations, mainly performed by Viking, that have led to the conception of a model of wave generation by ﬁnite geometry sources. The corresponding theoretical analysis is also presented. This model has certainly a very broad domain of applications, including radio emissions from other planets, from the Sun and more distant astrophysical objects.

Key words: Auroral Kilometric Radiation, free space radiation, magnetoionic modes, eﬀect of inhomogeneity, cyclotron maser mechanism, loss cone distribution, horseshoe, ring distribution

3.1 Introduction As seen from deep space, the Earth is a powerful natural radio source emitting 107 to 108 W with a maximum spectral density at ∼250 kHz. Its dominant emission – the Auroral Kilometric Radiation (AKR hereafter) – is generated in the night sector of the auroral zone, at magnetic latitudes larger than 65◦ and altitudes ranging from 5000 km to 15000 km. Its power increases with magnetospheric activity, especially when substorms develop. This has been realized ﬁrst by Benediktov et al. [6] and conﬁrmed by others [9, 20, 21, 23]. This radiation is characterized by a brightness temperature (>1015 K) much larger than any plasma temperature in planetary magnetospheres. It cannot be explained by a thermal generation mechanism and coherent emission processes such as plasma instabilities must be invoked for its generation. The AKR is not a singular phenomenon. In its non-thermal origin, high polarization, temporal variability and spectral ﬁne structure AKR is similar P. Louarn: Generation of Auroral Kilometric Radiation in Bounded Source Regions, Lect. Notes Phys. 687, 55–86 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

56

P. Louarn

to the radio emissions that emanate from other magnetospheres, at Jupiter, Saturn, Uranus and Neptune [see, e.g., Zarka, 60]. It can be compared to solar radio bursts (microwave spike bursts) [1, 15] and stellar radio emissions [32, 39]. The AKR generation mechanism has thus to be considered as a general process able to eﬃciently convert some forms of free energy present in magnetized plasmas into radiating electromagnetic waves. Due to the proximity of its sources, the study of the AKR oﬀers a splendid opportunity to test all the details of this radiation process. It was actually possible to determine some important characteristics of AKR from early space measurements [6, 21, 22]. The evidence for a dominant X polarization came from Voyager measurements [Kaiser et al., 26], as the probes began their odyssey. This was conﬁrmed later from DE-1 measurements published by Shawhan and Gurnett [55] and Mellott et al. [38], although the existence of a fainter O mode was also detected. The good correlation between inverted V precipitation and AKR [Green et al., 20] has given support to the idea that the AKR is generated by electrons accelerated above the auroral regions. Important progress has been made by the crossing of the sources regions themselves, ﬁrst reported by Benson and Calvert [7] using ISIS 1 spacecraft. These direct in-situ measurements have revealed that the AKR is generated: • in strongly magnetized regions (fp /fc 1 where fp is the plasma frequency and fc the electron gyrofrequency), • at frequencies close to the local cut-oﬀ of the X mode: fx = 12 [fc + (fc2 + 4fp2 )1/2 ], • with waves propagating at large angles with respect to the geomagnetic ﬁeld B0 [7, 8]. Numerous theories have been proposed for the interpretation of this coherent radiation: nonlinear mode coupling, soliton radiation and linear instabilities. In view of the observational constraints summarized above, a linear instability – the cyclotron maser – ﬁrst proposed by Wu and Lee [59] and Lee and Wu [27] and further developed by several authors [28, 29, 30, 40, 41, 42], has been acknowledged by a majority as being the most promising generation mechanism. Though this relativistic instability was known before [see Trubnikov, 57], the main progress realized by Wu and Lee [59] was to show that the relativistic correction in the electron motion is crucial even at the modest keV energies of auroral electrons. In the tenuous auroral plasmas, the X mode cut-oﬀ frequency is indeed just a few percent above fc and the relativistic correction must be considered in the wave/particle resonant condition: ω − k v − ωc /Γ = 0 where Γ = (1 − 1 v 2 /c2 )− 2 is the Lorentz factor and k the wave vector component parallel to the static magnetic ﬁeld. The relativistic eﬀects permit an exchange of energy between X mode waves and electrons. It can be shown that the plasma may act as a coherent radiating source if the electron distribution presents an inversion of population in the form of positive gradients ∂f /∂v⊥ > 0. Since the most common distribution functions observed in the auroral zone – the loss-cone

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

57

distributions – precisely present this type of free energy, an eﬃcient X mode ampliﬁcation may take place at frequencies close to fc . This good agreement between theoretical predictions and observations has explained the success of this direct ampliﬁcation mechanism in the community. The cyclotron maser model was thus soon recognized as the most plausible generation mechanism of non-thermal radio emission radiated by highly magnetized planetary, solar and even stellar plasmas. Our understanding of the AKR generation was greatly improved by the measurements made by the Swedish satellite Viking in the heart of the AKR sources. It was possible to understand the connection between the particle acceleration and the production of the radiations and to get a precise knowledge of the plasma conditions inside the sources and in their near vicinity. As it is reviewed here, these observations have motivated new theoretical analyses that were severely tested. These studies have contributed to establish a sophisticated model of the AKR generation and, by extension, of the planetary radio emissions. Ten years after Viking, the superb FAST measurements have conﬁrmed and updated this picture of wave ampliﬁcation in thin laminar sources. In Sect. 3.2, we present some of the Viking observations performed in the AKR sources. The observational eﬀects linked to the ﬁnite geometry of the sources are discussed in Sect. 3.3. The theoretical analysis in made in Sect. 3.4 before the discussion and the conclusion proposed in Sect. 3.5.

3.2 Spacecraft Observations in the Sources of the Auroral Kilometric Radiation 3.2.1 Structure of Sources and Wave Properties The Viking spacecraft was the ﬁrst to explore, with a complete set of experiments, the central part of the AKR sources, at altitudes of the order of one Earth’s radius (RE ) [see the review by Roux et al., 53]. Frequency-time spectrograms corresponding to AKR source crossings by Viking have been presented by Bahnsen et al. [2, 3], Pottelette et al. [44], Louarn et al. [33], and Ungstrup et al. [58]. Figure 3.1 shows such a source crossing. The source is deﬁned in the data by an intensiﬁcation of the signal at frequencies very close to fc . For the example shown in Fig. 3.1 (orbit 165), this occurs from ∼2033:00 to ∼2033:30 UT. During this 30 s time period, the spectral density of the AKR is maximum just at the frequency channel that contains fc . Two additional sources are also crossed during short time periods, at ∼2032:30 and 2034:00 UT. For the purpose of a statistical study, Hilgers et al. [24] computed the integrated electric energy EAKR in the vicinity of the electron gyrofrequency (from fc − 5 kHz to fc + 10 kHz) for the whole set of Viking data. They deﬁned the sources as the regions along the orbit where this quantity is maximum.

58

P. Louarn

Fig. 3.1. An example of source crossed by Viking. The AKR source is indicated and is characterized by an increase of the AKR wave power at and, in some cases, even below the gyrofrequency

Almost 40 source crossings have been identiﬁed using this method. The statistical wave properties inside or near the sources were then determined with the following results: • The lowest frequency peak fpeak of the AKR is on average a few percent above fc , but it may also be up to 3% below it. • To determine the lowest frequency of the AKR, flc is deﬁned as the ﬁrst frequency below fpeak where the spectral energy has decreased by 10%. This frequency is signiﬁcantly below fpeak , by 2% in the average. A significant power is thus produced down to and often below fc . • The refractive index is close to unity everywhere but close to the cutoﬀ where values smaller than unity are measured [Bahnsen et al., 3]. • Concerning the polarization, Hilgers et al. [25] showed that the orientation of the electric ﬁelds is not the same within the source at f ∼ fc and outside the sources at f fc . Inside the sources, the modulation pattern due to the spin of the spacecraft is consistent with a wave electric ﬁeld being conﬁned within 10◦ in the plane perpendicular to B0 , as expected for X mode waves propagating close to the direction perpendicular to B0 . Outside the sources, the modulation pattern is less pronounced. The waves being transverse, this suggests that they propagate at a more oblique angle with respect to B0 . Concerning the source extension and its lifetime, the fact that wellidentiﬁed AKR patterns are seen in the dynamic spectrograms over several tens of minutes shows that the source exists for at least that long, essentially at the same location and, even on the same ﬁeld lines. In most situations, the spectrum covers a broad frequency range. Since the emission is generated at or very close to fc , it is possible to translate the frequency bandwidth into an altitude range for the source. The 100–200 kHz bandwidth of the AKR then corresponds to an extension of the order of 2500 km below the spacecraft (at a typical altitude of 6000 km). As discussed by De Feraudy et al. [13], the

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

59

observation of broad spectra also implies that propagation takes place in a ﬁlled cone or, more likely, that the source is extended in east-west direction along the auroral oval. The north-south extent is always small, often less than 100 km, and is directly measured since the satellite orbit is nearly in the meridian plane. Altogether, this suggests that the AKR sources are relatively long lived regions, at a mean altitude of 6000 km, with a dimension along the Earth magnetic ﬁeld of ∼2000 km and more. They would thus be ribbons or laminar structures, limited to a width of 20–200 km north-south and much more extending east-west. 3.2.2 AKR Sources as Regions of Particle Accelerations An example of an AKR source was discussed in detail by Louarn et al. [33]. It occurred during the orbit 849 and has lasted a suﬃciently long time to permit the complete measurements of the distribution functions. The corresponding wave and particle measurements are displayed in Fig. 3.2. Using the criteria presented in Sect. 3.2.1, the source crossing occurred in between the two vertical red lines. This time period corresponds to important modiﬁcations in the electron and ions energy spectrograms. The energy of the electrons is maximum just outside the source where it reaches 10 keV. This energy decreases to 5 keV inside the source as upward propagating ion beams are simultaneously detected with energies of ∼5 keV. As discussed in Louarn et al. [33], this can be explained by assuming that the AKR source is an acceleration region characterized by V-shaped iso-potential surfaces. These observations indeed suggest that Viking was successively below, inside, and again below a zone of nonzero parallel electric ﬁeld, this acceleration region coinciding with the AKR source (see the lower panels in Figs. 3.2 and 3.3). As discussed below, the fact that the sources are regions of parallel particle acceleration has several important consequences regarding the wave generation mechanism. 3.2.3 AKR Sources as Plasma Cavities That regions of nonzero parallel electric ﬁeld are regions where the plasma density is perturbed is not a surprise. This has, nevertheless, a special importance here. The plasma density is a crucial parameter for the cyclotron maser instability: it contributes to determine the diﬀerent regimes of the process and the properties of the most unstable waves. Diﬀerent methods can be used to obtain the density. For example, the low frequency emission seen below 50 kHz in Figs. 3.1 and 3.2 (the auroral “hiss”) are whistler waves with a upper-cutoﬀ frequency at the local plasma frequency [see the comparison with relaxation sounder measurements by Perraut et al., 43]. This cut-oﬀ is larger than 20 kHz outside the source. This corresponds to densities above 6 cm−3 . It clearly decreases in the source (see the line in Fig. 3.2). A careful analysis shows that it at that point approaches or even

60

P. Louarn

kHz 400

Source

200

100

Maximum AKR level Plasma cavity

Energy decrease

Upward ion beams

Source

U-shaped potential Spacecraft trajectory

E||

Fig. 3.2. Wave and particle measurements performed by Viking inside a source of AKR. The particle measurements are fully consistent with the crossing of a region of particle acceleration. The white line in the upper panel is the gyro frequency

goes below 10 kHz, the lower- frequency threshold of the receiver. The density of the plasma is then below 1.1 cm−3 in the central part of the source. This observation can be generalized to other source crossings. Hilgers et al. [24] have even shown that the auroral small-scale cavities with densities typically of the order of or less than 1.5 cm−3 detected by Viking coincide almost systematically with AKR sources. To get more indications on the plasma parameters, Louarn et al. [33] have estimated the electron density within and in the vicinity of the source from the electron detector. While the density of high-energy (1–40 keV) electrons remains almost constant across the source, the density of low-energy electrons (below 1 keV) decreases by a factor larger than 10. The density variation is thus associated with a large decrease of the density of cold plasma. A

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

61

Fig. 3.3. A schematic representation of the AKR source/auroral acceleration region

simple interpretation would be that the low energy particles coming from the ionosphere have been evacuated from the source by the accelerating electric ﬁeld. 3.2.4 Free Energy for the Maser Process Let us determine the free energy that may drive the maser. At ﬁrst glance, the problem could be reduced to the search for positive ∂f /∂v⊥ slopes in the distribution functions. Nevertheless, the determination of the non-thermal feature that drives the maser is ambiguous: • several ∂f /∂v⊥ features are able to feed the instability, and • the diﬀusion associated with the waves can level out, at least partly, the “operating” ∂f /∂v⊥ slopes. Theoretical arguments are thus useful here for the identiﬁcation of the most eﬃcient form of free energy. In Fig. 3.4 examples of non-thermal auroral electron distribution functions are given, together with the properties of the waves that can be ampliﬁed via

62

P. Louarn

v

Loss Cone

T

Resonant Ellipse

v|| upward radiation f > fce , in dominant cold plasma v

Hole, Horseshoe

T

Resonant Ellipse

v|| upward radiation f > fce , in dominant cold plasma v

T

Resonant Ellipse

Trapped

v||

perpendicular radiation f < fce , in dominant energetic plasma Fig. 3.4. Forms of free energy that may drive the maser process and expected properties of the unstable waves. In gray: regions of positive ∂f /∂v⊥

the cyclotron maser instability by such free energy sources. A crucial parameter for determining the dominant free energy source is the proportion of hot electrons in the plasma. In a plasma dominated by cold electrons, X mode waves propagate above the electron gyrofrequency and the resonance condition, ω − k v − ωc /Γ = 0, can only be fulﬁlled if k is not null. Therefore, the resonant curve – an ellipse in the v , v⊥ phase space – is not centered at v = 0, v⊥ = 0 and the growth rate which is proportional to the integral of v⊥ ∂f /∂v⊥ along the resonant curve is large only if free energy exists for v = 0. This is the case of the “loss cone” and the “hole” distribution functions, as indicated in Fig. 3.4. Conversely, in an energetic plasma with no or a very tenuous cold component, X mode propagation is possible below the electron gyro-frequency. The Doppler shift term (k v ) is no longer required to

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

63

fulﬁll the resonant condition and free energy centered at v = 0 can be used. The detailed theoretical analysis even shows that, for a given energy of the particles, inversions of populations centered at v = 0 are the most eﬃcient to drive the maser. The three-dimensional plots displayed in Fig. 3.5 show the electron distribution functions just outside and inside an AKR source. The distribution function measured just outside the source is characterized by a dense thermal core and an enhanced population of energetic particles up to 10 keV. Two kinds of ∂f /∂v⊥ non-thermal features are clearly apparent: a loss cone, for electrons moving upward and a “hole” (Fig. 3.5b) for electrons moving downward. Note that the “hole” distribution functions are similar to the “horse-shoe” distributions described by Delory et al. [14], using FAST data. The source of the AKR, as identiﬁed from the criteria discussed in Sect. 3.2.1, does not coincide with these non-thermal features. This suggests that neither a loss cone nor a hole alone suﬃces to generate the AKR. Inside the potential structure (distributions D, E, F), the thermal core is highly depleted. The loss cone and the hole are still present but less pronounced than outside the source. The most prominent feature is a broad plateau in v⊥ (with sometimes a faint indication of positive ∂f /∂v⊥ slopes), which indicates that electrons tend to accumulate in this region of the phase space. This characteristic feature of the distribution functions inside the source leads Louarn et al. [33] to propose that the trapped electrons are the free energy source of the AKR. As shown by Eliasson et al. [16] and by Louarn et al. [33, 34], this trapped electron component can be produced by a time-varying or by a spatially-varying parallel electric ﬁeld. 3.2.5 FAST Observations To date, the FAST satellite, launched in 1996, was the most recent spacecraft to explore the high altitude auroral zones. It crossed the source regions of the AKR with experiments characterized by a remarkable time resolution, the electron distributions being measured in less than a few 10 ms for example [see, 12, 19]. FAST has largely conﬁrmed the results of Viking with even more precision on several crucial points [Ergun et al., 17, 18]: • the correspondence between the acceleration regions and the sources of AKR, • the fact that only accelerated electrons are present in the source, the low energy component being completely evacuated, • the presence of the trapped distribution, • the relatively small dimension of the sources, • the ﬁne temporal/spectral structuring of the radiations, in bursts of several tenths of seconds and frequency bandwidth thinner than 1 kHz. This point was not accessible to Viking observations. These observations have reinforced the status of the cyclotron maser process as the most plausible wave generation mechanism. It was nevertheless

64

P. Louarn

Fig. 3.5. Distribution functions measured outside and inside the AKR source. Different forms of free energy are indicated. The trapped population in only found in the source

advocated that horseshoe distribution functions, thus diﬀerent from the trapped distribution, would be used by the maser [see Delory et al., 14]. Horseshoe distributions result from the same processes as trapped distributions: a combination of electric ﬁeld acceleration and the motion in phase space due to the conservation of the ﬁrst adiabatic invariant. This cannot be considered as a fundamental diﬀerence between the Viking and the FAST observations. 3.2.6 Summary To conclude, an important result from Viking was to reveal that the AKR sources correspond to acceleration regions. They are laminar structures with

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

65

a small north-south extension (a few 10 km, typically). The plasma in these regions is tenuous (n ∼ 1 cm−3 ) and essentially constituted by energetic particles ( E = 3 − 10 keV). The sources are separated from the denser and colder external plasma (n ∼ 5−10 cm−3 , E ∼ a few 10 eV) by sharp density gradients with typical scale lengths smaller than 1 km. An important feature of the electron distribution functions observed inside the sources is an electron accumulation at low parallel velocities and keV energies. These electrons are trapped between their magnetic mirror point and an electrostatic reﬂection point, leading to distributions presenting positive ∂f /∂v⊥ slopes at low v and large v⊥ . The theory shows that this constitutes a very eﬃcient source of free energy for the maser. In the tenuous plasma that ﬁlls the sources and even at moderate energy ( E < 1 keV), it can be shown that relativistic eﬀects must be taken into account in the full X mode propagation and not only in the perturbation eﬀect that leads to wave ampliﬁcation [47, 48, 49, 56]. The relativistic dispersion is then the “zero order” theory for any realistic model of the generation of the AKR. The cyclotron maser instability is just the relativistic negative-absorption eﬀect linked to the presence of positive ∂f /∂v⊥ slopes in the electron distributions. However, the limited extension of the sources modiﬁes the properties of the generation process, with observable consequences regarding the radiated waves. An ensemble of papers [35, 36, 50, 51, 52] have considered this speciﬁc problem of AKR generation in ﬁnite geometry sources. The observation of ﬁnite geometry eﬀects, the model of laminar sources, and its mathematical analysis are discussed in the next sections.

3.3 Cyclotron Maser in Finite Geometry Sources The ﬁrst detailed mentioning of ﬁnite geometry eﬀects on AKR generation was due to Calvert [10, 11]. He proposed that the ﬁnest structures in the AKR observed by ISIS 1 could be explained by a feedback lasing eﬀect due to wave reﬂection in the sources. However, this model is considerably diﬀerent from the one discussed here. Before Viking, the AKR source regions were indeed assimilated to the auroral “cavity”, a generic term that designates a wide region of low plasma density corresponding, as a whole, to the region of auroral precipitations. The lasing eﬀect studied by Calvert was supposed to take place due to reﬂections on the edges of this large scale region. The sources explored by Viking are actually much smaller and are embedded in this large scale auroral cavity. 3.3.1 A Simple Model of AKR Sources We have chosen the simplest model of the sources (see Fig. 3.6). They are supposed to be laminar structures, limited in one direction perpendicular to the static B ﬁeld (x direction, the source width being 2l) and unlimited in the y-z directions (z being the direction of B0 ). Inside the source, the electron

66

P. Louarn

Fig. 3.6. A simple slab model for the sources of AKR

population is constituted by energetic particles with an idealized ring-like distribution: f (v , v⊥ ) = (2πv0 )−1 δ(v )(v⊥ − v0 ). The external plasma is cold and the source boundaries are sharp density gradients. The idealized distribution neglects the thermal eﬀects and could appear oversimpliﬁed. However, as shown by Pritchett [47], Strangeway [56], and Le Qu´eau and Louarn [31], it leads to dispersion equations that take into account the relativistic eﬀects without mathematical complications. The use of more realistic distributions does not signiﬁcantly improve the model. For a ﬁrst approach of ﬁnite geometry eﬀects, this simple description can be considered as precise enough. Only four parameters must be taken into account: • • • •

the the the the

density inside the source: ni (plasma frequency: ωpi ), energy of the internal electron population: E = 12 mv02 , density of the external plasma: n0 (plasma frequency: ωp0 ) and, width (l) of the source.

The ionic component is supposed to be a motionless neutralizing background. We will use the following normalized parameters: i = (ωpi /ωc )2 , o = (ωpo /ωc )2 , δ = (v0 /c)2 and L = ωc l/c. For a typical source, ni ∼ 1 cm−3 , no ∼ 10 cm−3 , energy of the electron ∼5 keV, fc ∼ 200 kHz and l ∼ 15 − 30 km; these parameters are: i = 0.2×10−2 , o = 2×10−2 , δ = 10−2 and L = 60. The non-homogeneity of the geomagnetic ﬁeld is another important parameter. For a dipolar ﬁeld: one has B(z) ∼ B0 (l − z/H); with H = R/3 where R is the distance to the center of the planet. 3.3.2 Generation in Plasma Cavities: A Simple Approach Some of the physical eﬀects linked to ﬁnite geometry can be discussed in a simple way by comparing the relativistic dispersion curve and the “cold” one. As it is calculated in Sect. 3.4.1, the relativistic equation that describes the

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

67

wave propagation at frequencies close to fc has two solutions [Le Qu´eau and Louarn, 31]:

2 1 − 2χ − N2 k⊥ c 2 2 (3.1) = 1 − χz and N⊥ = N⊥ = ωc 1−χ where χz = (ωp /ωc )2 , and χ = 12 [(∆ω−δN2 )]/(∆ω+δ)2 . The ﬁrst equation is the usual dispersion equation of the ordinary mode (O mode), here developed at the zero order in ∆ω = (ω−ωc )/ωc . The second equation corresponds to the extraordinary (X or Z) mode. In Fig. 3.7, the corresponding dispersion curves are displayed, both for an energetic plasma (left panel) and a cold plasma (right panel). At leading order in ∆ω and i,o (which are small quantities of the order of 10−2 throughout this study), the cut-oﬀ and the resonance frequencies assume simple expressions: (1) in cold plasma, the X mode cut-oﬀ and Z mode resonance are ∆ωX = o /2 and ∆ωZ = o /4, (2) in relativistic plasma, they become: ∆ωX = o /2 − δ and ∆ωZ = o /4 − δ.

Re ω

Re ω

Re ω

Im ω

Unstable

X-mode O-mode N Z-mode

X-mode

O-mode

N Z-mode

Relativistic plasma inside source

Cold plasma outside source

Fig. 3.7. Dispersion curves in relativistic and cold plasmas, for quasi-perpendicular propagation. Only a narrow frequency band centered at the gyrofrequency is considered here. The width of the unstable domain is δ

At ﬁrst order in δ, the O-mode dispersion is not modiﬁed by relativistic eﬀects and is stable. Conversely, due to the resonant denominator (∆ω + δ)2 , an instability develops on the respective frequency domains [∆ωX , δ] for the X mode and [−δ, ∆ωZ ], for the Z mode [Le Qu´eau and Louarn, 31]. In the complex frequency plane, the unstable frequencies lie on the circle ||∆ω|| = δ. The maximum growth rate (Im [∆ω] = δ) is obtained for Re [∆ω] = 0. Let us note the existence of an energy threshold: for δ < 3i /8 the maximum growth rate is on the Z mode, when for δ > 3i /8 it is on the X mode. Taking i = 2.5 × 10−2 (corresponding to fp = 10 kHz and fc = 200 kHz), this critical electron energy is δ = 0.094 × 10−3 corresponding to 500 eV. This is below the typical electron energy measured in the majority of sources. The most

68

P. Louarn

frequent regime is rather δ > i which is consistent with the observation of a dominant X mode inside the source. The examination of the dispersion curves suggests that the internally unstable waves are not easily connected to the external X mode, when the connections to the external Z and O modes are systematically possible. The direct internal/external X connection would imply: δ > o /2. For an external plasma frequency of ∼30 kHz, this corresponds to δ = 1.12 × 10−2 or, to energies of ∼5.5 keV. In most cases, the energy measured by Viking is smaller, with δ < o /2. Let us also note that the maximum instability takes place at Re(∆ω) = 0 thus always below the external X mode cut-oﬀ. One concludes that the most unstable internal waves are rarely directly connected to the external X mode. This leads to a few questions: • If the most unstable waves are directly connected to the external Z and O modes only, why is the level of Z and O modes as low as it is observed? • Can the electromagnetic energy created at Re(∆ω) = 0 be transmitted to the external X mode, and by what mechanism? • How important is the condition δ > o /2 in the generation of AKR? These questions are addressed by analyzing the dynamic spectra measured by Viking during the crossing of AKR sources. 3.3.3 Observations of Finite Geometry Eﬀects Selection of Source Crossings The dynamic spectra corresponding to four source crossings are presented in Fig. 3.8. They occurred during the orbits 165, 176, 237 and 1260. The electron gyrofrequency is indicated by a white line. Two frequency ranges have been selected for each orbit. The 10–60 kHz band contains the “auroral hiss”. As mentioned before, these whistler waves have a frequency cut-oﬀ at the local plasma frequency providing the plasma density. The radiation in the upper band is the AKR (typically 200–400 kHz). During small periods of time (a few 10 s at most, in between the vertical bars), the AKR intensiﬁes, and its low-frequency cut-oﬀ shifts down to fc . These are the AKR sources. They always correspond to a decrease of the power of the hiss, its cut-oﬀ being less apparent and often shifted below the band of analysis (10 kHz). This directly illustrates that the sources are plasma cavities. For each source, the conditions of mode connection can be studied at the two interfaces corresponding to the crossings of the source frontiers. The ratio fp /fc at each interface, just outside the sources, is deduced from the position of the hiss cut-oﬀ. The lower value of this ratio is observed for orbit 1260 (0.06) and the higher for orbit 176 (0.208). This corresponds to frequency-gaps between fc and the cut-oﬀ of the external X mode (fxo ) that vary from 1.2 kHz to more than 7 kHz. This is also the gap between the domain of wave ampliﬁcation and the frequency above which the connection with the external X mode

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

69

Fig. 3.8. Four examples of source crossings. The diﬀerent components of the AKR are indicated. The black line is the gyro frequency

becomes possible. The density inside the source is not precisely known. Since the hiss cut-oﬀ is below 10 kHz, it is expected to be smaller than ∼1 cm−3 . The duration of the crossing gives an upper limit for the source width. The thinnest sources would be 20–40 km wide, the widest 150–200 km. Note that oblique crossings being possible, these numbers must thus be considered as upper limits. Mode Identiﬁcations, Internal/External Connections Two components in the AKR can be distinguished in Fig. 3.8. The most intense has a bowl shaped cut-oﬀ at a few frequency channels above fc (see

70

P. Louarn

Fig. 3.9. Detailed views of the sources. The modulation due to the spin is wellapparent and is used in the mode identiﬁcation

orbits 165 and 176). The analysis of the polarization shows that it is the X mode component. Between this component and fc , a fainter O mode radiation is often detected. It can be relatively powerful as for orbit 176. In Fig. 3.9, detailed views of the same source crossings are presented. For each orbit, three panels are shown: (1) a dynamic spectra near fc , (2) a compressed “hiss” dynamic spectra and (3) the angle between the antenna and the geomagnetic ﬁeld. This angle will be used for the study of the polarization. The temporal resolution is 2.4 s and the spectral one is 2 kHz (4 kHz above

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

71

256 kHz). The source frontiers are indicated by tick marks. We also indicate the gyrofrequency and the X mode cut-oﬀ. Between fc and the X cut-oﬀ in the external plasma (fx0 ) the level of the AKR is more than 20 dB greater inside the sources than outside. This feature of the dynamic spectra shows that the electromagnetic energy is conﬁned inside the sources, at least close to fc . This is clearly apparent for obits 165 and 176. This diﬀerence between the internal and the external wave level decreases as the frequency increases. At frequencies a few percent above fc , the energy propagates outward more easily or even freely. Outside the sources, the most intense external waves are then observed above fxo . The modes of propagation can be determined from the spin modulation. Inside the sources and whatever the frequency, the maxima of the AKR Eﬁeld are observed for a perpendicular orientation of the antenna which is consistent with the extraordinary mode polarization. Outside the sources and for f > fxo , the maxima of the E-ﬁeld are also observed when the antenna is perpendicular (see the example shown in the panels of orbit 165). The internal and the external ﬁelds have thus a similar polarization which explains the easy energy escape in this frequency domain. Still outside the sources but for fc < f < fxo , the maxima are now observed for a parallel orientation of the antenna. This is the expected polarization of O mode waves. In this frequency domain, the internal/external polarization then diﬀers which could explain the attenuation observed at the crossing of the source frontier. It is interesting to note that the relative importance of the X and O components varies with the ratio fc /fp at the interfaces. For high values of this ratio, the O component is relatively intense. The O mode level is even larger than the X one for the densest case (left part of orbit 176). For low values of this ratio, the O component is comparatively less powerful and can even be undetectable (as in the case of orbit 1260). The relative proportion of O mode thus increases for sources embedded in dense plasma. The angle of propagation of the diﬀerent AKR components can also be estimated from the dynamic spectra. Given the geometry of the Viking orbit, radiation with a largely opened bowl-shaped dynamic spectra propagates at larger angles with respect to B0 than radiation with close bowl shape. One sees that the O mode essentially propagates perpendicular to the geomagnetic ﬁeld when the X mode propagates upward at oblique angles. This angle of propagation has been evaluated at the interfaces of the diﬀerent sources. It clearly decreases, i.e. the propagation is more parallel, when the ratio fp /fc increases. For large fp /fc (case of the orbit 176), very oblique angles of propagation (20◦ ) are observed [see Louarn and Le Qu´eau, 35]. Structure of the Electromagnetic Field and Radiating Diagram The ratio E /E⊥ (ratio between the electric ﬁeld measured when the antenna is respectively parallel and perpendicular to B0 ) can be used to follow the wave refraction in the sources. This ratio increases as the frequency increases. On

72

P. Louarn

average for the sources presented here, one gets E /E⊥ ∼0.1 for (f − fc )/fc = 0.01 and E /E⊥ ∼ 0.8 for (f −fc )/fc = 0.1. This increase can be related to the refraction of the waves inside the source. Since the waves are generated very close to fc , the analysis of waves with increasing frequencies is equivalent to the analysis of waves generated farther below. It is then possible to describe the evolution of the wave vector due to the upward propagation from the evolution of the spin modulation with the frequency. This nevertheless requires some theoretical considerations. The three components of the wave electric ﬁeld are related by the following relations: 1 − χ − N2 Ey χ N N⊥ (1 − χ − N 2 ) Ey , Ez = i 2) χ(1 − N⊥

Ex = −i

(3.2) (3.3)

where z is the direction of B0 , and the wave vector is in the x, z-plane. It 1 can be shown that E /E⊥ [Ez /(Ex2 + E − y 2 ) 2 ] depends more strongly on the variations on N than of the frequency. The evolution of this ratio provides thus actually an indication of the wave refraction in the source. As discussed in Louarn and Le Qu´eau [35], the increase of the ratio E /E⊥ from 0.1 to 0.8 would indicate that the wave vector rotates from a nearly perpendicular direction to an oblique one (∼45–60◦ ), over an altitude range of 100–200 km. The measurements of E /E⊥ oﬀer another indication. In the case of an isotropic radiation ( Ex ∼ Ey ), it can be veriﬁed that the theoretical E /E⊥ value hardly exceeds 0.4 even for very oblique propagation. This is signiﬁcantly smaller than the observed values. This can be solved by noting that the polarization can be very elliptic in the x-y plane. For example, for N ∼ 0.1 and ky ∼ 0, one gets Ex /Ey ∼ 0.05. If the radiating diagram is anisotropic in the x-y plane, the measurements of E may vary signiﬁcantly with the orientation of the antenna in the x-y plane. Much larger E /E⊥ could be measured by antenna almost perpendicular to the dominant x/y polarization or, equivalently, if the plane of the antenna is close to the main direction of the radiating diagram (see Fig. 3.10). Given the geometry of the orbit and the cartwheel spin mode of Viking, this means that the radiation coming from the sources studied here mainly propagates in the meridian plane (north-south direction). It could appear surprising that the four orbits examined here present this particular orientation. The bias certainly comes from the choice of long-duration source crossings. This indeed leads to the selection of tangential-to-the-source crossings, thus presenting a similar meridian orientation. This cannot be considered as a generality. Despite this particularity of these source crossings, the study shows that the waves are rather emitted tangentially to the sources which suggests that they optimize their ampliﬁcation path in the sources.

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

73

Fig. 3.10. Radiating diagram of AKR sources. The waves are preferentially emitted tangentially to the source

Summary The main conclusions of this observational study are the following: • The energy is systematically produced on the X mode. It is conﬁned inside the source from fc (the frequency of wave ampliﬁcation) to fx0 (the frequency of possible connection with the external X mode). In this frequency range, the connection is possible with the external O and Z modes. The eﬃciency of transmission to the O mode is small (attenuation by 25 dB). Nevertheless, it increases as fp /fc increases in the external plasma. Sources embedded in a particularly dense plasma can thus even emit a dominant O mode. The Z mode transmission coeﬃcient can be considered as null. • The energy generated in the source propagates upward before escaping. This corresponds to a progressive refraction of the waves that is actually measured. As discussed later, the range of upward propagation is linked to the frequency gap that separates fc and fxo . • The angle of propagation in the external plasma varies from 80◦ –70◦ to less than 20◦ . This angle decreases as the density gap at the source frontiers increases. A quite parallel propagation is observed for large fp /fc . • The radiation diagram of a source is not isotropic: the preferred direction of emission is the tangential-to-the-source direction. As shown by these observations, the ﬁnite geometry of the sources deeply inﬂuences the properties of the radiations. We will examine below how these eﬀects are explained by the theory of the cyclotron maser in ﬁnite geometry.

74

P. Louarn

3.4 The Cyclotron Maser in Finite Geometry 3.4.1 Mathematical Formulation A priori, the classical dielectric tensor – the one obtained in homogeneous plasma – is not adapted to the present analysis. Given the geometry chosen here, a Fourier transform is indeed not possible in the x direction and general solutions of the form H(x) · exp[i(ky y + k z − ωt)] must be considered. The corresponding general current perturbation is: J(x, ky , k , ω) = dx σ(x, x ky , k , ω)E(x , ky , k , ω) (3.4) The x convolution is an important mathematical complication. Nevertheless, as discussed now, the possibility of neglecting Larmor radius eﬀects greatly simpliﬁes the problem. The conductivity tensor indeed becomes purely local: σ(x, x , ky , k , ω) = σ(ky , k , ω)δ(x − x ) and the convolution is reduced to a simple multiplication. The conductivity tensor can also be directly computed starting from the Vlasov equation ∂f ∂f ∂f +v· − e (E + v × B) · =0 ∂t ∂x ∂p

(3.5)

where p is the relativistic momentum. Using cylindrical coordinates, p = (p cos θ, p sin θ, p ), the zero and ﬁrst order linear equations are: ∂f ∂f ∂f0 v· − e (v × B0 ) · =0 ∂t ∂x ∂p ∂ ∂ −iω + ik v + vx − vy ∂x ∂y ωc ∂ ∂f0 + δf = e [δE + v × δB] · Γ ∂θ ∂p

(3.6)

(3.7)

The zero order equation reads Γ v⊥ ∂f0 ∂f0 + =0. cos θ ωc ∂x ∂θ

(3.8)

If the Larmor radius (ρ) is neglected, one notes that the idealized ring-like distribution (see Sect. 3.2.1) is a solution of (3.8). Concerning the ﬁrst order (3.7), supposing that the transverse scale of the perturbation (L = 1/k for a wave) is large compared to ρ, one can neglect the terms vx ∂/∂x and vy ∂/∂y. They are indeed of the order ωc ρ/L and thus very small compared to ω and ωc . The ﬁrst-order equation then reduces to ωc ∂ ∂f0 . (3.9) −iω + ik v + δf = e [δE + v × δB] · Γ ∂θ ∂p

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

75

This is simply the classical equation obtained in an homogeneous plasma supposing k⊥ = 0. The conductivity tensor can be deduced from (3.9). Equivalently, one can make the assumption k⊥ = 0 in the general expression of this tensor that writes [see, e.g., Bekeﬁ, 5, p. 229] ωp2 σ = i2π0 ω

∞

∞ p⊥ dp⊥

0

with:

Mm

−∞

∞ dp Mm Γ m=−∞ ω − k v − mωc /Γ

(3.10)

2 mJ 2 mJm Jm mJm m v U −iv U W v ⊥ ⊥ ⊥ m b 2 b mJm Jm 2 v = iv⊥ U U (J ) −iv W J J ⊥ ⊥ m m m m b 2 mJm 2 −iv v U U J J v W J m m m m 2

where U Wm

∂f ∂f ∂f = me Γ ω + k p⊥ − p ∂p⊥ ∂p ∂p⊥

∂f ∂f ∂f mωc = me Γ ω − − p p⊥ ∂p Γ p⊥ ∂p ∂p⊥

where the Jm are the Bessel functions of order m, with argument b = Γ k⊥ v⊥ /ωc , and me is the rest mass of the electrons. For the problem under consideration here, i.e. the generation of supra-luminous mode in a moderately energetic plasma, the argument b is small. Retaining only the leading term in the development of the Bessel functions, one gets: ∞ ∞ ωp2 ωv ∂f v⊥ dv⊥ dv Mz σ = −i2π0 ω ω − k v ∂v −∞ 0

∂f v⊥ ∂f + + k v⊥ − v ω 4 ∂v⊥ ∂v M− M+ + × ω − k v − ωΓc ω − k v + ωΓc where

0 0 0 Mz = 0 0 0 001

1 −i 0 M− = i 1 0 0 0 0

1 i M+ = −i 1 0 0

(3.11)

0 0 0

Assuming that the distribution function is an idealized ring-like function: f (v , v⊥ ) = (2πv0 )−1 δ(v )δ(v⊥ − v0 ), this becomes:

76

P. Louarn

0 1 − χ1 iχ2 0 −iχ2 1 − χ1 = 0 0 1 − χz with

(3.12)

δ(N2 − 1) 1 1 + ± , 1 + ωc /ω 1 − (1 − δ)ωc /ω [1 − (1 − δ)ωc /ω]

χ1,2

ωp2 =± 2 2ω

χz

= −(ωp /ω)2

With the additional assumption that the generation takes place close to the gyrofrequency or ∆ω = (ω − ωc )/ωc 1, one obtains the simpliﬁed dielectric tensor that will be systematically used here: 0 1 − χi,o iχi,o 0 −iχi,o 1 − χi,o i,o = (3.13) 0 0 1 − χzi,o where the abbreviations are 2

χi =

i ∆ω + δN , 2 (∆ω + δ)2

χo =

o 1 , 2 ∆ω

χzo,i = o,i

and the subscripts i, o refer to the internal or the external plasma. Let us now take into account the ﬁnite geometry. In each of the three homogeneous regions that constitute the slab geometry, the Maxwell equations read: ω2 ·E =0 c2 ∇ × (∇ × H) + iω0 ∇ × ( · E) = 0

∇ × (∇ × E) −

(3.14)

It can be shown that the components of the electromagnetic ﬁeld can be deduced from the parallel components only (E and H , with H = µ0 B): ∂E ∂E i Ex = − + iχ k (N2 − 1 + χ) ∆ ∂x ∂y ∂H ∂H + (N2 − 1 + χ) +ωµ0 −iχ (3.15) ∂x ∂y ∂E ∂E i + (N2 − 1 + χ) k −iχ Ey = − ∆ ∂x ∂y ∂H ∂H + iχ −ωµ0 (N2 − 1 + χ) (3.16) ∂x ∂y and Hx = −

1 ωµ0

k Ey + i

∂E ∂y

,

Here ∆ = (ω 2 /c2 )[(1 − χ − N2 )2 − χ2 ].

Hy =

1 ωµ0

k Ex + i

∂E ∂x

(3.17)

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

77

The parallel components of the electromagnetic ﬁeld are related by the equations χ ω 2 1 − χz µ0 2 2 (1 − χ + N ) E = i ω 2 N H ∇⊥ + 2 (3.18) c 1 − chi c 1−χ

χ(1 − χ2 ω 2 1 − 2χ 0 − N2 H E (3.19) ∇2⊥ + 2 = −i ω 2 N c 1−χ c 1−χ where ∇2⊥ is the “perpendicular” Laplacean. For perpendicular propagation (N = 0), these equations respectively correspond to the dispersion of the O and the extraordinary (X and Z) mode. They can be combined to obtain a fourth-order equation for either the parallel electric or magnetic components. This equation must be solved in the diﬀerent plasma regions. Using the continuity conditions for Ey , Hy , E and H , one gets the compatibility condition that determines the set of discrete acceptable solutions. At the leading order in ∆ω, one ﬁnds: N2 ∂ χ ∂ + + 1− H = icµ0 N (3.20) E ∂x ˜ ∂ y˜ 1−χ 1−χ

∂ ω 2 1 − 2χ ∂ + + 2 − N2 H = −ic0 N χ1 − χE (3.21) ∂x ˜ ∂ y˜ c 1−χ where (˜ x, y˜) = (ωc /c)(x, y). We assume that χz = 0 which is justiﬁed by the low value of (ωp /ωc )2 inside and near the sources of AKR. With this simpliﬁcation, the fourth-order diﬀerential operator can be combined into two second-order wave operators corresponding respectively to the extraordinary and O modes: 1 − 2χ − N2 ∂ ∂ ∂ ∂ H + + + + 1 − N2 = 0 . (3.22) E ∂x ˜ ∂ y˜ 1−χ ∂x ˜ ∂ y˜ At this order of simpliﬁcation, the “O mode” operator simply becomes the “vacuum” operator. The O mode then “sees” no diﬀerence between the source and the external plasma. In this limiting case, it is stable, it freely propagates, and it is not coupled with the X mode. Let us consider the extraordinary mode: 1 − 2χ − N2 ∂ ∂ + + H = 0 . (3.23) ∂x ˜ ∂ y˜ 1−χ Let us choose symmetric solutions in each portion of the slab geometry. Inside the source ! ˜ exp i(Ny y˜ + N z˜ − ωt) . (3.24) Hi (x, y, z, t) = cos (si x Outside the source (+ on the right side, – on the left side)

78

P. Louarn

! Ho (x, y, z, t) = A exp(±iso x ˜) exp i(Ny y˜ + N z˜ − ωt) .

(3.25)

A is a normalization factor. si,o are the transverse wave numbers outside and inside the source, respectively. They satisfy the equation 1 − 2χi,o − N2

s2i,o + Ny2 =

1 − χi,o

.

(3.26)

Now, using the condition of continuity of Ey , Hy , E and H , one gets a relation between the amplitude inside and outside the sources and a compatibility condition: A (N2

= cos(si L) exp[−isi,o L] si so − 1 − χi ) sin(si L) + i(N2 − 1 + χo ) cos(si L) ∆i ∆

o χo χi = Ny − cos(si L) . ∆o ∆i

(3.27)

(3.28)

Equation (3.28) can be considered as the dispersion equation that contains the physical eﬀects due to the ﬁnite geometry of the source region. 3.4.2 Solutions of the Dispersion Relation The main results concerning the solutions of (3.28) are summed up below: • The solutions are discrete. They are localized both in the complex frequency and Re(ω)/Re(k) dispersive plane along curves presenting strong analogies with those obtained in the homogeneous case. In particular, the frequency of the X mode cut-oﬀ and of the Z mode resonance are not modiﬁed. This is illustrated in Fig. 3.11, where the solutions (for N = 0 and Ny = 0) are plotted. Two sets of discrete solutions are obtained (i) supraluminous solutions (s < 1) with the same cut-oﬀ as the “homogeneous” X Re ∆ω

Re ∆ω

Instability

0 0.5

- 0.5

Im ∆ω

X-mode

2 N

- 0.5 X-mode

Z-mode - 1.0

1

0

Z-mode

- 1.0

Fig. 3.11. Repartition of the discrete solutions in the complex frequency and dispersion plane. Normalized frequencies are used (see text)

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

•

• •

•

79

mode and (ii) infra-luminous solutions (s > 1) that accumulate near the Z mode resonance. The cut-oﬀ and the resonance do not depend on the width of the source. The wave vectors of two successive solutions diﬀer by 2π/L, the quantization rule being si (n) ∼ 2nπ/L. For s ∼ 1, the frequency gap between two successive solutions is thus of the order of 2cπ/L which corresponds to 200 Hz for a source of 30 km width. The cyclotron maser instability presents the same regimes as in the homogeneous case. They correspond to parameter ranges independent on the source width: for δ > i /2, the most unstable solutions are on the X mode near fc and N ∼ 1. For 3i /8 < δ < i /2, the most unstable solutions are on the X mode near the cut-oﬀ and N ∼ 0, for 3i /8 > δ, the most unstable solutions are on the Z mode near the resonance and N 1. Nevertheless, compared to the homogeneous case, both the maximum growth rate and the domain of instability are slightly reduced. Whatever the value of the electron energy, there is no possibility of a direct connection between the internal unstable waves and the external X mode waves. From the simple approach (see Sect. 3.3.2), one could conclude that for very energetic sources, such that δ > o /2, some internal unstable waves might directly escape the source on the X mode. The complete analysis shows that this is not the case. When δ is increased, the instability domain is indeed shifted towards lower frequencies which further increases the frequency gap with the external X mode cut-oﬀ. In fact, the direct connection is only possible for overdense source regions (i > o ). Polarization of the discrete mode. Inside the source, for s ∼ 0, the polarization is circular (|Ex | ∼ |Ey |). For s ∼ 1 it is elliptic (|Ex | < |Ey |) and, for s 1 it becomes quasi-longitudinal (|Ex | > |Ey |). The external part of the solutions corresponds to Z mode wave. The amplitudes of the internal and of the external parts of the solutions are of the same order, the level of the external Z mode is similar to that of the internal X mode. This result was also obtained in the simulation studies of Pritchett [50, 52]. It is in contradiction with the observations since no Z mode escapes from the sources.

3.4.3 The Problem of Energy Escape The problem of a too large theoretical production of the Z mode can be solved by considering an oblique propagation (Ny = 0). The component of the Poynting ﬂux that corresponds to the energy escape is proportional to Sx = Ey H . As already mentioned, the polarization is strongly elliptic for N ∼ 1, and Ey can be decreased by rotating the wave vector towards y direction. The X/Z transmission coeﬃcient can then be reduced by a factor larger than 20, from Sx ∼ 0.9 if Ny = 0 to Sx < 0.05 for Ny ∼ 0.95. Wave generation at Ny ∼ 1 would thus explain the very low level of Z mode production. This implies that the radiation would be preferentially generated tangentially

80

P. Louarn

rather than perpendicularly to the source. This is precisely what was deduced from the analysis of the polarization. As discussed in Louarn and Le Qu´eau [36], the fact that waves escape from the sources on the X mode can be explained by considering the upward propagation inside the source. When it propagates upward, the electromagnetic energy generated at a given frequency f indeed “sees” a decreasing fc and thus fxo . At some altitude above the region of generation, f becomes larger than fxo and the connection with the external X mode is possible. For this mechanism to be eﬃcient, the connections between the internal X mode and the external O and Z modes must be as ineﬃcient as possible. Otherwise, the energy would be converted into O and Z modes prior to the possible escape on the X mode. From both the observations and theoretical considerations, we thus get a new scenario of generation of the diﬀerent components of the AKR. During the upward propagation inside the source, the internal X mode waves are connected to the external O and Z modes and a part of the initial energy is converted into Z and O modes. Nevertheless, the transmission coeﬃcients (internal X mode → external O or Z modes) being small, the electromagnetic energy remains conﬁned inside the source until the connection with the external X mode becomes possible (see Fig. 3.12). This indirect production of the observable radio emission by a mode conversion at the source frontiers is one of the speciﬁc properties of the laminar source model. Let us quantify this eﬀect. If W is the density of electromagnetic energy inside the source, L the width of the source, T the coeﬃcient of energy transmission across the interfaces, vg⊥ and vg the perpendicular and parallel group velocities, one

X-mode

O-mode

upward propagation

Source

X-mode Plasma density Production of O-mode

Plasma cavity Fig. 3.12. Schematic of wave refraction and propagation inside and outside the sources. Sources corresponding to deep plasma cavities produce more O mode

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

has: L

vg⊥ ∂W = −2T W ∂h vg

81

(3.29)

where h is the altitude. The energy radiated from the source, assuming an upward propagation from ho to h is: h dz v g⊥ (3.30) W (h) = W0 exp −2 T vg L ho

T can be split into three terms: Txx the X/X transmission, Txo the X/O and, Txz , the X/Z ones. No Z mode escapes from the source which, as already presented, could be explained if the radiation is produced tangentially to the source. Thus, with this polarization, one has Txz ∼ 0. Txx is zero from ho to the altitude of possible connection with the X mode (hx ). This altitude is such that: (3.31) fxo (hx ) = fc (hx )(1 + o /2) ≤ fc (ho ) , where fxo (h) is the cut-oﬀ of the external X mode at altitude h. Assuming a linear variation of the magnetic ﬁeld, h = hx − ho is of the order of Ro /6. This altitude range is ∼100 km if R = 15000 km and o = 4 × 10−2 . Txo can be calculated from the continuity of the electromagnetic ﬁeld components. Using the dispersion relation in Sect. 4.1, it is possible to perform a parametric study of Txo . One approximately gets: Txo = 10−5 o δ 3

(3.32)

Using the equation of transfer, one obtains the relative proportion of X and O mode produced by the source, assuming that the energy not converted into O mode escapes as X mode whenever this becomes possible: O/X = (1 − A)/A

with

A ∼ exp[−10−4 o δ 3 (o + δ/2)R/L]

(3.33)

This expression shows that the proportion of O mode increases if: • • • •

the energy of the emitting particles increases (role of δ), the density of the external plasma increases (role of o ), the distance to the planet increases (R), or the width of the source decreases (L).

In conclusion, narrow energetic sources, embedded in a dense plasma, produce a more powerful O mode radiation. Again, as explained in Sect. 3.3, this is an observational fact.

82

P. Louarn

3.5 Discussion and Conclusion 3.5.1 O and X Mode Production in Narrow Sources An important aspect of the laminar source model is to separate the initial generation of the electromagnetic energy (the maser inside the source) and the production of the observable radio emission (by conversion across the source frontiers). One may consider that the maser operates inside the sources with maximum eﬃciency: this produces X mode waves at the gyrofrequency in a fully relativistic plasma. The produced electromagnetic energy is conﬁned in the source, and mode conversion must be taken into account for explaining the energy escape. During this stage, radiation in the O mode is produced. To illustrate this scenario, we have performed a parametric study corresponding to terrestrial kilometric and Jovian decametric radiation respectively [Louarn, 37]. These two cases diﬀer by the distances from the center of the planet (R = 15000 km for the Earth and R = 70000 km for Jupiter) and the value of the gyrofrequency, a factor of 100 larger at Jupiter than at Earth. Since fc is used for normalization, one obtains normalized parameters (i,o , L . . .) much smaller in the Jovian than the terrestrial cases. For the terrestrial case, one shows that only relatively extreme conditions (width smaller than 10 km, strong density: o ∼ 0.3 and high energy: 10 keV) lead to the generation of a large fraction of O mode, as is actually observed with Viking. For the Jovian case, the expected very low values of o due to the strong magnetization of the plasma make O mode production even more diﬃcult. Even for very narrow sources and energetic particles, the X mode is thus expected to dominate. It is nevertheless possible that energetic very low altitude sources (so that o could be larger) emit in the O mode. This discussion could be extended to other types of radio emission. For example, in the solar corona, the laminar source model is consistent with the idea of fragmented energy release. During a ﬂare, the global energy release could well take place over small scale regions that could also be the sources of non-thermal emission. The polarization of the radiation coming from such a strongly structured ﬁbrous-like corona would simply result from the conversion processes at the frontiers of the sources, not from speciﬁc generation processes [see Sharma et al., 54]. 3.5.2 Fine Structure of Radiation What could be the radiating diagram of a laminar source? Could it explain some of the ﬁne structures observed in the radio spectra by Baumback and Calvert [4], Ergun et al. [17], and Pottelette et al. [45]? As already discussed, the ﬁlamentary geometry creates an asymmetry around the magnetic ﬁeld. The x and y directions are indeed not equivalent: ky is a free parameter when kx is quantized (kx = 2πn/L). For a given ky , the unstable waves are thus emitted at discrete frequencies, along preferential directions of propagation, the radiation diagram being a succession of

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

83

narrow beams in the plane perpendicular to B0 . At a diﬀerent ky , a slightly modiﬁed diagram will be obtained with diﬀerent quantized frequencies. Due to the possibility of continuous variation of ky , the narrow beams likely overlap and, at some distance from the source, a continuous dynamic spectrum would be observed. This point is discussed by Pritchett et al. [52]. Without other ingredients as, for example, a mechanism that would explain wave generation at preferred altitudes or an interaction with non-linear structures [see Pottelette and Treumann, 46, this issue], it seems unlikely that the laminar source model alone may explain the ﬁne structures. Nevertheless, the possibility of the formation of conics has still not been analyzed. Due to the quantization of kx , it is indeed not excluded that the energy radiated by the source concentrates along deﬁned ray-paths. A spacecraft crossing these ray-paths would see a more powerful emission at deﬁned frequencies. Another possibility would be that both ky and kx are quantized as it would be the case for sources limited in both the x and the y directions. This 3-D model also remains to be studied. 3.5.3 Conclusion and Pending Questions To conclude, the laminar source model has been precisely tested, both observationally and theoretically, and it is certainly not exaggerated to aﬃrm that the AKR generation is today one of the best-known magnetospheric phenomena. Given the similarity of the terrestrial auroral phenomena with those at Jupiter and Saturn, the laminar source model can also certainly be generalized. One fundamental point is still lacking: the prediction of the power of the emission, in relation with the energy and the number of the accelerated particles. An important diﬃculty here is the identiﬁcation of the process that saturates the maser. Is this just the convection of the energy out of the source? Is it some non-linear eﬀect, for example the relaxation of the free energy by non-linear wave/particle interactions? The full quantitative understanding of the coherent radio wave production remains an exciting goal for the future, with many potential applications regarding the interpretation of radio emission from distant astrophysical objects.

References [1] Aschwanden, M.J. and A. Benz: On the electron cyclotron instability: II. Pulsation in the quasi-stationary state, Astrophys. J. 332, 466, 1988. [2] Bahnsen, A., M. Jespersen, E. Ungstrup, and I.B. Iversen: Auroral hiss and kilometric radiation measured from the Viking satellite, Geophys. Res. Lett. 14, 471, 1987. [3] Bahnsen A., B.M. Pedersen, M. Jespersen, E. Ungstrup, L. Eliasson, J.S. Murphree, D. Elphinstone, L. Blomberg, G. Hohmgren, and L.J. Zanetti: Viking observations at the source of the AKR, J. Geophys. Res. 94, 6643, 1989.

84

P. Louarn

[4] Baumback, M.M. and W. Calvert: The minimum bandwidths of auroral kilometric radiation, Geophys. Res. Lett. 14, 119, 1987. [5] Bekeﬁ, G.: Emission processes in plasma, p. 229, Gordon & Breach, New York, 1963. [6] Benediktov, E.A., G.G. Getmantsev, Y.A. Sazonov, and A.F. Tarojov: Preliminary results of measurements of the intensity of distributed extraterrestrial radio frequency emission at 725 and 1525 kHz (in Russian), Kosm. Issled. 1, 614, 1965. [7] Benson, R.F. and W. Calvert: ISIS-1 observations at the source of auroral kilometric radiation, Geophys. Res. Lett. 6, 479, 1979. [8] Benson, R.F., W. Calvert, and D.M. Klumpar: Simultaneous wave and particle observations in the auroral kilometric radiation source region, Geophys. Res. Lett. 7, 959, 1980. [9] Benson, R.F. and S.I. Akasofu: Auroral kilometric radiation/aurora correlation, Radio Sci. 19, 527, 1984. [10] Calvert, W.: The auroral plasma cavity, Geophys. Res. Lett. 8, 919, 1981. [11] Calvert, W.: A feedback model for the source of auroral kilometric radiation, J. Geophys. Res. 87, 8199, 1982. [12] Carlson, C.W., R.F. Pfaﬀ, and J.G. Watzin: The fast auroral snapshot (FAST) mission, Geophys. Res. Lett. 25, 2017, 1998. [13] De Feraudy, H., B.M. Pedersen, A. Bahnsen, and M. Jespersen: Viking observations of AKR. from plasmasphere to night auroral oval source region, Geophys. Res. Lett. 14, 511, 1987. [14] Delory, G.T., R.E. Ergun, C.W. Carlson, L. Muschietti, C.C. Chaston, W. Peria, J.P. McFadden, and R.J. Strangeway: FAST observations of electron distributions within AKR source regions, Geophys. Res. Lett. 25, 2069, 1998. [15] Dulk, G.A.: Radio emission From the sun and the stars, Ann. Rev. Astron. Astrophys. 23, 169, 1985. [16] Eliasson, L., G.A. Holmgren, and K. R¨ onnmark: Pitch-angle and energy distributions of auroral electrons measured by the ESRO-4 satellite, Planet. Space Sci. 27. 87, 1979. [17] Ergun, R.E., et al.: FAST satellite wave observations in the AKR source region, Geophys. Res. Lett. 25, 2061, 1998. [18] Ergun, R.E., C.W. Carlson, J.P. McFadden, G.T. Delory, R.J. Strangeway, and P.L. Pritchett: Electron-cyclotron maser driven by charged particle acceleration from magnetic ﬁeld-aligned electric ﬁelds, Astrophys. J. 538, 456, 2000. [19] Ergun, R.E., et al.: The FAST satellite ﬁelds instrument, Space Sci. Rev. 98, 67, 2001. [20] Green, J.L., D.A. Gurnett, and R.A. Hoﬀman: Correlation between auroral kilometric radiation and inverted-V electron precipitations, J. Geophys. Res. 84, 5216, 1979. [21] Gurnett, D.A.: The Earth as a radio source: terrestrial kilometric radiation, J. Geophys. Res. 79, 4227, 1974. [22] Gurnett, D.A. and J.L. Green: On the polarization and origin of auroral kilometric radiation, J. Geophys. Res. 83, 689, 1978. [23] Gurnett, D.A. and R.R. Anderson: The kilometric radio emission spectrum; Relation to auroral acceleration processes, in Physics of Auroral Arc Formation, Geophys. Monogr. Ser., vol. 25, edited by S.-l. Akasofu, and J. R. Kan, p. 341, AGU, Washington, D. C., 1981.

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

85

[24] Hilgers, A., A. Roux, and R. Lundin: Characteristics of AKR sources: a statistical description, Geophys. Res. Lett. 18, 1493, 1991. [25] Hilgers, A., H. de Feraudy, and D. Le Qu´eau: Measurement of the direction of the auroral kilometric radiation electric ﬁeld inside the source with the Viking satellite, J. Geophvs. Res. 79, 8381, 1992. [26] Kaiser, M.L., J.K. Alexander, A.C. Riddle, J.B. Pierce, and J.W. Warwick: Direct measurements by Voyager 1 and 2 of the polarization of terrestrial kilometric radiation, Geophys. Res. Lett. 5., 857, 1978. [27] Lee, L.C. and C.S. Wu: Ampliﬁcation of radiation near cyclotron frequency due to electron population inversion, Phys. Fluids 22, 1348, 1980. [28] Le Qu´eau, D., R. Pellat, and A. Roux: Direct generation of the auroral kilometric radiation by the maser synchrotron instability: An analytical approach, Phys. Fluids 27, 247, 1984a. [29] Le Qu´eau, D., R. Pellat, and A. Roux: Direct generation of the auroral kilometric radiation by the maser synchrotron instability: Physical discussion of the mechanism and parametric study, J. Geophys. Res. 89, 2841, 1984b. [30] Le Qu´eau, D., R. Pellat, and A. Roux: The maser synchrotron instability in an inhomogeneous medium: Application to the generation of the auroral kilometric radiation, Ann. Geophys. 3, 273, 1985. [31] Le Qu´eau, D. and P. Louarn: Analytical study of the relativistic dispersion: application to the generation of the AKR, J. Geophys. Res. 94, 2605, 1989. [32] Louarn, P., D. Le Qu´eau, and A. Roux: A new mechanism for stellar radio bursts: the fully relativistic electron maser, Astron. Astrophys. 165, 211, 1986. [33] Louarn, P., A. Roux, H. de Feraudy, D. Le Qu´eau, M. Andr´e, and L. Matson: Trapped electrons as a free energy source for the auroral kilometric radiation, J. Geophys. Res. 95, 5983, 1990. [34] Louarn, P., D. Le Qu´eau, and A. Roux: Formation of electron trapped population and conics inside and near auroral acceleration region, Ann. Geophys. 9, 553, 1991. [35] Louarn, P. and D. Le Qu´eau: Generation of the auroral kilometric radiation in plasma cavities, I, Experimental study, Planet. Space Sci. 44, 199, 1996a. [36] Louarn, P. and D. Le Qu´eau: Generation of the auroral kilometric radiation in plasma cavities, II, The cyclotron maser instability in small size sources, Planet. Space Sci. 44, 211, 1996b. [37] Louarn, P.: Radio emissions from ﬁlamentary sources: a simple approach, Planetary radio emissions IV, Ed. Rucker H.O., Bauer S.J. and Lecacheux A., Verlag ¨ der Osterreichischen Akademie der Wissenschaften, 1997. [38] Mellott, M.M., W. Calvert, R.L. Huﬀ, D.A. Gurnett, and S.D. Shawhan: DE 1 observation of ordinary mode and extraordinary mode auroral kilometric radiation, Geophys. Res. Lett. 11, 1188, 1984. [39] Melrose, D.B. and G.A. Dulk: Electron-cyclotron masers as the source of certain solar and stellar radio bursts, Astrophys. J., 259, 844, 1982. [40] Melrose, D.B., K.G. R¨ onnmark, and R.G. Hewitt: Terrestrial kilometric radiation: The cyclotron theory, J. Geophys. Res. 87, 5140, 1982. [41] Omidi, N. and D.A. Gurnett: Growth rate calculations of auroral kilometric radiation using the relativistic resonance condition, J. Geophys. Res. 87, 2377, 1982. [42] Omidi, N. and D.A. Gurnett: Path-integrated growth of auroral kilometric radiation, J. Geophys. Res. 89, 10,801, 1984.

86

P. Louarn

[43] Perraut, S., H. de Feraudy, A. Roux, P.M.E. D´ecr´eau, J. Paris, and L. Matson: Density measurements in key regions of the Earth’s magnetosphere: Cusp and auroral region, J. Geophys. Res. 95, 5997, 1990. [44] Pottelette, R., M. Malingre, A. Bahnsen, L. Eliasson, and K. Stasiewicz: Viking observations of bursts of intense broadband noise in the source region of auroral kilometric radiation, Ann. Geophys. 6, 573, 1988. [45] Pottelette, R., R.A. Treumann, and M. Berthomier: Auroral plasma turbulence and the cause of auroral kilometric radiation ﬁne structure, J. Geophys. Res. 106, 8465, 2001. [46] Pottelette, R. and R.A. Treumann: Auroral Acceleration and Radiation, in: part 1 of this book, 2005. [47] Pritchett, P.L.: Relativistic dispersion, the cyclotron maser instability, and auroral kilometric radiation, J. Geophys. Res. 89, 8957, 1984. [48] Pritchett, P.L. and R.J. Strangeway: A simulation study of kilometric radiation generation along an auroral ﬁeld line, J. Geophys. Res. 90, 9650, 1985. [49] Pritchett, P.L.: Electron-cyclotron maser instability in relativistic plasmas, Phys. Fluids 29, 2919, 1986a. [50] Pritchett, P.L.: Cyclotron maser radiation from a source structure localized perpen dicular to the ambient magnetic ﬁeld, J. Geophys. Res. 91, 13,569, 1986b. [51] Pritchett, P.L. and R.M. Winglee: Generation and propagation of kilometric radiation in the auroral plasma cavity, J. Geophys. Res. 94, 129, 1989. [52] Pritchett, P.L., R.J. Strangeway, R.E. Ergun, and C.W. Carlson: Generation and propagation of cyclotron maser emissions in the ﬁnite auroral kilometric ra-diation source cavity, J. Geophys. Res. 107, 12, doi:10.1029/2002JA009403, 2002. [53] Roux, A., A. Hilgers, H. de Feraudy, D. Le Qu´eau, P. Louarn, S. Perraut, A. Bahnsen, M. Jespersen, E. Ungstrup, and M. Andr´e: Auroral kilometric radiation sources: in situ and remote sensing observations from Viking, J. Geophys. Res. 98, 11657, 1993. [54] Sharma, R.R., L. Vlahos, and K. Papadopoulos: The importance of plasma eﬀects on electron-cyclotron maser emission from ﬂaring loops, Astron. Astrophys. 112, 377, 1982. [55] Shawhan, S.D. and D.A. Gurnett: Polarization measurements of auroral kilometric radiation by DE-1, Geophys. Res. Lett. 9, 913, 1982. [56] Strangeway, R.J.: Wave dispersion and ray propagation in a weakly relativ-istic electron plasma: Implications for the generation of auroral kilometric radiation, J. Geophys. Res. 90, 9675, 1985. [57] Trubnikov, B.A.: In Plasma physics and the problem of controlled thermonuclear reactions, ed. M.A. Leontovich, Pergamon Press Inc., New york, 1959. [58] Ungstrup, E., A. Bahnsen, H.K. Wong, M. Andr´e, and L. Matson: Energy source and generation mechanism for AKR, J. Geophys. Res. 95, 5973, 1990. [59] Wu, C.S. and L.C. Lee: A theory of the terrestrial kilometric radiation, Astrophys. J. 230, 621, 1979. [60] Zarka, P.: The auroral radio emissions from planetary magnetospheres: what do we know, what don’t we know, what do we learn from them? Adv. Space Res. 12, 99. 1992.

4 Generation of Emissions by Fast Particles in Stochastic Media G.D. Fleishman National Radio Astronomy Observatory, Charlottesville, VA 22903 [email protected]

Abstract. We demonstrate the potential importance of small-scale turbulence in the generation of radio emission from natural plasmas. This emission, being reliably detected and interpreted, probes small-scale turbulence in remote sources in the most direct way. The radiation emitted is called “diﬀusive synchrotron radiation” because it is related to shaking the electron distribution when passing through randomly distributed small-scale plasma inhomogeneities caused in turbulence. The emissivity is calculated and shown to be in the observable range. A further eﬀect of inhomogeneities is transition radiation arising from fast particles which interact with small-scale density inhomogeneities. This emission process generates continuum emission below synchrotron and is applicable to some solar radio bursts. It serves for probing number densities. The eﬀect of inhomogeneities on coherent emissions is either broadening or splitting of the spectral peaks generated by the electron cyclotron maser mechanism.

Key words: Solar radio bursts, random inhomogeneities, turbulent radiation, diﬀuse synchrotron radiation

4.1 Introduction Interaction of charged particles with each other and/or with external ﬁelds results in emission of electromagnetic radiation. This article considers the emission arising as fast (nonthermal) particles move through media with random inhomogeneities. The nature of these inhomogeneities might be rather arbitrary. One of the simplest examples of inhomogeneities is a distribution of the microscopic particles (atoms or molecules) in an amorphous substance, so the medium is inhomogeneous at microscopic scales (of the order of the mean distance between particles), perhaps remaining uniform (on average) at macroscopic scales. More frequently, however, real objects are inhomogeneous on macroscopic scales as well. The irregularities might be related to the interfaces between G.D. Fleishman: Generation of Emissions by Fast Particles in Stochastic Media, Lect. Notes Phys. 687, 87–104 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

88

G.D. Fleishman

inhomogeneities, variations of the elemental composition, temperature, density, electric, and magnetic ﬁeld. Random inhomogeneities of any of these parameters may strongly aﬀect in various ways the generation of electromagnetic emission. For example, the presence of the density inhomogeneities implies that the dielectric permeability tensor is a random function, as well as the refractive indices of the electromagnetic eigen-modes. As a result, the eigen-modes of the uniform medium are not the same as the eigen-modes of the real inhomogeneous medium. The irregularities of the electric and magnetic ﬁelds aﬀect primarily the motion of the fast particle (although the eﬀect of the ﬁeld ﬂuctuations on the dielectric tensor exists as well). Below we consider one of many emission processes appearing due to or affected by small-scale random inhomogeneities, namely, diﬀusive synchrotron radiation arising as fast particles are scattered by the small-scale random ﬁelds. This emission process is of exceptional importance since current models of many astrophysical objects (see, e.g., [6, 7] and references therein) imply generation of rather strong small-scale magnetic ﬁelds. The eﬀect of the inhomogeneities on other emission processes is discussed brieﬂy as well.

4.2 Statistical Methods in the Theory of Electromagnetic Emission The trajectories of charged particles and the ﬁelds created by them are random functions as the particles move through a random medium. Thus, the use of appropriate statistical methods is required to describe the particle motion and the related ﬁelds. 4.2.1 Spectral Treatment of Random Fields For a detailed theory of the random ﬁelds we refer to a monograph [Toptygin, 20] and mention here a few important points only. To be more speciﬁc, let us discuss some properties of the random magnetic ﬁelds. Assume that total magnetic ﬁeld is composed of regular and random components B(r, t) = B 0 (r, t) + B st (r, t), such as B 0 (r, t) = B(r, t) and

B st (r, t) = 0, where the brackets denote the statistical averaging. Note, that the method of averaging depends on the problem considered. The statistical properties of the random ﬁeld might be described with a (inﬁnite) sequence of the multi-point correlation functions, the most important of which is the (two-point) second-order correlation function (2)

Kαβ (R, T, r, τ ) = Bst,α (r 1 , t1 )Bst,β (r 2 , t2 ) , where R = (r 1 + r 2 )/2, r = r 2 − r 1 , T = (t1 + t2 )/2, and τ = t2 − t1 .

(4.1)

4 Generation of Emissions by Fast Particles in Stochastic Media

89

Since the regular and random ﬁelds are statistically independent ( B0 Bst = B0 Bst = 0), each of them satisﬁes the Maxwell equations separately. In particular ∇ · B st = 0, so only two of three vector components of the random ﬁeld are independent. For a statistically uniform random ﬁeld the Fourier transform of the cor(2) relator Kαβ (r, τ ) over spatial and temporal variables r and τ gives rise to the spectral treatment of the random ﬁeld drdτ i(ωτ −kr) (2) e Kαβ (r, τ ) . (4.2) Kαβ (k, ω) = (2π)4 In the case of isotropic turbulence we easily ﬁnd

1 kα kβ Kαβ (k, ω) = K(k)δ(ω − ω(k)) δαβ − 2 , 2 k

(4.3)

which, in particular, satisﬁes the the Maxwell equation ∇ · B st = 0, since the tensor structure of the correlator is orthogonal to the k vector: kα Kαβ = 0. Although the spectral shape of the correlators is not unique and may substantially vary depending on the situation, we will adopt for the purpose of the model quasi-power-law spectrum of the random measures: " 2# ν−1 Γ (ν/2 + 1)kmin A Aν , Aν = , (4.4) K(k) = 2 2 ν/2+1 3/2 (kmin + k ) 3π Γ (ν/2 − 1/2) where ν is the spectral index of the turbulence, and the spectrum K(k) is normalized to d3 k: kmax " # K(k)d3 k = A2 , for kmin kmax , ν > 1 , (4.5) 0

where A2 is the mean square of the corresponding measure of the random 2 2 , Est , ∆N 2 etc. ﬁeld, e.g., Bst 4.2.2 Emission from a Particle Moving along a Stochastic Trajectory The intensity of the emission of the eigen-mode σ σ En,ω = (2π)6

ω 2 nσ (ω) |(eσ · j ω,k )|2 c3

(4.6)

depends on the trajectory of the radiating charged particle since the Fourier transform j ω,k of the corresponding electric current has the form: ∞ dt v(t) exp(iωt − ikr(t)) , (4.7) j ω,k = Q (2π)4 −∞

90

G.D. Fleishman

where Q is the charge of the particle. For the stochastic motion of the particle, we have to substitute (4.7) into (4.6) and perform the averaging of the corresponding expression: σ En,ω =

Q2 ω 2 nσ (ω) 4π 2 c3 T ×Re dt −T

(4.8) ∞

dτ eiωτ e−ik[r(t+τ )−r(t)] (e∗σ · v(t + τ ))(eσ · v(t)) ,

0

where 2T is the total time at which the emission occurs, and eσ is the polarization vector of the eigen-mode σ. It is convenient to perform the averaging denoted by the brackets with the use of the distribution function of the particle(s) F (r, p, t) at the time t and the conditional probability W (r, p, t; r , p , τ ) for the particle to transit from the state (r, p) to the state (r , p ) during the time τ . For statistically uniform random ﬁeld we obtain: σ En,ω = ×Re

Q2 ω 2 nσ (ω) (4.9) 4π 2 c3 T ∞ dt dτ eiωτ drdpdp (e∗σ · v )(eσ · v)F (r, p, t)Wk (p, t; p , τ ) .

−T

0

Then, the integration over τ gives rise to the temporal Fourier transform of W , so the spectrum of emitted electromagnetic waves is expressed via the spatial and temporal Fourier transform of the distribution function of the particle in the presence of the random ﬁeld. 4.2.3 Kinetic Equation in the Presence of Random Fields The conditional probability W , which substitutes the particle trajectory in the presence of random ﬁelds, can be obtained from the kinetic Boltzmannequation: ∂f ∂f ∂f +v· + FL · =0, (4.10) ∂t ∂r ∂p where F L = QE + Qc (v × B) is the Lorentz force, while the electric (E) and magnetic (B) ﬁelds contain in the general case both regular and random components. Let us express the Lorentz force as a sum of these two components explicitly: (4.11) F L = F R + F st . Accordingly, we’ll seek a distribution function in the form of the sum of the averaged (W ) and ﬂuctuating (δW ) components: f (r, p, t) = W (r, p, t) + δW (r, p, t) .

(4.12)

The equation for the averaged component W can be derived from (4.10) applying the Green’s function method [Toptygin, 20]:

4 Generation of Emissions by Fast Particles in Stochastic Media

91

2 Qc ∂W ∂W → − +v· − (Ω · O )W = (4.13) ∂t ∂r E ∞ × dτ Oα Tαβ [∆r(τ ), τ ]Oβ W [r − ∆r(τ ), p − ∆p(τ ), t − τ ] , 0

where

QBc (4.14) E is a vector pointing in the direction of the magnetic ﬁeld and whose magnitude equals the rotation frequency of the charged particle with energy E, Ω=

2 $

Bst rα rβ % . ψ(r)δαβ + ψ1 (r) 2 3 r (4.15) To derive (4.13) we transformed the terms with the magnetic ﬁeld using the following property of the scalar triple product: Q ∂ QB ∂ ∂ QBc (v × B) · =− · v× · O, O= v× =− , c ∂p c ∂p E ∂v (4.16) where O is the operator of the angular variation of the velocity. The equation (4.13) is rather general and can be applied to the study of both emission by fast particles and particle propagation in the plasma [Toptygin, 20]. Further simpliﬁcations of equation (4.13) can be done by taking into account some speciﬁc properties of the problems considered. The theory of wave emission involves a fundamental measure called the coherence length (or the formation zone) that refers to that part of the particle path where the elementary radiation pattern is formed. The coherence length is much larger than the wavelength for the case of relativistic particles, e.g., the coherence length for synchrotron radiation in the presence of the uniform magnetic ﬁeld is ls = RL /γ = M c2 /(QB), where RL is the Larmor radius, γ = E/M c2 is the Lorentz-factor of the particle. Length ls is by the factor of γ 2 larger than the corresponding wave length. The eﬀect of magnetic ﬁeld inhomogeneity on the elementary radiation pattern is speciﬁed by the ratio of the spatial scale of the ﬁeld inhomogeneity and the coherence length. If the scale of inhomogeneity is much larger than the coherence length, the eﬀect of the inhomogeneity is small and can typically be discarded. However, if the magnetic ﬁeld changes noticeably at the coherence length, the inhomogeneity aﬀects the emission strongly, so the spectral and angular distributions of the intensity and polarization of the emission can be remarkably diﬀerent from the case of the uniform ﬁeld. This means, in particular, that in the presence of magnetic turbulence with a broad distribution over the spatial scales, the large-scale spatial irregularities should be considered like the regular ﬁeld, while the small-scale ﬂuctuations should be properly taken into account as the random ﬁeld. Since the variation

Tαβ (r, τ ) = Bst,α (r 1 , t1 )Bst,β (r 2 , t2 ) =

92

G.D. Fleishman

of the particle speed (momentum) over the correlation length of the smallscale random ﬁeld is small, then we can adopt ∆p(τ ) = 0, ∆r(τ ) = vτ in the right-hand-side of (4.13). Then, the kinetic equation takes the form ∂W ∂W

B 2 → − +v· − (Ω · O )W = st ∂t ∂r 3

Qc E

2 ∞ O2 dτ ψ(vτ )W (r − vτ, p, t − τ ) , 0

(4.17) 4.2.4 Solution of Kinetic Equation Let us outline the solution of the kinetic equation (4.17) for the averaged distribution function W . First of all, the stochastic ﬁeld has to be split into large& and small-scale (B st ) components. To see this, consider a purely scale (B) sinusoidal spatial wave of the magnetic ﬁeld with the strength B0 and the wavelength λ0 = 2π/k0 . If the wavelength λ0 is less than the coherence length ls0 calculated for the emission in a uniform ﬁeld B0 , ls0 ∼ M c2 /(QB0 ), this wave represents the small-scale ﬁeld, whose spatial inhomogeneity is highly > ls0 , it is a large-scale important for the emission; in the opposite case, λ0 ∼ ﬁeld. The splitting is less straightforward when the random ﬁeld is a superposition of the random waves with a quasi-continuous distribution over the spatial scales. Let us consider the eﬀect provided by a random magnetic ﬁeld corresponding to a small range ∆k in the spectrum (4.4) on the charged particle trajectory. The energy of this magnetic ﬁeld is δEst ∼ K(k)k 2 ∆k .

(4.18) √ The corresponding non-relativistic “gyrofrequency” is δωst ∼ Q δEst /(M c). For a truly random ﬁeld, when the harmonics with k and k+dk are essentially uncorrelated, we can arbitrarily select the value ∆k to be small enough to satisfy δωst kc for any k, so all the independent ﬁeld components represent the small-scale ﬁeld. However, in a more realistic case the Fourier components of the random ﬁeld with similar yet distinct k are typically correlated, so they disturb the particle motion coherently and ∆k in estimate (4.18) cannot be arbitrarily small any longer. Accordingly, all components of the random ﬁeld < δωst /c (where δωst is calculated for the smallest allowable ∆k) must with k ∼ be treated as a large-scale ﬁeld. & together with the regular ﬁeld B 0 speciﬁes the The large-scale ﬁeld B vector Ω in the left-hand-side of equation (4.17): Ω=

& Q(B 0 + B)c . E

(4.19)

For the analysis of the emission process (and, respectively, for the solution of the kinetic equation (4.17)), we treat the large-scale ﬁeld (which is the sum of the regular and large-scale stochastic ﬁelds) as uniform, (Ω = const);

4 Generation of Emissions by Fast Particles in Stochastic Media

93

the actual inhomogeneity might be taken into account by averaging the ﬁnal expressions of the emission if necessary. Equation (4.17) has been solved in the presence of a uniform magnetic ﬁeld and small-scale random magnetic ﬁelds [cf. 13]: 2 ωpe 1 ωv 1− τ w(θ 0 , θ, τ ) , (4.20) Wk = 2 δ(p − p0 ) exp −i p c 2ω 2 where

x exp − x(θ2 + θ02 ) coth zτ π sinh zτ 2 τ (θ − θ 0 ) · (n × Ω) Ω⊥ − + 2xθθ 0 sinh−1 zτ − , (4.21) 2q 4q

θ2 θ2 v0 v −n 1− 0 , θ0 = θ = −n 1− , (4.22) v 2 v 2

1 1 ω 2 , z = (1 − i)(ωq) 2 , (4.23) x = (1 − i) 16q 2 Qc (4.24) dk K(k )δ[ω − (k − k )v] . q(ω, θ) = π E

w(θ 0 , θ, τ ) =

If there is only a random ﬁeld and no regular ﬁeld, the function w reads x −1 2 2 exp −x(θ +θ0 ) coth zτ +2xθθ 0 sinh zτ , (4.25) w(θ 0 , θ, τ ) = π sinh zτ while in the opposite case, when there is no random ﬁeld, we have iωτ 2 Ω2 τ 2 w(θ 0 , θ, τ ) = δ(θ − θ 0 + [n × Ω]τ ) exp θ0 − θ 0 · [n × Ω]τ + ⊥ . 2 3 (4.26) Calculation of the emission with the use of this distribution function leads evidently to the standard expressions of synchrotron radiation in the uniform magnetic ﬁeld. Finally, the distribution function of the free particle (moving without any acceleration), which does not produce any emission in the vacuum or uniform plasma, is iωτ 2 θ . (4.27) w0 (θ 0 , θ, τ ) = δ(θ − θ 0 ) exp 2

94

G.D. Fleishman

4.3 Emission from Relativistic Particles in the Presence of Random Magnetic Fields 4.3.1 General Case Let us consider the energy emitted by a single particle (regardless of polarization) based on the general expression (4.9), i.e., we take the sum of (4.9) over the two orthogonal eigen-modes: ' Q2 ω 2 ε(ω) (4.28) En,ω = 2π 2 c3 T ∞ × Re dt dτ eiωτ dr dp dp [n × v ] · [n × v]F (r, p, t)Wk (p, t; p , τ ) , −T

0

where we neglected the diﬀerence between 'the two refractive indices in the magnetized plasma and adopted nσ (ω) ≈ ε(ω). In the presence of a statistically uniform and stationary magnetic ﬁeld the emitted energy (4.28) is proportional to the time (on average), although the intensity of emission at a given direction n depends on time since the angle between the instantaneous particle velocity v(t) and n changes with time as described by the dependence of the function F (r, p, t) on time t. This kind of the temporal dependence is not of particular interest, e.g., it represents periodic pulses provided by the rotation of the particle in the uniform magnetic ﬁeld, so it is more convenient to proceed with time-independent intensity of radiation emitted into the full solid angle ∞ 2 2 γ ωpe Q2 ω 2 ' iωτ ε(ω) Re dτ exp 1+ (4.29) Iω = 2π 2 c 2γ 2 ω2 0 × d2 θ d2 θ (θθ )(w(θ, θ , τ ) − w0 (θ, θ , τ )) , where w0 (θ, θ , τ ) is the distribution function of the free particle (4.27), which does not contribute to the electromagnetic emission (since the VavilovCherenkov condition cannot be fulﬁlled in the plasma or vacuum). Then, calculation of (4.29), described in detail in [21], results in 2 γ 2 ωpe Q2 ω 8Q2 q (ω) γ 2 Φ1 (s1 , s2 , r) + 1+ Φ2 (s1 , s2 , r) , (4.30) Iω = 3πc [1 + γ 2 (ωpe /ω)2 ] 4πcγ 2 ω2 where Φ1 (s1 , s2 , r) and Φ2 (s1 , s2 , r) stand for the integrals: ∞ 6|s|4 Φ1 (s1 , s2 , r) = Im dt exp(−2st) × (4.31) s1 s2 0

t 1 × coth t exp −2rs3 coth t − sinh−1 t − − , 2 t

4 Generation of Emissions by Fast Particles in Stochastic Media

95

∞ cosh t − 1 × (4.32) Φ2 (s1 , s2 , r) = 2r|s|2 Re dt sinh t 0

t × exp −2st − 2rs3 coth t − sinh−1 t − − iφ , 2

which depend on the dimensionless parameters s1 , s2 , r: φ

12 π 2 2 ωpe γ ω e−i( 4 + 2 ) √ s = s1 − is2 = 1+ , |q(ω)| ω2 4 2γ 2 −3

2 2 2 ωpe γ Ω⊥ 6 r = 32γ . 1+ ω ω2

(4.33)

(4.34)

The parameter s depends on the rate of scattering of the particle by magnetic inhomogeneities q(ω), which has the form √ 2 ν−1 2 πΓ (ν/2)ωst ω0 α (ν − 1)ωst q(ω) = , (4.35) +i 2 2 2 2 2 ν/2 3γ ω0 (1 + (ν − 1)α2 /ω02 ) 3Γ (ν/2 − 1/2)γ (α + ω0 ) for power-law distribution of magnetic irregularities over scales: P (k) = 2 2 4πK(k)k 2 ∝ k −ν at k kmin = ω0 /c, where ωst = Q2 Bst /(M c)2 is the square of the cyclotron frequency in the random magnetic ﬁeld, α = 2 /ω 2 , and a is a factor of the order of unity. (aω/2) γ −2 + ωpe In the general case the integrals (4.31, 4.32) cannot be expressed in terms of elementary functions. However, there are convenient asymptotic expressions of these integrals. In particular, if r|s|3 1 and r|s|3 |s| we have 1 1 1 2 |s|4 (2π − 3φ), Φ2 ≈ 2 3 3 6 Γ (4.36) Φ1 ≈ − r3 , s1 s2 3 while for |s| 1, r 1, the functions Φ1 and Φ2 contain exponentially small terms: ( 1 1 3π 2 r 4 |s|4 8 2 , (4.37) Φ1 ≈ 1 − exp − 5 3 r 2 4 s1 s2 ( 3 1 1 r 8 2 . (4.38) Φ2 ≈ + 2 4 π 2 r 4 exp − 32|s|2 3 r Complementary, for s 1 and rs3 1 we obtain Φ1 ≈ 6s

1 − rs2 , 1 + r2 s4

Φ2 ≈

rs 1 + rs2 . 2 1 + r2 s4

(4.39)

4.3.2 Special Cases The radiation intensity (4.30) depends on many parameters, allowing many diﬀerent parameter regimes. It is clear that in the absence of the random ﬁelds we arrive at standard expressions for synchrotron radiation in a uniform magnetic ﬁeld. Let us consider here a few interesting cases when the presence of small-scale random ﬁeld results in a considerable change of the emission.

96

G.D. Fleishman

Weak Random Magnetic Inhomogeneities Superimposed on Regular Magnetic Field Consider the case of weak magnetic irregularities with a broad (power-law) 2 & 2 , so that distribution over spatial scales, with B⊥ B ωc ω &st ω0 .

(4.40)

Here ωc is the gyro-frequency, and ω &st is the gyro-frequency related to the & 2 12 . total random ﬁeld B Radiation from highly relativistic particles [Toptygin and Fleishman, 21] is mainly speciﬁed by the regular ﬁeld, since either |s| 1 or r|s|3 1. However, at high frequencies, where synchrotron radiation decreases exponentially, the spectrum is controlled by the small-scale ﬁeld: the spectral index of radiation is equal to the spectral index of the random ﬁeld, see Fig. 4.1.

Fig. 4.1. Spectra of radiation by a relativistic particle with γ = 104 for diﬀering value of the random magnetic ﬁeld (left) and with diﬀerent γ in the presence of weak 2 /B02 = 10−4 (right) random magnetic ﬁeld Bst

However, a more interesting regime, which has not been considered so far, takes place for moderately relativistic (and non-relativistic) particles moving in a dense plasma (the case typical for solar and geospace plasmas), when synchrotron radiation is known to be exponentially suppressed according to (4.37), (4.38) by the eﬀect of plasma density (Razin-eﬀect [5, 19]) at all frequencies. The contribution of the small-scale random ﬁeld, which we refer to as diﬀusive synchrotron radiation, in this conditions takes the form: 2 ν−1 2ν ωst ω0 γ Q2 2ν+1 Γ (ν/2)(ν 2 + 7ν + 8) . (4.41) Iω = √ 2 3 πΓ (ν/2 − 1/2)(ν + 2) (ν + 3) c ω ν 1 + ω 2 γ 2 /ω 2 ν+1 pe

It is important that this radiation decreases with the increase of the plasma −ν ) than density (plasma frequency) much more slowly (as a power-law, ∼ ωpe synchrotron radiation. As a result, the diﬀusive synchrotron radiation can

4 Generation of Emissions by Fast Particles in Stochastic Media

97

Fig. 4.2. Spectra of radiation by a relativistic particle with diﬀerent γ in a dense 2 /B02 = 10−6 (left), and plasma in the presence of weak random magnetic ﬁeld Bst 4 with γ = 10 in small-scale random magnetic ﬁeld (right). If ω0 is big enough (e.g., ω0 /ωce = 107 in the ﬁgure) the spectral region provided by multiple scattering, ω 1/2 , disappears

dominate the entire spectrum even if the random ﬁeld is much weaker than the regular ﬁeld, as is evident from the left part of Fig. 4.2: the emission by < 10 is deﬁned exclusively by the small-scale ﬁeld. particles with γ ∼ Small-Scale Magnetic Field Consider an extreme case, which might be relevant in the physics of cosmological gamma-ray bursts, when there is only a small-scale random magnetic ﬁeld but no (very weak) regular ﬁeld, so that ω0 ωst [4]. Now, the parameter q depends substantially on ω0 , the particle motion is similar to the random walk, so the radiation spectrum is similar to some extent to bremsstrahlung provided by multiple scattering of the fast particle by randomly located Coulomb centers. In particular, the spectrum of diﬀusive synchrotron radiation can 1 contain a ﬂat region (as standard bremsstrahlung) and a ∝ ω 2 region (like bremsstrahlung suppressed by multiple scattering), Fig. 4.2 right. However, at suﬃciently high frequencies (ω > ω0 γ 2 ), the ﬂat spectrum gives way to a power-law region ∝ ω −ν typical for the diﬀusive synchrotron radiation. We should note, that the spectrum depends signiﬁcantly on the energy of radiating particle (compare the left part in Fig. 4.3). For low-energy particles some parts of the spectrum (e.g., ﬂat region) might be missing. 4.3.3 Emission from an Ensemble of Particles The results presented in the previous section can be directly applied to monoenergetic electron distributions, which can be generated in the laboratory, but are rare exceptions in nature (e.g., in astro- and geo-plasmas). Natural particle distributions can frequently be approximated by power-laws, say, as function of the dimensionless parameter γ:

98

G.D. Fleishman

Fig. 4.3. Spectra of radiation by a relativistic particle with diﬀerent γ = 30, 3 · 103 , 106 in the presence of small-scale random magnetic ﬁeld (left). Emissivity by fast electron ensemble with diﬀerent energetic spectra (ξ = 2.5, 4.5, 6.5) for the case of dense plasma, ωce /ωpe = 3 · 10−3 (right)

dNe (γ) = (ξ − 1)Ne γ1ξ−1 γ −ξ , γ1 ≤ γ ≤ γ2 ,

(4.42)

where Ne is the number density of relativistic electrons with energies E ≥ mc2 γ1 , and ξ is the power-law index of the distribution. Evidently, the intensity of incoherent radiation produced by the ensemble (4.42) of electrons from the unit source volume is Pω = Iω dNe (γ) . (4.43) Hard Electron Spectrum As we will see, the radiation spectrum produced by an ensemble of particles diﬀers for hard (ξ < 2ν + 1) and soft (ξ > 2ν + 1) distributions of fast electrons over energy. Let us consider ﬁrst the case of hard spectrum [Toptygin and Fleishman, 21], which is typical, e.g., for supernova remnants and radio galaxies. Assuming the small-scale ﬁeld to be small compared with the regular ﬁeld, we may expect the contribution of diﬀusive synchrotron radiation to be noticeable only in those frequency ranges where synchrotron 'emission is small. 2 /ωc⊥ , ωpe ωpe γ1 /ωc⊥ ), synIn particular, at low frequencies ω max(ωpe chrotron radiation is suppressed by the eﬀect of density. Diﬀusive synchrotron radiation is produced by relatively low energy electrons at these frequencies, and each electron produces the emission according to (4.41), which peaks at ω ∼ ωpe γ. Evaluation of the integral (4.43) gives rise to

ν+1−ξ 2 ν−1 (ξ − 1)Γ ν2 (ν 2 + 7ν + 8) e2 Ne γ1ξ−1 ωst ω ω0 Pω √ ν c ωpe ωpe 3 πΓ ν2 − 12 (ν + 2)2 (ν + 3) (4.44) in agreement with Nikolaev and Tsytovich [14]. The spectrum can either increase or decrease with frequency depending on the spectral indices ν and ξ.

4 Generation of Emissions by Fast Particles in Stochastic Media

99

This expression holds for ω ωpe γ1 . If there are no particles with γ < γ1 , the spectrum at even lower frequencies drops as 2 ν−1 ν+2 ω0 ω e2 Ne ωst (ξ − 1)2ν+1 Γ (ν/2)(ν 2 + 7ν + 8) √ . (4.45) 2ν+2 3(ξ + 1) πΓ (ν/2 − 1/2)(ν + 2)2 (ν + 3) c γ12 ωpe ' 2 /ωc⊥ , ωpe ωpe γ1 /ωc⊥ ) ω ωc⊥ γ22 , where At high frequencies max(ωpe the eﬀect of density is not important, the spectrum is speciﬁed by standard synchrotron radiation. However, at higher frequencies, ω ωc⊥ γ22 , the intensity of synchrotron radiation decreases exponentially, and the contribution of diﬀusive synchrotron radiation dominates again. Adding up contributions from all particles described by (4.41) at these frequencies, we obtain

Pω =

2 ν−1 2ν−ξ+1 2ν+1 (ξ − 1)Γ ( ν2 )(ν 2 + 7ν + 8) e2 Ne γ1ξ−1 ωst ω0 γ2 ν . Pω = √ c ων 3 π(2ν − ξ + 1)Γ 2 − 12 (ν + 2)2 (ν + 3) (4.46) Thus, power-law spectrum of relativistic electrons with a cut-oﬀ at the energy E = mc2 γ2 produces diﬀusive synchrotron radiation at high frequencies, whose spectrum shape is deﬁned by the small-scale ﬁeld spectrum. Remarkably, the corresponding ﬂattening in the synchrotron cut-oﬀ region has recently been detected in the optical-UV range for the jet in the quasar 3C273 [8], which would imply the presence of a relatively strong small-scale ﬁeld there in agreement with the model of Honda and Honda [6]. Although formally the spectrum (4.46) is valid at arbitrarily high frequencies, there is actually a cut-oﬀ related to the minimal scale of the random ﬁeld lmin . Accordingly, the largest frequency of the diﬀusive synchrotron radiation is about ωmax ∼ (c/lmin )γ22 .

Soft Electron Spectrum Let us now turn to the case of suﬃciently soft electron spectra, ξ > 2ν + 1, which are typical, e.g., in many solar ﬂares. The contribution of synchrotron radiation is described by the standard expression, Pω ∝ ω −α , α = (ξ − 1)/2, which is steeper for soft spectra than the spectrum of diﬀusive synchrotron radiation, Pω ∝ ω −ν . Hence, for soft electron spectra, diﬀusive synchrotron radiation can dominate even at ω < ωc⊥ γ22 . The spectrum of diﬀusive synchrotron radiation has the same shape as before but its level is deﬁned by lower-energy electron contribution: 2 ν−1 2ν+1 (ξ − 1)Γ ν2 (ν 2 + 7ν + 8) e2 Ne γ12ν ωst ω0 Pω = √ . (4.47) ν 1 c ων 3 π(ξ − 2ν − 1)Γ 2 − 2 (ν + 2)2 (ν + 3) At low frequencies, ω ωpe γ1 , the radiation is still speciﬁed by expression (4.44), see the corresponding curves on the right in Fig. 4.3. One may note that in the case of soft electron spectra, the emission produced by the electron ensemble is similar to the emission from a mono-energetic electron distribution

100

G.D. Fleishman

with γ = γ1 , which is the main diﬀerence between the cases of hard and soft electron spectra. Let us estimate the ratio of the diﬀusive synchrotron radiation intensity to the synchrotron radiation intensity. For simplicity, we neglect factors of the order of unity, assume ω0 = ωce and γ1 ∼ 1, and introduce frequency 2 /ωce (ω ≡ (ω/ω∗ )ω∗ ), where synchrotron radiation has a peak, then ω∗ = ωpe Psh ω2 ∼ st 2 Psyn ωce

ωpe ωce

(

ω ω∗

ξ−2ν−1 .

(4.48)

Evidently, this ratio increases with frequency, so that diﬀusive synchrotron radiation can become dominant well before the frequency reaches ωc⊥ γ22 . Moreover, in the case of dense plasma, ωpe ωce , diﬀusive synchrotron radiation can dominate at all frequencies under the condition 2 ωst 2 ωce

ωpe ωce

ξ−2ν−1 >1,

(4.49)

even if the random ﬁeld is small compared with the regular ﬁeld ωst ωce . On top of this, the radiation spectrum produced from the dense plasma depends critically on the highest energy of the accelerated electrons. Indeed, if γ2 ωpe /ωce (e.g., γ2 = 4 in Fig. 4.4), then the radiation spectrum is entirely 2 /B02 = 10−4 in set up by the small-scale ﬁeld, in spite of its smallness ( Bst Fig. 4.4). Evidently, the standard synchrotron emission increases and becomes observable as far as γ2 increases.

Fig. 4.4. Left: Same as in Fig. 4.3, right, for less dense plasma, ωce /ωpe = 3 · 10−2 . The contribution from the uniform ﬁeld (synchrotron radiation) decreases for softer electron spectra (i.e., as ξ increases). Right: Emissivity by fast electron ensemble with (ξ = 6) from dense plasma (ωce /ωpe = 3 · 10−2 ) in the presence of weak magnetic 2 /B02 = 10−4 for diﬀerent high-energy cut-oﬀ values γ2 . When inhomogeneities Bst γ2 is small enough, the uniform magnetic ﬁeld does not aﬀect the radiation spectrum

4 Generation of Emissions by Fast Particles in Stochastic Media

101

Diﬀusive Synchrotron Radiation from Solar Radio Bursts? According to microwave and hard X-ray observations of solar ﬂares, the energetic spectra of accelerated electrons are frequently rather soft [9, 15]. Consequently, the diﬀusive synchrotron radiation can dominate the microwave emission for dense enough radio sources. Nevertheless, as a rule the microwave emission from solar ﬂares meets reasonable quantitative interpretation as synchrotron (gyrosynchrotron) radiation by moderately relativistic electrons (moving in non-uniform magnetic ﬁeld of the coronal loop). An example of microwave burst produced by gyrosynchrotron emission is given in Fig. 4.5, left.

Fig. 4.5. Two microwave bursts recorded by Owens Valley Solar Array in the range 1–18 GHz with 40 spectral channels and 4 sec temporal resolution. The ﬁrst one (left bottom) displays evident spectral hardening with time, while the second one shows remarkable constancy of the high-frequency spectral slope (courtesy of D.E. Gary)

Note, that the high-frequency spectral index δ (the radio ﬂux is ﬁtted by a power-law, F ∝ f δ , at high frequencies, Fig. 4.5, bottom left) decreases in value with time. Such spectral evolution typical for solar microwave bursts is well-understood in the context of the energy-dependent life time of electrons against the Coulomb collisions. Indeed, higher energy electrons have longer life times, which results in spectral hardening of the trapped electron population [Melrose and Brown, 12], and, respectively, hardening of the produced gyrosynchrotron radiation as observed by Melnikov and Magun [11]. However, if microwave emission is produced by diﬀusive synchrotron radiation as fast electrons interact with small-scale magnetic (and/or electric)

102

G.D. Fleishman

ﬁelds, then the radio spectrum is speciﬁed by the spectrum of random ﬁelds rather then of fast electrons. Thus, no spectral evolution (related to electron distribution modiﬁcation) is expected. Indeed, there is a minority of solar microwave bursts, which do not show any spectral evolution (e.g., no spectral hardening). An example of such a burst, demonstrating constancy in time of the high-frequency spectral index, is shown in Fig. 4.5 at the right. Curiously, the spectral index is δ = −1.5 to − 1.7 in agreement with standard models [Vainshtein et al., 22] and measurements of the turbulence spectra, e.g., in interplanetary [see, e.g., Toptygin, 20] and interstellar [Cordes et al., 1] space. Although it has not been ﬁrmly proven so far, such microwave bursts are possibly produced by diﬀusive synchrotron radiation mechanism. Since (to be dominant) this mechanism requires relatively dense plasma at the source site and soft spectra of accelerated electrons, the observational evidence can be found from analysis of simultaneous observations of soft and hard X-ray emissions from the same ﬂares.

4.4 Discussion The analysis presented demonstrates the potential importance of small-scale turbulence in the generation of radio emission from natural plasmas. This emission, being reliably detected and interpreted, provides the most direct measurements of small-scale turbulence in the remote sources. The diﬀusive synchrotron radiation is only one of the observable eﬀects of the turbulence on the radio emission. Indeed, the presence of density inhomogeneities aﬀects the properties of bremsstrahlung, because the Fourier transform of the square of the electric potential produced by background charges in a medium depends on spatial distribution of the charges through the double sum: * ) |∆N |2q 2 −iq(RA −RB ) 3 = N + (2π) , (4.50) e | ϕq0 ,q | ∝ V A,B

where RA and RB are the radius-vectors of the particles A and B, respectively, |∆N |2q is the spectrum of the inhomogeneity, and V is the volume of the system. In statistically uniform media the positions of various particles are uncorrelated and this double sum equals the total number of particles N . However, the macroscopic inhomogeneities make the positions correlated, so the double sum deviates from N . The second term in (4.50) gives rise to coherent bremsstrahlung, which in a certain spectral range dominates the incoherent bremsstrahlung [Platonov et al., 17]. Another important radiation process in the turbulent plasma is transition radiation arising when fast particles interact with small-scale density inhomogeneities of the background plasma (see Platonov and Fleishman [18], and references therein), whose potential importance for ionospheric conditions has been pointed out long ago by Yermakova and Trakhtengerts [23] (see also the

4 Generation of Emissions by Fast Particles in Stochastic Media

103

discussion in LaBelle and Treumann [10]). This emission process, giving rise to enhanced low-frequency (at frequencies lower than the accompanying synchrotron emission) continuum radio emission, has recently been reliably conﬁrmed in a subclass of two-component solar radio bursts [3, 16]. This ﬁnding is of particular importance for diagnostics of the number density, the level of small-scale turbulence, and the dynamics of low-energy fast particles in solar ﬂares. In addition, the turbulence can also aﬀect the coherent emissions from unstable electron populations [Fleishman et al., 2], e.g., providing strong broadening (or splitting) of the spectral peaks generated by electron cyclotron maser (ECM) emission. The typical bandwidth of the broadened ECM peaks and its distributions are found to be quantitatively consistent with those observed for narrowband solar radio spikes [Fleishman et al., 2].

Acknowledgements The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. This work was supported in part by the Russian Foundation for Basic Research, grants No. 03-02-17218, 04-02-39029. I am very grateful to T. S. Bastian for his numerous comments on this paper.

References [1] Cordes, J.M., M. Ryan, J.M. Weisberg, D.A. Frail, and S.R. Spangler: Nature 354, 121 (1991). [2] Fleishman, G.D.: Astrophys. J. 601, 559 (2004). [3] Fleishman, G.D., G.M. Nita, and D.E. Gary: Astrophys. J. 620, 506 (2005). [4] Fleishman, G.D.: Astro-ph/0502245 (2005). [5] Ginzburg, V.L. and S.I. Syrovatsky: Ann. Rev. Astron. Astrophys. 3, 297 (1965). [6] Honda, M. and Y.S. Honda: Astrophys. J. 617, L37 (2004). [7] Jaroschek, C. H., H. Lesch, and R. A. Treumann: Astrophys. J. 618, 822 (2005). [8] Jester, S., H.-J. R¨ oser, K. Meisenheimer, and R. Perley: Astron. Astrophys. 431, 477 (2005). [9] Kundu, M.R., S.M. White, N. Gopalswamy, and J. Lim: Astrophys. J. Suppl. 90, 599 (1994). [10] LaBelle, J. and R. A. Treumann: Space Sci. Rev. 101, 295 (2002). [11] Melnikov, V.F. and A. Magun: Solar Phys. 178, 153 (1998). [12] Melrose, D.B. and J.C. Brown: Monthly Notic. Royal Astron. Soc. London 176, 15 (1976). [13] Migdal, A.B.: Dokl. Akad. Nauk SSSR (in Russian) 96, 49 (1954); Phys. Rev. 103, 1811 (1956). [14] Nikolaev, Yu.A. and V.N. Tsytovich: Phys. Scripta 20, 665 (1979).

104

G.D. Fleishman

[15] Nita, G.M., D.E. Gary, and J. Lee: Astrophys. J. 605, 528 (2004). [16] Nita, G.M., D.E. Gary, and G.D. Fleishman: Astrophys. J. 629, L65 (2005). [17] Platonov, K.Yu., I.N. Toptygin, and G.D. Fleishman: Uspekhi Fiz. Nauk (in Russian) 160, 59 (Engl. transl.: Sov. Phys. Uspekhi, 33, 289) (1990). [18] Platonov, K.Yu. and G.D. Fleishman: Physics-Uspekhi 45, 235 (2002). [19] Razin, V. A.: Radioﬁzika 3, 584 (1960). [20] Toptygin, I.N.: Cosmic rays in interplanetary magnetic ﬁelds (DordrechtHolland, D. Reidel, 1985) 387 pp. [21] Toptygin, I.N. and G.D. Fleishman: Astroph. Space Sci. 13, 213 (1987). [22] Vainshtein, S.I., A.M. Bykov, and I.N. Toptygin: Turbulence, Current Sheets, and Shocks in Cosmic Plasma (The Fluid Mechanics of Astrophysics and Geophysics, Vol. 6, Langhorne: Gordon and Breach Science Publ., 1993) 398 pp. [23] Yermakova, E.N. and V.Yu. Trakhtengerts: Geomagn. Aeron. (Engl. Transl.) 21, 56 (1981).

5 Auroral Acceleration and Radiation R. Pottelette1 and R.A. Treumann2,3 1

2

3

CETP/CNRS, 4 av. de Neptune, 94107 St. Maur des Foss´es Cedex, France [email protected] Ludwig-Maximilians Universit¨ at M¨ unchen, Sektion Geophysik, Theresienstr. 37-41, 80333 M¨ unchen, Germany, [email protected] Department of Physics and Astronomy, Dartmouth College, Hanover, New Hampshire 03755, USA

Abstract. A brief review is given of the recent achievements in understanding the connection between processes in the generation of auroral acceleration and processes taking place at the tailward reconnection site. It is shown that most of the acceleration in the aurora is due to local ﬁeld-aligned electric potentials which are located in vertically narrow double layers along the magnetic ﬁeld of the order of ∼10 km and which are the site of preferential excitation of phase space holes of kilometer size extension along the magnetic ﬁeld which by themselves sometimes represent local potential drops and accelerate electrons and ions antiparallel to each other such that the energy modulation of the electron and ion energy ﬂuxes are in antiphase. Auroral kilometric radiation observations suggest that these structures may be the elementary radiation sources which build up the entire spectrum of the auroral kilometric radiation. Leaving open the very generation mechanism of the double layer whether produced by locally applied shear ﬂows as recently suggested in the literature (and reviewed here as well) we argue that the ﬁeld aligned current generator responsible for the production of the initial auroral current system is non-local but is related to reconnection in the tail. The ﬁeld aligned currents are interpreted as the closure currents required to close the recently observed electron Hall current system in the ion diﬀusion region at the tail reconnection site. Such a model is very attractive as it does not need any other secondary current disruption mechanism. Coupling to the ionosphere may be provided by kinetic Alfv´en waves emanating from the Hall reconnection region as surface waves and generating local shear ﬂow when focussing close to the ionosphere and transforming into shear-kinetic Alfv´en waves. A main problem still remains in how the decoupling of the two hemispheres observed in the aurora is produced at reconnection site. Multiple reconnection would be one possible solution.

Key words: Auroral processes, auroral particle acceleration, double layers, auroral kilometric radiation, elementary radiation sources, electron holes, reconnection, generator region

R. Pottelette and R.A. Treumann: Auroral Acceleration and Radiation, Lect. Notes Phys. 687, 105–138 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

106

R. Pottelette and R.A. Treumann

5.1 Introduction For many years it has been a mystery why electrons and ions are accelerated in the auroral zone. The idea of St¨ ormer [40] that cosmic ray particles spiral in along Earth’s magnetic ﬁeld to excite the atoms and ions of the upper atmosphere and ionosphere and generate the aurora has readily been proven wrong. Cosmic rays are too energetic to cause anything like an aurora. For them the Earth’s magnetic ﬁeld is practically invisible. They impinge on the ionosphere isotropically causing showers of secondary elementary particles and become absorbed in the atmosphere at altitudes way below those of the aurora. Solar energetic particles on the other hand are not energetic enough to reach anywhere deep enough into the ionosphere. According to their magnetic hardness they are excluded from the inner magnetosphere, while when ﬂowing in from their forbidden zone boundary they are still too energetic to cause substantial aurora. Moreover, they appear at too low latitudes to be associated with auroral phenomena. In addition, aurora is located most frequently on the antisolar side of the magnetosphere even though there is also aurora on the dayside under the cusp and driven by particles accelerated in magnetopause reconnection. These dayside aurorae complete the auroral oval. The energy range required for electrons to become eﬀective in generating aurorae lies between several 100 eV and several keV. The electron distribution in the magnetospheric tail, i.e. the plasma sheet, is at the lower end of this range. It thus can serve as the source for the auroral electron distribution. However, it must become additionally accelerated by some secondary acceleration process which up till now has only been badly identiﬁed. There is a high probability that this process does not consist of one unique step but that the acceleration of electrons takes place in a primary step followed by a sequence of a few or several secondary acceleration steps until the electrons have enough energy to play their role in the aurora. Presumably these two steps are located in spatially separate regions of the magnetosphere. Since aurorae occur during magnetic substorms and storms which are believed to be the result of violent reconnection going on in the magnetospheric tail current sheet, it is reasonable to assume that the ﬁrst step of acceleration is directly related to the tailward reconnection site. The second step, on the other hand, has been found to occur much closer to the ionosphere at altitudes of ∼2000–8000 km. Both steps act in concert but from the point of view of the aurora the second step is the more interesting one. It is also related to the generation of the famous auroral kilometric radiation which, under special plasma conditions, is emitted from the acceleration region. In the present paper we brieﬂy review the processes in the second step acceleration region adding some founded speculations on the processes in the ﬁrst step acceleration region.

5 Auroral Acceleration and Radiation

107

5.2 Morphology of the Auroral Acceleration Region During the past decade various spacecraft have passed the auroral zone at diﬀerent altitudes. The lowest ﬂying spacecraft was the Swedish satellite Freja (at a nominal altitude of ∼2000 km), followed by the NASA satellite FAST (at nominal altitude ∼4000 km) and the Swedish satellite Viking (at ∼8000 km altitude). Also, Geotail and Polar have passed magnetic ﬁeld lines connected to the auroral zone at much larger distances from Earth thus providing occasional information about auroral region-magnetosphere connections. According to these observations we may divide the auroral-magnetosphere connection into two diﬀerent regions: the lower magnetospheric auroral region and the tail auroral region. We will ﬁrst discuss the former as the processes taking place there have recently been illuminated best. Traditionally this region is divided into a sequence of ﬁeld-aligned current elements: the upward current and the downward current regions, respectively. Since the auroral ﬁeld-aligned currents are carried mainly by electrons with the ions contributing only little, the upward current region is the region where electrons from the magnetosphere ﬂow down into the ionosphere. Traditionally this region is called the “inverted-V” a name adapted from the Λ-like shape of the electron energy ﬂuxes versus time – or space – during satellite crossings of the auroral upward current region. An example is shown in Fig. 5.1 for a relatively long-lasting “inverted-V” event which in the low time resolution looks rather like a step function with sharp onset in the electron energy ﬂux around a few keV at the low latitude boundary of the event followed by a long-lasting plateau and a correspondingly sharp dropout of the ﬂux at the high latitude boundary. In high time resolution this increase is rather gradual indicating that the potential drops on a somewhat longer spatial distance. The whole event lasts for roughly 30 s, a very short time which for a spacecraft velocity of ∼6 km s−1 corresponds to the short horizontal distance of only ∼180 km in the corresponding northern hemispherical auroral passage of FAST. We will return to this particular fact in Sect. 5.6 below. In fact, the length of the “inverted-V” event is marked even better by the ion energy ﬂuxes than by the electron ﬂuxes. This is very clearly seen in Fig. 5.1 where the sharp increase in the ion energy accompanies the changes in energetic electron energy ﬂux. Before entering into a discussion of the entire sequence of data during this short passage we note that this short “inverted-V” event is a section of a longer auroral disturbance shown in Fig. 5.2 in very low time resolution. This entire event lasts for ∼5 min, corresponding to a horizontal distance of ∼1800 km on the northern hemispherical auroral region in the altitude range between 4000 km and 3500 km. It covers a latitudinal range of roughly 5 degrees in invariant latitude and a longitudinal range of roughly 1 hour in magnetic local time (or 15 degrees) in the pre-midnight sector. Thus the entire event is located fully on the northern hemisphere. This is important to remember for our later discussion. However, the self-sustained short time section shown in Fig. 5.1 which is an “inverted V” on its own shows that the full event consists

108

R. Pottelette and R.A. Treumann

Fig. 5.1. An overview of data from FAST orbit 1843 during a passage of the region above an aurora on the northern hemisphere

of a sequence of several sub-events more or less well covered by the spacecraft orbit, each of them a separate “inverted V”. Inspection of the overall event thus suggests that it consists of a mixture of a sequence of more or less well developed upward and downward current regions, and in order to obtain a true impression of its nature it would be necessary to investigate its higher

5 Auroral Acceleration and Radiation

109

Fig. 5.2. The whole auroral disturbance on FAST orbit 1843 as seen in the electron ﬂux. Shown are the downward (upper panel), the perpendicular (second panel), the upward electron ﬂux (third panel), and pitch angle distributions selected for two electron energy ranges (fourth and ﬁfth panels) in very low time resolution overview (courtesy C. W. Carlson, UCB)

110

R. Pottelette and R.A. Treumann

time resolution in order to separate it into parts which physically represent regions of either upward or downward currents. With this idea in mind we return to Fig. 5.1. The ﬁrst panel in Fig. 5.1 shows the perpendicular component of the magnetic ﬁeld which is caused by ﬁeld aligned currents. The positive slope of the ﬁeld indicates downward currents corresponding to upward electron ﬂux, while negative slopes occur in upward currents. The former are narrow in time and space and highly variable, while the latter are less variable and broader in time and space, corresponding to the dilute downward energetic auroral electron beam emanating from somewhere in the magnetosphere. These electron ﬂuxes are shown in the ﬁfth panel. Combined with the pitch angle distribution in the sixth panel one indeed realizes that intense upward highly variable electron ﬂuxes correlate with the downward current region, and relatively stable energetic electron ﬂuxes correlate with the upward current region. In this particular event the downward electron energies are close to 10 keV, while the upward electron energies cover a broad energy range from 10 eV upward up to 10 keV. Figure 5.3 shows a schematic of the auroral current system inferred from a sequence of upward and downward electron ﬂuxes [Elphic et al., 11]. Such a ﬁgure suggests that the upward and downward currents form one closed current system. If one assumes that the generator of the currents is in the magnetosphere, a reasonable assumption, then the upward currents are the primary currents while the downward currents are the return currents which close the system. The connection between the two is given by ionospheric Pedersen currents ﬂowing perpendicular to the magnetic ﬁeld ﬂux tubes in the ionosphere. Depending on the direction of the perpendicular ionospheric electric ﬁeld these currents may ﬂow in the plane of the upward and downward electron ﬂuxes or deviate from it forming a three-dimensional current system. In fact the direction of the electric ﬁeld will be imposed by the mapping of the magnetospheric electric potential along the magnetic ﬁeld into the ionosphere which in the most general case will be 3d as the transport of the potentials is done by shear Alfv´en waves. Figure 5.3 thus is an idealization which assumes that everything happens in the plane containing the upward and downward currents. It is, however, clear that in order to maintain both upward and downward currents in a closed current system the electrons, which carry the current must have been accelerated both from the magnetosphere down into the ionosphere and from the ionosphere up into the magnetosphere in order to arrive at closed and divergence free currents. 5.2.1 Upward Current Region The upward current region or downward electron acceleration region is in our view the primary acceleration region. Ignoring for the moment the primary magnetospheric acceleration processes, the data suggest that it contains a warm energetic primary electron beam of a few keV energy and several 100 eV velocity spread or temperature. This beam is not necessarily Maxwellian.

5 Auroral Acceleration and Radiation

Subauroral Latitude

Auroral Zone

111

Polar Cap

FAST Orbit

ionospheric density

Fig. 5.3. Schematic of the auroral ﬁeld-aligned current system inferred from a path of FAST across a sequence of downward and upward electron ﬂuxes [Elphic et al., 11] (Reprinted with permission of the American Geophysical Union)

Figure 5.4 shows on its right a cut through the electron distribution function (for FAST orbit 1773) along the magnetic ﬁeld. The auroral beam sits on a broad plateau in the distribution function. The increase at low velocities (energies) is due to electrons 0 which is the condition for the presence of free energy and a “lifted” to higher energy level distribution function, and the particle resonance condition

Perpendicular velocity v

v (km/s)

10 5

0

los s 0 cone

-105 10 5

0 Parallel velocity v

=

=

0 v (km/s)

Fig. 5.18. Left: Measured [after Delory et al., 9, with permission of the American Geophysical Union] and Right: model horseshoe electron distribution in the auroral region. The measured distribution consists of the spread electron beam moving down along the ﬁeld line and a plateau which is produced by the quasilinear interaction of the distribution with the self-generated VLF waves. The loss cone on the left in both distributions is practically empty

5 Auroral Acceleration and Radiation

131

Fig. 5.19. Left: Schematics of the mechanism of generation of horseshoe distributions in a converging magnetic ﬁeld. The converging ﬁeld reﬂects the particles while the parallel electric potential accelerates the particles. These gain in this way perpendicular energy. The result is the distribution on the Right: which is a horseshoe distribution. The formal analytical treatment of this process has been given by Chiu and Schulz [8]

k vr /2π = f − fce /γr

(5.4)

where k is the parallel wave number, f the wave frequency, fce the local 1 electron cyclotron frequency, and γr = [1 − v 2 /c2 ]− 2 the relativistic energy factor. Clearly, for strictly perpendicular radiation k = 0, the resonance line becomes a circle in the (v , v⊥ )-plane which can be positioned along the nearly circular horseshoe distribution where the derivative in the above equation is positive and maximum. This is the simple idea of emission in the x-mode. A theory like this generates very broadband emission unless the distribution is deformed locally very strongly. Inspection of the high-resolution emission spectrum however demonstrates that strongest emission is generated in very narrow bands which by themselves drift on variable speed across the frequency-time spectrogram. Such situations are shown in Figs. 5.20 and 5.21 for two diﬀerent cases: when the narrow band emission is standing and when it is moving. Figure 5.21 in addition shows the observation of two such interacting narrow band AKR sources of high intensity. In this case the two move initially at diﬀerent velocities. When interacting, they obviously repel each other and move together at the same speed to higher frequencies downward along the magnetic ﬁeld. This can be interpreted as both emission regions being of same polarity and thus not mixing.

132

R. Pottelette and R.A. Treumann

Fig. 5.20. An example of a narrow emission band in AKR which in this case is stationary in space and frequency FAST orbit 1761 kHz 445

Frequency

440

435

430

1997-01-31/06:44:56.002 UT

425 0

2

4 Time (s)

6

Fig. 5.21. Left: Two narrow band AKR emission lines moving across the frequencytime spectrogram approach each other, collide and move together down along the magnetic ﬁeld at same speed. Right: The explanation of this case as two electric ﬁeld structures of similar amplitude and equal polarity approaching each other [adapted from Pottelette et al., 32] (Reprinted with permission of the American Geophysical Union)

Another very interesting observation can be made from these two ﬁgures. They show that the emission of AKR is not continuum but consists of the superposition of a large number of “elementary radiation events” (ERE’s) the emitted radiation of which superimposes to generate the apparently

5 Auroral Acceleration and Radiation

133

continuous AKR spectrum. The nature of these elementary radiators has not yet been determined. However, in light of the above discussion of the dynamics of the auroral plasma it seems quite natural to assume that the elementary radiators are the electron holes which are generated in the plasma and move on the background of the horseshoe distribution. If this is true then the same theory applies as before with two modiﬁcations: • the electron holes must generate steep gradients on the distribution in perpendicular velocity which implies that they must be bent in phase space; • the emission from each electron hole is then not necessarily perpendicular as this is no longer required by the resonance condition. It will, however, be inclined, and the strongest emission is detected when the spacecraft will be in the emission cone of the elementary radiators. The existence of strong emission lines which move together may suggest that many such radiators have become attracted to a certain region and move approximately together over relatively long times. This is possible when realizing that the electron holes are eﬀectively positive charges on the electron background which implies that they will interact in a certain way with the main electron component becoming attracted by the bulk of the electrons and thus trapped in the electron distribution while at the same time growing in amplitude as their absolute depth will be conserved. This can lead to a collection of holes in a more narrow space region which then acts as the radiation source. The questions related to these problems have not yet been solved and are subjects of ongoing research.

5.5 The Tail Acceleration Region We now come to the discussion of the ultimate source of the auroral electron beam. This discussion will be rather brief as there is no consensus about this hot problem yet. It seems, however, reasonable to assume that processes in the tail of the magnetosphere provide the primary energy source since aurora are most prominent during substorms which are directly related to reconnection in the tail current sheet. It is thus natural to consider the structure of the reconnection site and to speculate about the reconnection site being the ultimate energy source of aurorae. It has recently been shown by Øieroset et al. [30] that the reconnection diﬀusion site in the tail current sheet is the site of separate electron and ion dynamics. On the small scale of the ion diﬀusion length c/2πfpi the motion of electrons and ions becomes diﬀerent. The ions in the plasma carrying the tail current are eﬀectively non-magnetic on this scale while the electrons are still tied to the magnetic ﬁeld. This diﬀerence gives rise to the generation of Hall currents since the magnetized electrons follow the convective motion of the magnetic ﬁeld from the

134

R. Pottelette and R.A. Treumann

Fig. 5.22. Schematic of the earthward half of the ion diﬀusion region in the magnetospheric tail current sheet during reconnection and its relation to the auroral electron ﬂux regions

tail lobe into the tail current sheet. These Hall currents must close somewhere. However, since such a closure cannot be done locally, the only possibility for them to close is through ﬁeld-aligned currents. Inﬂow of electrons on the boundary far from the current sheet and outﬂow from the boundary in the current sheet are required. This is demonstrated in Fig. 5.22 which shows a schematic of the connection of a part of the ion diﬀusion region in the tail and the auroral electron ﬂuxes which have been measured during one particular path of FAST. In this ﬁgure the electron inﬂows and outﬂows in the ion diﬀusion region in reconnection are seen as large arrows, while the corresponding Hall current is indicated in oppositely directed lines. The electron speed is essentially the inward-convection and outward reconnection-jet speed, respectively, to both

5 Auroral Acceleration and Radiation

135

Fig. 5.23. The schematic connection between tail reconnection and the auroral region. Tail reconnection is shown as a simulation result for the magnetic ﬁeld xpoint region, when one dominant x-point has evolved and several secondary x-points are visible in the tail current sheet. In this case the current sheet is very thin, of the order of the ion skin depth, such that the entire current sheet is aﬀected by reconnection. The Hall coupling of the reconnection region to the auroral zone causes the coupling of the reconnection to the processes in the aurora. It is not yet clear in which way this really happens

sides of the diﬀusion region. The electron outﬂow is related to the downward auroral electron beam ﬂuxes, while the electron inﬂow is fed by or at least related to the ionospheric outﬂow of electrons in the downward current region. In this picture the downward currents are the Hall currents emanating from the near-Earth reconnection site in the tail. Figure 5.23 shows the speculative geometry how activity in the tail connects to the auroral ionosphere. This picture does not take into account fast particles generated during reconnection in the tail since these particles are only a fraction of the entire electron distribution. It also suppresses the real process of coupling between the reconnection site and the ionosphere. This is a process which depends on the real conditions, on the propagation speed of the currents and particles and how fast particles can be extracted from the ionosphere to feed the Hall current in the tail reconnection region. The coupling might be produced by kinetic Alfv´en waves which are generated in the Hall region with transverse scales just of the ion inertial scale. When these waves, which are surface waves, move down to the ionosphere they constitute a current pulse as well as a transverse electric ﬁeld pulse which may cause the required shear ﬂow potential at the topside ionosphere which, as has been discussed here, will drive a U-shaped potential in the ionosphere, causing the double layer and, depending on its polarity, extracting electrons out of the ionosphere. An important problem in this connection is that in all available models the magnetospheric tail reconnection site is so large that it should aﬀect both hemispheres in the same way. Hence aurorae should be symmetrical with

136

R. Pottelette and R.A. Treumann

respect to the equator if a coupling like the one proposed here will exist. This cannot be the case for the auroral processes. The observations of FAST and other spacecraft show that in most cases the closure is nearly local. At ﬁrst glance this suggests that the auroral ﬁeld-aligned current system should be local and should have little in common with a tail current system generated in reconnection. One might therefore argue that the above picture is incorrect as it requires a global closure of the magnetic ﬁeld-aligned currents in the ionosphere similar to those closure processes which have been predicted for decades in the literature [for a review of the various auroral ﬁeld-aligned current system generator mechanisms see Borovsky, 4]. Though this may sometimes be indeed the case, there are possibilities for small scale reconnection in the tail being restricted solely to one hemisphere and even to part of it when reconnection is multiple [as has recently been argued by Pottelette and Treumann, 34] or when, as has been observed recently with Cluster [as can be concluded from observations by Runov et al., 36, 37], the current sheet in the tail bifurcates into narrow current layers. The mechanism of such a bifurcation is not understood yet. It is probably related to the preference of the current layer to generate conditions which are in favor of reconnection to develop on a very fast time scale. For this to happen it is required that the current sheet is very thin. This, however, causes again problems with the topology of the magnetic ﬁeld which becomes very complicated. In principle bifurcation can become possible only when the initially two-dimensional magnetic ﬁeld and current structure develops into a three-dimensional conﬁguration. If this is the case, as most recent numerical simulations of reconnection have shown, then a model of the connection between tail reconnection in the near Earth tail and auroral processes can be developed which is restricted solely to small scale auroral phenomena taking place on one hemisphere only in the auroral region.

5.6 Conclusions We have given a brief overview of the current state of the art of our knowledge in the fundamental acceleration, radiation, and source processes in the nighttime auroral ﬁeld-aligned current system. After a quarter of a century in situ observations from S3-3 to Viking, Freja, Polar, FAST, Geotail and ﬁnally Cluster a stage has been reached on which we can ﬁrmly conclude that our understanding of the auroral processes has stepped up from a purely descriptive one to a semi-quantitative understanding of the dominant acceleration processes of charged particles, both electrons and ions, in aurorae, of the structure of the auroral current system that consists of upward and downward ﬁeld-aligned currents which close in the ionosphere through Pedersen currents on surprisingly small scales, on the radiation mechanism in auroral kilometric radiation, and on the generation of the parallel potential drops. We know by now that indeed such parallel potential drops are generated which cause

5 Auroral Acceleration and Radiation

137

tremendous density decreases in the ionosphere while being highly dynamic. Usually in an auroral acceleration region not only one such drop exists but several are present along the magnetic ﬁeld. Their longitudinal scale is of the order of not more than few 10 km. They thus comprise quasi-stationary but small-scale double layers containing potential drops of order 100 eV to ∼1 keV, in rare cases few keV. These double layers move along the magnetic ﬁeld and can interact with each other. They are the sources of electron holes, small scale structures in the electric ﬁeld and electron phase space distribution of enormous dynamics. These electron holes may themselves contain smaller potential drops which erase the large double layer potential and at the same time cause out of phase variability in the acceleration of electrons and ions. Moreover, these electron holes seem to be the very sources of the auroral kilometric radiation, serving as elementary radiation sources. Thus they are of enormous dynamical importance in the auroral processes and in analogous applications under astrophysical conditions. Nevertheless, a number of open questions still remain which in near future will have to be attacked by multi-spacecraft missions and numerical simulations of the auroral and magnetospheric processes. One of the most interesting of these problems is the relation of the auroral processes and the current generator processes in the geomagnetic tail. These are most probably related to reconnection in the tail current sheet under thinning conditions during substorms. A direct relation has been argued for in this review which is based on the realization that the reconnection Hall current system in the tail current sheet must be closed by ﬁeld aligned currents which connect down to the ionosphere. Many aspects of such a model are in agreement with observation. However, one basic property is still barely understood: this is the fact that aurorae are very obviously local phenomena on one hemisphere only. How this can happen when reconnection is the driving force has still to be clariﬁed. Since reconnection is the most probably generator one may thus ask how reconnection can be imagined being restricted to one hemisphere only. We have suggested ideas on a possible resolution of this puzzle only without going into more detail since models of this kind are still under evolution.

References [1] Bale, S. J., et al.: Astrophys. J. Lett. 575, L25 (2002). [2] Bernstein, I.B., J.M. Greene, and M.D. Kruskal, Phys. Rev. 108, 546 (1957). [3] Berthomier, M., R. Pottelette, and M. Malingre: J. Geophys. Res. 103, 4261 (1998). [4] Borovsky, J.E.: J. Geophys. Res. 98, 6101 (1993). [5] Carlson, C.W., et al.: Geophys. Res. Lett. 25, 2017 (1998). [6] Cattell, C., et al.: Geophys. Res. Lett. 26, 425 (1999). [7] Cattell, C., et al.: J. Geophys. Res. 110, A01211 (2005). [8] Chiu, L. and M. Schulz: J. Geophys. Res. 83, 629 (1978). [9] Delory, G.T., et al.: Geophys. Res. Lett. 25, 2069 (1998).

138 [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42]

R. Pottelette and R.A. Treumann Dupree, T.: Phys. Fluids 26, 2460 (1983). Elphic, R., et al.: Geophys. Res. Lett. 25, 2033 (1998) Ergun, R.E., et al.: Geophys. Res. Lett. 25, 2041 (1998a). Ergun, R.E., et al.: Geophys. Res. Lett. 25, 2025 (1998b). Goldman, M.V., M.M. Oppenheim, D.L. Newman: Geophys. Res. Lett. 13, 1821 (1999). Goldman, M.V., D.L. Newman, and R.E. Ergun: Nonlin. Proc. Geophys. 10, 37 (2003). Gray P., et al.: Geophys. Res. Lett. 17, 1609 (1990). Kindel, J.F. and C.F. Kennel: J. Geophys. Res. 76, 3055 (1971). LaBelle, J. and R.A. Treumann: Space Sci. Rev. 101, 295 (2002). Louarn, P.: this volume (2005). Lysak, R.L. and C.T. Dum: J. Geophys. Res. 88, 365 (1983). Marklund, G., T. Karlsson, and J. Clemmons: J. Geophys. Res. 102, 17509 (1997). Marklund, G., et al.: Nature 414, 724 (2001). McFadden, J.P., et al.: Geophys. Res. Lett. 25, 2045 (1998). McFadden, J.P., C.W. Carlson, and R.E. Ergun: J. Geophys. Res. 104, 14453 (1999). Muschietti, L., et al.: Geophys. Res. Lett. 26, 1093 (1999). Muschietti, L., et al.: Nonlin. Proc. Geophys. 9, 101 (2002). Newman, D.L., et al.: Phys. Rev. Lett. 87, 255001 (2001). Newman, D.L., M.V. Goldman, and R.E. Ergun: Phys. Plasmas 9, 2337 (2002). Newman, D.L., et al.: Comp. Phys. Comm. 164, 122 (2004). Øieroset, M., et al.: Nature 412, 414 (2001). Pickett, J.S., et al.: Ann. Geophys. 22, 2515 (2004). Pottelette, R., R. A. Treumann, and M. Berthomier: J. Geophys. Res. 106, 8465 (2001). Pottelette, R. and R. A. Treumann: Geophys. Res. Lett. 32, L12104, doi:10.1029/2005GL022547 (2005a). Pottelette, R. and R. A. Treumann: Geophys. Res. Lett. 32, submitted (2005b). Pottelette, R., R. A. Treumann, and E. Georgescu: Nonlin. Proc. Geophys. 11, 197 (2004). Runov, A., et al.: Geophys. Res. Lett. 30, 1036, doi: 10.1029/2002GL016136 (2003). Runov, A., et al.: Ann. Geophys. 22, 2535 (2004). Sagdeev, R.Z.: Rev. Mod. Phys. 51, 1 (1979). Singh, N., et al.: Nonlin. Proc. Geophys. 12, in press (2005) St¨ ormer C.: Ergebn. kosm. Physik 1, 1 (1931). Treumann, R. A. and W. Baumjohann: Advanced Space Plasma Physics (Imperical College Press, London, 1997). Wu, C.S. and L.C. Lee: Astrophys. J. 230,621 (1979).

Part II

High-Frequency Waves

6 The Inﬂuence of Plasma Density Irregularities on Whistler-Mode Wave Propagation V.S. Sonwalkar Department of Electrical and Computer Engineering, University of Alaska Fairbanks, Fairbanks, AK, 99775, USA [email protected] Abstract. Whistler mode (W-mode) waves are profoundly aﬀected by FieldAligned Density Irregularities (FAI) present in the magnetosphere. These irregularities, present in all parts of the magnetosphere, occur at scale lengths ranging from a few meters to several hundred kilometers and larger. Given the spatial sizes of FAI and typical wavelength of W-mode waves found in the magnetosphere, it is convenient to classify FAI into three broad categories: large scale FAI, large scale FAI of duct-type, and small scale FAI. We discuss experimental results and their interpretations which provide physical insight into the eﬀects of FAI on whistler (W) mode wave propagation. It appears that FAI, large or small scale, inﬂuence the propagation of every kind of W-mode waves originating on the ground or in space. There are two ways FAI can inﬂuence W-mode propagation. First, they provide W-mode waves accessibility to regions otherwise not reachable. This has made it possible for W-mode waves to probe remote regions of the magnetosphere, rendering them as a powerful remote sensing tool. Second, they modify the wave structure which may have important consequences for radiation belt dynamics via wave-particle interactions. We conclude with a discussion of outstanding questions that must be answered in order to determine the importance of FAI in the propagation of W-mode waves and on the overall dynamics of wave-particle interactions in the magnetosphere.

Key words: Field aligned irregularities, whistler propagation, ducting, plasma inhomogeneity, ionosphere

6.1 Introduction The whistler mode is a cold plasma wave mode with an upper cutoﬀ frequency at the plasma frequency (fpe ) or cyclotron frequency (fce ), whichever is lower. Waves propagating in whistler mode (W-mode) are found in all regions of the Earth’s magnetosphere. They are also found in the magnetospheres of other planets. These waves may originate in sources residing outside the magnetosphere, such as lighting or VLF transmitters, or they may originate within the magnetosphere as a result of resonant wave-particle interactions. W-mode V.S. Sonwalkar: The Inﬂuence of Plasma Density Irregularities on Whistler-Mode Wave Propagation, Lect. Notes Phys. 687, 141–191 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

142

V.S. Sonwalkar

waves have been detected on every spacecraft carrying a plasma wave receiver and at numerous ground stations [see, e.g., 48, 51, 54, 74, 107, 115]. W-mode waves are important because they inﬂuence the behavior of the magnetosphere and partly because they are used as experimental tools to investigate the upper atmosphere. W-mode waves and their interactions with energetic particles have been a subject of interest since the discovery of the radiation belts. These interactions establish high levels of ELF/VLF waves in the magnetosphere and play an important role in the acceleration, heating, transport, and loss of energetic particles in the magnetosphere via cyclotron and Landau resonances [see, e.g., 56, 67, 78, 104]. Propagation, including reﬂection, refraction, guiding, and scattering, determines the extent to which whistler mode energy can remain trapped in the magnetosphere and inﬂuence the nature of wave-particle interactions. Plasma density irregularities in turn determine to a large extent the nature of whistler mode propagation. The role of plasma density irregularities on whistler mode propagation and their eﬀects on the overall dynamics of the magnetosphere is the subject of this paper. The magnetosphere is a highly structured magnetoplasma containing ﬁeld aligned irregularities (FAI) ranging in size from meters to hundreds of kilometers in the cross-B0 direction [30, 39, and references therein] where B0 is the geomagnetic ﬁeld. These irregularities are present in all parts of the magnetosphere. Plasma density and density structures are believed to play an important role in many physical processes at low and high latitudes: wave particle interactions, mode conversion, particle acceleration and precipitation, including auroral precipitation, visible and radar aurora, and substorm activity [e.g., 66, 105, 113]. On the practical side, these irregularities are important because they contribute to the fading of high frequency trans-ionospheric signals and to the degradation of ground-satellite communication. Forecasting and speciﬁcation of these irregularities is a major component of space weather programs. Though the subject of this paper is to understand the role of FAI on whistler mode propagation, the knowledge of these irregularities themselves is incomplete and is a subject of ongoing research [e.g., 30, 41]. Moreover, a wealth of information about FAI has actually been obtained by studying their eﬀects on plasma wave propagation. Thus the research on W-mode waves, and in general on plasma waves, and plasma density irregularities are intimately tied to one another. Evidence that FAI play an important role in the propagation of W-mode waves has been accumulating since the 1950s and 1960s [e.g., 26, 58, 60, 65, 112, and references therein]. In these earlier studies, W-mode propagation in the presence of FAI was considered mainly to explain the occurrence rates of lightning-generated whistlers. In recent years it has been realized that FAI may play a role more important in the overall physics of the magnetosphere than initially envisioned. For example, lighting-generated whistlers refracting through large scale ionospheric horizontal density gradients in the topside ionosphere may lead to generation of plasmapheric hiss, believed to be

6 Density Irregularities and Whistler Propagation

143

primarily responsible for the slot region in the radiation belts [44, 79, 118]. On the other hand, experimental observations of burst electron precipitation by lightning generated whistlers, which propagate through FAI called ducts, indicate that discrete whistlers may play a signiﬁcant role in the loss rates for the radiation belts in the mid-to-low latitudes within the plasmasphere [20, 134]. The general subject of whistler mode wave generation, propagation, and their interaction with energetic particles is a broad subject and is discussed in various books [40, 54] and review papers [48, 51, 74, 115]. Plasma density irregularities also aﬀect waves propagating in free space R-X and L-O modes and in Z-mode, which are discussed elsewhere [e.g., 14, 15, 30, 31, 123]. We focus in this review paper on particular aspects of whistler mode propagation motivated by the following questions: • How is whistler mode propagation aﬀected by FAI of various scale sizes? • What is the signiﬁcance of FAI inﬂuence on W-mode propagation from the magnetosphere to the ground and vice versa? • What are the implications of W-mode wave refraction or scattering by FAI for the generation of new kinds of waves? • What are the implications for mode conversion? • What are the implications for wave-particle interaction? We shall ﬁnd that: • FAI are responsible for providing W-mode waves accessibility to various regions in the magnetosphere otherwise not reachable in a smooth magnetosphere; • FAI guide, reﬂect, and scatter W-mode waves and, in general, modify wave structure, which may have profound eﬀects on wave-particle interactions and the general dynamic equilibrium of the radiation belt particles. It is impossible in this brief review to provide a comprehensive review of the vast amount of literature available on these topics. Our approach is to provide physical insight into various ways FAI may aﬀect W-mode propagation with the help of qualitative explanations and a few illustrative examples. This chapter is organized as follows: Section 6.2 brieﬂy describes magnetospheric plasma and ﬁeld aligned irregularities; Section 6.3 provides a theoretical background on the propagation of whistler mode waves; Section 6.4 presents ground and spacecraft observations of whistler mode waves illustrating the role played by ﬁeld aligned irregularities in the propagation of these waves, and Sect. 6.5 follows with a discussion and concluding remarks.

6.2 Magnetospheric Plasma Distribution: Field Aligned Irregularities The Earth’s magnetosphere is described in many texts and monographs [e.g., 66, 69]. Here we brieﬂy review aspects of magnetospheric cold plasma

144

V.S. Sonwalkar

Fig. 6.1. Schematic showing in the noon-midnight meridian various regions and features of the magnetosphere. Also shown are the locations where whistler-mode waves of various types are observed

distribution important for understanding how various magnetospheric boundaries and density irregularities aﬀect the whistler mode propagation. Figure 6.1 shows schematically various regions of the magnetosphere in the noonmidnight meridian and the locations where whistler mode waves of various types are observed. Immediately surrounding the Earth is the non-conducting atmosphere roughly 60–80 km thick, transparent to the propagation of radio waves. The next layer above the neutral atmosphere is called the ionosphere which extends up to ∼1000 km, completely encircles the Earth, and then merges into the magnetosphere. The boundary between the ionosphere and the neutral atmosphere below it is called the Earth-ionosphere boundary and the region between this boundary and the Earth is called the Earth-ionosphere waveguide. The ionosphere is a conductive medium, in which a small fraction ( fce/2 c)

Fig. 6.3. The refractive index as a function of wave normal direction for (a) f < flhr , (b) flhr < f < fce /2, and (c) f > fce /2. The ray direction is perpendicular to the refractive index surface and in general is in the direction diﬀerent from the wave normal direction as illustrated in (a) and (b)

150

V.S. Sonwalkar

frequency increases in value from below fLHR to a value higher than fLHR . In a closed refractive index surface (f < fLHR ), shown in Fig. 6.3a, the refractive index values remain ﬁnite for all values of the wave normal angle θ. In an open refractive index surface, the refractive index tends to inﬁnity at a certain angle called the resonance angle, θRES ; the corresponding surface of revolution is called resonance cone. For frequencies fLHR < f < fce /2, as shown in Fig. 6.3b, the refractive index surface is concave downward for smaller values of θ and then concave upward for large values of θ. Refractive index surface undergoes another topological change as the frequency increases beyond fce /2 and becomes concave upward for all wave normal angles. These changes in refractive index surface topology have important consequences for wave propagation in the magnetosphere, as discussed in the next subsection. As Fig. 6.3b and Fig. 6.3c show, for f > fLHR , the refractive index n → ∞ as θ → θRES . At wave normal angle close to resonance cone angle, the refractive index becomes large and W-mode waves become short wavelength quasielectrostatic waves and may suﬀer Landau damping. For cold magnetospheric plasma with temperatures fLHR,max , where fLHR,max is the maximum lower hybrid frequency along the ray path, will propagate to the other hemisphere and will undergo total internal reﬂection at the Earth-ionosphere boundary. A ray such as B or B at a low frequency will undergo magnetospheric reﬂection. In general, W-mode waves injected from the ground will remain trapped in the magnetosphere. Also, rays can reach any given location by more than one path after undergoing magnetospheric reﬂections, such as the rays B and B arriving at the satellite as shown in the Fig. 6.4a. Figure 6.4b shows eﬀects of large scale and small scale FAI on the W-mode propagation. A large scale FAI, satisfying condition (1) and thus W.K.B., can bend the upgoing ray such that at a given location rays may arrive by more than one path, each starting at a slightly diﬀerent latitude (rays A and A in Fig. 6.4b). A W-mode wave (ray B in Fig. 6.4b) incident on small scale irregularity can scatter into W-mode waves of both small and large wave normal angles (rays collectively labelled C in Fig. 6.4b). FAI with scale sizes comparable or smaller than W-mode wavelengths can scatter electromagnetic W-mode waves into quasi-electrostatic W-mode propagating with wave normal angles close to resonance cone angle θRES . In general, when condition (2) is satisﬁed, W.K.B. fails and strong scattering of W-mode waves occurs. The excited quasi-electrostatic waves are cut oﬀ at fLHR and are a type of lower hybrid (LH) wave. A number of mechanisms, both linear and nonlinear, have been proposed for producing LH waves from W-mode waves [6, 7, 8, 45, 86, 133]. Figure 6.6 schematically shows the linear mode conversion mechanism ﬁrst proposed by Ngo and Bell [1985]. This mechanism is attractive because it is simple and because it explains most experimental observations better than others do, as shown by Bell and Ngo [7]. Ducted Propagation through the Magnetosphere Figure 6.4c shows W-mode propagation guided by a duct (ray B) and by the plasmapause (ray D). Some of the energy injected from the ground (rays A in Fig. 6.4c) can couple into whistler mode duct or other duct type FAI such as plasmapause and be guided to the other hemisphere with relatively small wave normal angles. At the other hemisphere, part of the energy can propagate down to the Earth and part may be reﬂected back into the magnetosphere, some of it getting coupled into the same duct. Because the refractive index surface goes through a topological change (see Fig. 6.4) at f /fce = 12 , conditions for ducting also change at that frequency [22, 54, 112]. Snell’s construction shows that the density gradients on both sides of a crest tend to rotate the wave normal toward the geomagnetic ﬁeld direction. For frequencies below fce /2, the refractive index surface is concave downward. This geometry requires a density enhancement (crest) for ducting. For frequencies above fce /2, the refractive index surface is concave upward. This geometry requires a density depletion (trough) for ducting. However,

6 Density Irregularities and Whistler Propagation

155

Fig. 6.6. A geometric description of linear mode conversion of whistler mode waves. An incident small wave-normal (θi1 < θR ) whistler mode wave (labelled INC WM in region 1) is partly reﬂected as a small wave normal angle wave (labelled REF WM in region 1) and as a quasi-electrostatic wave with a large wave-normal angle close to resonance cone (labelled REF ES in region 1) and partly transmitted as a small wave-normal whistler mode wave (labelled TRANS WM in region 2), and as a quasi-electrostatic wave with a large wave normal angle close to resonance cone (labelled TRANS ES in region 2). The linear mode conversion is possible because W-mode refractive index has multiple solutions for the parallel component (nzi ) of the refractive index in the fLHR ≤ f < fce range. Thus W-mode waves scattered by this mechanism exhibit a low frequency cutoﬀ at fLHR . The mechanism of mode conversion also works when the incident wave is at large wave normal angle as shown by dotted arrow (θi 1 ∼ θRES ) [adapted from Bell and Ngo, 7] (Reprinted with permission of American Geophysical Union)

since this mode of ducting requires f > fce /2, the wave frequency must be ∼0.5 MHz or higher in order to be ducted all the way down to the ionosphere. Thus this type of ducting does not apply to whistlers. We conclude that whistlers received on the ground require enhancement ducts and that ducting should be eﬀective up to one half of the minimum electron gyrofrequency along the path. Alternatively, a ducted whistler propagating along a certain geomagnetic ﬁeld line will show an upper cutoﬀ at half the gyrofrequency at a point where the ﬁeld line crosses the equator. Both ground and spacecraft observations of whistlers conﬁrm this upper frequency cutoﬀ [Carpenter et al., 28]. Whistler dispersion, time delay as a function of frequency, depends on both the path length and the electron density along the path. Several factors including duct L shell, width, and enhancement/depletion determine whether or not the duct will trap waves incident from below the ionosphere. Helliwell [54] has discussed in detail the trapping of rays in ducts as a function of density enhancement/depletion, the scale size (gradients), and the initial wave normal angle. The coupling of wave energy into and out of a duct may also depend on how low the duct extends downward in altitude [e.g., James, 65]. Using

156

V.S. Sonwalkar

numerical simulations, Bernhardt and Park [16] have estimated that ducts may extend down to 300 km altitude at night but usually terminate above 1800 km during the day. In summer, ducts terminate above 1000-km altitude at all local times. As a result of ducts ending above the topside ionosphere, large scale FAI may play an important role in coupling wave energy from the ground to the duct and vice versa [James, 65]. Depending on the conditions at the exit points of the duct a wave trapped in a duct can undergo reﬂections at the Earth-ionosphere boundary and then propagate back to the other hemisphere in the same duct. Such reﬂections lead to multiple traces of whistler, called echo train, each showing increasing dispersion. After reﬂection, a whistler may not be trapped back in the duct and can propagate subsequently in the non-ducted mode [Rastani et al., 93]. In the case of several ducts, lightning energy from the same lightning discharge may propagate along several ducts and can be detected on the ground station as a multi-component whistler. Ducts and duct-type large scale ﬁeld aligned density drop oﬀs in the magnetosphere can also aﬀect non-ducted propagation [Edgar, 37]. In such cases the ducts are not able to trap the waves, but can signiﬁcantly modify their ray paths such that signals such as MR whistlers show distinctive signatures. Equatorial anomaly and plasmapause are other examples of duct-type irregularities. Equatorial anomaly, in which electron concentration during the daytime is depleted near the equatorial ionosphere and is enhanced in two regions on either side of the equator, also acts like a duct [Sonwalkar, 115]. It can be shown that ducts need not have density gradients on both sides to guide the waves from hemisphere to hemisphere. Thus plasmapause, where the density decreases sharply with increasing radial distance, provides an excellent one-sided whistler duct [61]. Plasmapause can also trap nonducted W-mode waves in certain cyclic trajectories [Thorne et al., 131]. The ducts need not go from one hemisphere to another. They may start and terminate in the same hemisphere. Such ducts have been found at high latitude and have been used to explain the dispersion of impulsive auroral hiss [Siren, 109]. Propagation of W-Mode Waves in the Magnetosphere: Waves Originating in a Source in the Magnetosphere Whistler mode emissions generated near the magnetic equator including midlatitude hiss, discrete, quasi-periodic, periodic emissions, and chorus are also believed to be generated by gyroresonance instabilities, sometimes called whistler mode instabilities [Sonwalkar, 115, and references therein]. The energy sources in most of these cases are the energetic electrons (∼1–100 keV) trapped in the magnetosphere. In the case of chorus, the energy sources are 5–150 keV electrons injected in the dayside region beyond the plasmapause during the period of heightened magnetic activity. The W-mode emissions in the auroral region include auroral hiss which may be produced via Cherenkov mechanism by precipitating beams of electrons with energies greater than

6 Density Irregularities and Whistler Propagation

157

10 keV [LaBelle and Treumann, 74]. Recently, for the ﬁrst time, we have a manmade source of W-mode waves on-board the IMAGE satellite [94, 123]. The Radio Plasma Imager (RPI) on the IMAGE satellite is capable of radiating in W-mode during the low altitude portion of its orbit. The general principles of W-mode propagation for a source in the magnetosphere are similar to those discussed for the case of a ground source. The principal diﬀerence between these two types of sources is that the waves injected from the ground enter the ionosphere vertically, whereas the waves injected into the magnetosphere from space may start with a wide range of initial wave normal angles. Figure 6.7a schematically illustrates the propagation of whistler mode signals from a source location in the equatorial plane within the magnetosphere. Most of the rays, such as A and B, from a magnetospheric source cannot reach a ground receiver because they either undergo total internal reﬂection at the Earth-ionosphere boundary or they undergo LHR reﬂections in Propagation of Plasma Waves - Source of Waves in Space a)

b)

c)

B Large

A

Rx

Earth

C

B 15kHz

Small

Tx

C‘

C‘’ L=5 fLHR=1kHz

ducted propagation

CAH Generation 10 - 100 m Irregularties

Transmission Cone o o (5 - 10 )

PP L=4

Small

10 - 100 m Irregularties

1kHz

1kHz

B Large

IAH Generation

Transmission Cone o o (5 - 10 )

Earth

Earth Diffuse ZM Echo

d)

e)

B

Discrete Non-Ducted Multipath Echo

Earth

B

Diffuse WM Echo

Duct Width [~10 - 100 km]

1 - 10 km Scale Irregularities

Earth

B

f)

Discrete Ducted WM Echo

Earth

10 m - 1 km Scale Irregularity

Fig. 6.7. An illustration of various ray paths for whistler mode wave propagation in the magnetosphere from a source in the magnetosphere: (a) source at the equator, ducted and nonducted propagation; (b) source in the auroral region, nonducted propagation followed by scattering by small scale FAI and propagation to the ground; (c) source in the auroral region, ducted propagation followed by scattering by small scale FAI and propagation to the ground; (d) source on a low altitude satellite, reﬂection from the ionosphere leading to an echo; echoes arriving from multiple paths resulting from propagation through a large scale FAI; (e) source on a low altitude satellite, ducted echoes, (f) source on a low altitude satellite, W-mode and Z-mode diﬀuse echoes resulting from scattering by small scale FAI [adapted from Sonwalkar et al., 123] (Rerpinted with permission of American Geophysical Union)

158

V.S. Sonwalkar

the magnetosphere. However, in the presence of ducts, some of the rays, such as C, can be guided to the low altitude ionosphere with their wave normal at small angles with respect to the local vertical; these rays can be observed at a ground receiver after propagating within the Earth-ionosphere wave guide [Helliwell, 54]. The ducted signals that have backscattered from the ionosphere-waveguide boundary can propagate back into the duct or can propagate as nonducted signals [Rastani et al. 93]. In the topside ionosphere, FAI may refract or scatter the downcoming W-waves, which may play a crucial role in determining subsequent paths of these waves. The waves can also reach the ground directly from the source if the source is at low altitudes and radiates such that some of the wave energy reaches the Earth-ionosphere boundary with suﬃciently small wave normal angles with respect to the local vertical [Thompson and Dowden, 132]. Other modes of guiding energy from a magnetospheric source to the ground include propagation along the plasmapause [61] and a sub-protonospheric mode [92, 110]. Figures 6.7b and 6.7c show propagation from a source located on auroral ﬁeld lines and generating waves at large wave normal angles. It is believed that auroral hiss (AH) is generated in this manner. As shown in the ﬁgure, these W-mode waves can propagate down in ducted or nonducted mode and, in general, will be reﬂected back by magnetospheric or total internal reﬂection. However, if there are FAI present in the regions where these reﬂections occur, some of the large wave normal angle W-mode waves may be refracted or scattered into small wave normal angle waves which then can be observed on the ground. Figures 6.7d-f show various ways W-mode waves can propagate from a low altitude source, assumed here to be a transmitter on a satellite, towards the Earth-ionosphere boundary and be seen on the satellite as an echo or be seen on the ground. An echo can reach the satellite by single or multiple non-ducted paths (Fig. 6.7d), or propagate down to the Earth-ionosphere boundary in a duct and return back in the same duct (Fig. 6.7e), or be scattered by small scale FAI (Fig. 6.7f). The scattered waves generally show a range of time delay giving echo a diﬀuse appearance on a spectrogram.

6.4 Observations and Interpretations The literature describes numerous observations of W-mode waves, from ground stations and from low and high altitude spacecraft [e.g., 1, 48, 51, 54, 56, 74, 115]. We present here a few examples to illustrate the key features of W-mode propagation in the presence of ﬁeld-aligned irregularities. W-mode observations are frequently categorized as those found on the ground and those found on spacecraft. The reason for dividing ground and spacecraft observations into separate categories is that somewhat distinct and apparently uncorrelated activity is detected at ground stations and on satellite [Sonwalkar, 115]. However, in this paper, for the reasons discussed in the

6 Density Irregularities and Whistler Propagation

159

previous section, we classify observations of W-mode waves of magnetospheric origin into two main categories: • the observations of W-mode waves with sources on the ground, • the observations of W-mode waves with sources in the magnetosphere. 6.4.1 W-Mode Observations When the Source is Below the Ionosphere It is convenient to categorize W-mode observations into two subsections: (1) non-ducted propagation, (2) ducted propagation. Ducted propagation can be seen both on the spacecraft and on the ground, whereas non-ducted propagation can only be seen on the spacecraft. Historically ducted propagation was discovered ﬁrst (on the ground) and non-ducted later. It is, however, easier to start with observations of non-ducted signals as non-ducted propagation is the natural mode of propagation of W-mode waves in a smooth magnetosphere. Nonducted Propagation: Eﬀects of Large-scale FAI, Ducts, and Density Drop-oﬀs We begin with the simplest possible propagation scenario: propagation of a single frequency signal reaching a satellite as a plane wave in a smooth magnetosphere as illustrated by ray B in Fig. 6.4a. Figure 6.8a shows an example of near plane wave received on the DE 1 satellite. A 20-s-long, 4.0 kHz continuous wave Siple transmitter signal was received by the 200-m long electric antenna on the DE 1 satellite. The satellite was located outside the plasmasphere at L ∼ 4.5 and λm ∼ 7◦ N. The local electron density was about 15 el/cc corresponding to fpe ≈ 35 kHz and the local magnetic ﬁeld was about 0.0034 G corresponding to fce ≈ 9.5 kHz. The electric ﬁeld amplitude shows well deﬁned spin fading at half the spin period (6 s), indicating reception of a single plane wave. Reception of plane W-mode waves is relatively rare, but when detected, it can provide testing of some of the most fundamental ideas of W-mode propagation and antenna properties in space plasmas [114, 116]. The depth and the phase of spin fading pattern can be used to obtain the wave normal direction of about 55◦ with respect to geomagnetic ﬁeld. Ground transmitter signals received on the satellite often show fading patterns which cannot be explained as a result of satellite spin motion alone. In fact fading patterns have been observed on satellites that were not spin stabilized. Heyborne [59] and Scarabucci [102] observed amplitude fading at low altitude (∼1000 km) OGO 1, OGO 2, and OGO 4 satellites. Cerisier [32]

160

V.S. Sonwalkar

Fig. 6.8. Nonducted whistler mode propagation of Siple transmitter signals: (a) An example of near plane wave received on the DE 1 satellite. A 20-s-long, 4.0 kHz continuous wave Siple transmitter signal was received by the 200-m long electric antenna on the DE 1 satellite. (b) An example of two plane waves received on the ISEE 1 satellite. A 20-s-long, 4.0 kHz continuous wave Siple transmitter signal was received by the 215-m long electric antenna on the ISEE 1. (c) Fourier transform of the wave amplitude shown in (b) [adapted from Sonwalkar et al. 114, 120] (Reprinted with permission of American Geophysical Union)

observed two Doppler shifts in the FUB signal observed on the FR 1 (altitude ∼750 km). Neubert et al. [85] and Sonwalkar et al. [120] have observed pulse elongations and amplitude modulations in Omega and Siple transmitter pulses received on the GEOS 1 and ISEE 1 satellites, respectively. These

6 Density Irregularities and Whistler Propagation

161

authors have interpreted their observations as indicating the presence of two or more plane waves arriving from diﬀerent directions. Figure 6.8b shows an example of a 20-s-long, 4.0 kHz continuous wave Siple transmitter signal received by the 215-m long electric antenna on the ISEE 1 satellite. The fading pattern now contains frequencies other than twice the spin frequency of the ISEE 1 satellite (spin period = 3 s). This fading pattern can be explained as that arising from the satellite receiving two plane waves arriving from two diﬀerent directions. The Doppler frequency for each plane wave is diﬀerent, and thus the satellite receiver measures the beat pattern resulting from this diﬀerence. It can be easily shown that the number of spectral components in the mean square received voltage is given by (3N 2 −3N +4)/2, where N is the number of multiple paths [Sonwalkar and Inan, 114]. Figure 6.8c shows the amplitude of the Fourier transform of the electric ﬁeld amplitude envelope shown in Fig. 6.8b. We clearly notice peaks at ﬁve frequencies, which are identiﬁed as the d.c., diﬀerential doppler shift (∆ωd ), twice the spin frequency (ωs ), and the sum and diﬀerence of diﬀerential Doppler shift with twice the spin frequency. A detailed analysis of Siple transmitter signals received on the ISEE 1 satellite showed that, in general, at any point in the magnetosphere the direct signals (before reﬂections) transmitted from the ground arrive, within few hundred milliseconds of each other, along two or more closely spaced multiple paths, as illustrated by rays A and A in Fig. 6.4b . The multiple paths can be explained by assuming propagation through 1–10 km cross-B scale size FAI in the topside ionosphere with a few percent enhancement or depletion in the plasma density. Such FAI can refract the wave normal direction by ∼5–10◦ , giving rise to multiple path propagation as illustrated in Fig. 6.4b [Sonwalkar et al., 120]. The diﬀerential Doppler shift ∆f resulting from the observation of multipath W-mode ground signals on a satellite is of the order of ∆nVs f /c, where ∆n is the diﬀerence in the refractive index values of two closely spaced multiple paths, Vs is the satellite speed and c the velocity of light in vacuum. Assuming ∆n ∼ 1, f ∼ 4 kHz, Vs ∼ 8 km, we obtain ∆f ∼ 0.1 Hz, a fraction of Hz. Clearly to resolve such multiple paths we need either long duration ﬁxed frequency signals, such as the one discussed above, or signals originating in impulsive (duration ∼ ms) lightning discharges which can be resolved in time domain when they arrive at the satellite by more than one path, or we need measurements of multiple spacecraft separated by distances of the order of 100 km (e.g. Cluster). It is possible to misinterpret or have more than one interpretation of wave normal direction measurements of a short duration (∼1 s) ground transmitter signals received on a satellite. For example, Lefeuvre et al. [76] determined wave normal directions of seven Omega transmitter pulses observed on the GEOS 1 satellite. They found a 0.2- to 0.4-s periodicity in the observed wave normal directions, which Sonwalkar et al. [120] interpreted as an indication of two closely spaced multiple paths. Whistlers propagating in the non-ducted mode also show evidence of propagation along single direct paths in a smooth magnetosphere and that of

162

V.S. Sonwalkar

Fig. 6.9. Non-ducted whistler mode propagation of whistlers : (a) Spectrogram of a multi-component MR whistler received on OGO 1 satellite via in a smooth magnetosphere, (b) propagation in the presence of FAI [adapted from Smith and Angerami, 111] (Reprinted with permission of American Geophysical Union)

propagation along multiple paths in a magnetosphere containing large scale FAI. Figure 6.9a shows the spectrogram of a magnetospherically reﬂected or MR whistler received on OGO 1 satellite [Smith and Angerami, 111]. The MR whistler usually consists of a series of traces or components, each exhibiting a frequency of minimum travel time or nose frequency. The multi-component feature of MR whistlers can be explained in terms of propagation of lightning energy in relatively smooth magnetosphere by non-ducted paths which have undergone multiple MR reﬂections as illustrated by rays B and B in Fig. 6.4a [Kimura, 68]. MR whistlers, also called unducted or non-ducted whistlers, show a wide variety of frequency-time signatures, depending on the location of the receiver (spacecraft) with respect to the causative lightning discharge,

6 Density Irregularities and Whistler Propagation

163

and on the density distribution of electrons and ions in the magnetosphere. By the nature of their propagation mode, these unducted whistlers cannot be detected on ground. The spacecraft observations of unducted whistlers have led to a new understanding of the many subtle features of whistler-mode propagation and to the deduction of important plasma parameters in space [17, 35, 36, 37, 38, 111, 118, 129]. Edgar [37] has shown that ray tracing simulations in a smooth magnetosphere can explain the observed spectrogram shown in Fig. 6.9a. In general, MR spectrogram contain features that cannot be explained by propagation in a smooth magnetosphere. For example Fig. 6.9b shows the spectrogram of MR whistler observed on 18 March 1965 on OGO 1 satellite. This whistler has a spectral signature quite diﬀerent than that of the MR whistler shown in Fig. 6.9a. The traces O+ , 2− , and 2+ are expected for an MR whistler propagating in a smooth magnetosphere for the given location of the satellite and the hemisphere of the lightning. Smith and Angerami [111] speculated that extra traces, labelled 2∗− and 2∗+ , probably resulted from multi-path propagation, possibly due to large-scale irregularities in the magnetosphere. Using ray tracing simulations, Edgar [37] showed that the extra traces 2∗− and 2∗+ could be explained by assuming multi-path propagation resulting from a 30% density drop oﬀ at L ∼ 1.8. He also demonstrated that the distortion in the 2− , and 2+ components near 6 kHz can be explained by considering a duct with ∼ 20% enhancement in density near L = 2.4. In general Edgar [37, 38] and more recently Bortnik et al. [17] have shown that large and small scale FAI in the topside ionosphere and density drop oﬀs and ducts in the magnetosphere can explain speciﬁc observed features in MR whistler spectrograms. Some features such as the frequency cutoﬀ of MR whistlers may be explained in terms of their trapping in density gradients, Landau damping, and D-region absorption [17, 38, 131]. It is evident that the same mechanism, viz. bending of rays due to the presence of large scale FAI has been used to explain multi-path propagation of the ground signals (Fig. 6.8b) as well as multi-path propagation of MR whistlers (Fig. 6.9b). This bending permits more than one ray paths of similar type, such as two direct ray paths or two or more MR paths of similar type (e.g. 2− and 2∗− ), to reach the satellite for a given location in the magnetosphere. It is clear that this type of multi-path propagation (Rays A and A in Fig. 6.4b), resulting from the presence of large scale FAI in the magnetosphere or topside ionosphere, is diﬀerent from the one resulting from the multiple magnetospheric reﬂections in the smooth magnetosphere (Rays B and B in Fig. 6.4a). Non-Ducted Propagation: Eﬀects of Small-Scale FAI Small scale irregularities satisfying condition (2) can scatter W-mode waves into directions which may be drastically diﬀerent than their initial wave normal directions. For W-mode, with typical wavelengths of the order of a few km,

164

V.S. Sonwalkar

the condition (2) can be easily satisﬁed by small scale FAI with LF AI ∼10– 100 m and ∆Ne /Ne ∼ a few percent. The scattered waves are generally Wmode waves with both small and large wave normal angles. The large wave normal angle waves are quasi-electrostatic and show a lower cutoﬀ at fLHR and are therefore generally call lower hybrid (LH) waves. Earliest evidence, albeit indirect, of inﬂuence of small scale FAI on Wmode propagation came from a strong correlation of whistler activity at midlatitude with HF backscatter [Carpenter and Colin, 26] and the correlation of auroral hiss activity with HF backscatter [Hower and Gluth, 60]. HF waves are scattered by FAI with scale lengths comparable to HF-wavelength, a few tens of meters. A dramatic evidence of scattering of W-mode waves by small scale FAI came when the “spectral broadening phenomenon” was discovered for ground transmitter signals received on the low altitude satellites [Bell et al., 11]. Figure 6.10 shows an example of Omega transmitter signals received on the DE 1 satellite. The pulses of ∼ 1 s duration from the Omega transmitter in North Dakota are present at 10.2 kHz, 11.05 kHz, and 11.333 kHz. Spectrally broadened Omega transmitter pulses are seen both in Fig. 6.10a, showing a spectrogram of the electric ﬁeld measured by a 200 m long dipole antenna, and in Fig. 6.10b, showing spectrogram of the electric ﬁeld measured by a short 8 m long dipole antenna. The spectrally broadened pulses are direct pulses which have propagated from the ground along a ∼2000 km path directly to the spacecraft. Following the direct pulses are echoes of the direct pulses reﬂected from the conjugate hemisphere with a delay of ∼2 s. In both panels the apparent bandwidth of the direct pulses is larger than 100 Hz, ∼100 times the 1 Hz bandwidth of the transmitted pulses.

Fig. 6.10. Observations of lower hybrid waves excited by signals from the Omega (ND) transmitter. (a) and (b) respectively show reception by the Ex and Ez antennas. The Ex antenna consists of two 100 m wires deployed in the spin plane. The Ez antenna consists of two 4 m tubes deployed along spin axis [adapted from Bell et al., 10] (Reprinted with permission of American Geophysical Union)

6 Density Irregularities and Whistler Propagation

165

The LH waves can also be excited by non-ducted whistlers. Figure 11a shows an example of the excitation of LH waves above local fLHR by a series of non-ducted whistlers observed on the ISIS-2 satellite. Note that LH waves are excited for f ≥ fLHR , consistent with generation by linear mode conversion method (Fig. 6.6). Observations from many satellites and theoretical analysis show that the excited waves are short wavelength (5 m < λ < 100 m) electrostatic lower hybrid (LH) waves excited by electromagnetic whistler mode waves through scattering from magnetic ﬁeld-aligned irregularities located in the topside ionosphere and magnetosphere [e.g., Bell et al., 12, and references therein]. These LH waves exhibit very large Doppler shifts which cause the observed spectral broadening (∼300–1000 Hz) of the signal as received at the moving satellite [7, 10, 11, 62]. Earlier observations of this phenomenon occurred at altitudes ≤7000 km [Bell and Ngo, 6]. Recently, Bell et al. [12] report new observations from the Cluster spacecraft of strong excitation of lower hybrid (LH) waves by electromagnetic (EM) whistler mode waves at altitudes ≈20,000 km outside the plasmasphere. These observations provide strong evidence that EM whistler mode waves are continuously transformed into LH waves as the whistler mode waves propagate at high altitudes beyond L ∼ 4. This process may represent the major propagation loss for EM whistler mode waves in these regions, and may explain the lack of lightning generated whistlers observed outside the plasmapause [Platino et al., 90]. The excited LH waves represent a plasma wave population which can resonate with energetic ring current protons to produce pitch angle scattering on magnetic shells beyond L∼4. Thus linear mode coupling provides a new mechanism by which lightning generated whistler mode waves can aﬀect the lifetimes of energetic ring current protons. We discuss another interesting manifestation of the eﬀects of FAI, perhaps both small and large scale, on the propagation of whistlers. Figure 6.11b shows an example of whistler-triggered hiss emissions. Using data from the DE 1 satellite, Sonwalkar and Inan [118] showed that lightning generated whistlers

Fig. 6.11. Examples of eﬀects of large and small scale FAI on non-ducted whistler propagation. (a) ISIS-2 satellite observations of lower hybrid waves near 10 kHz excited by whistlers. (b) DE 1 satellite observations of hiss band excited by whistler (courtesy of VLF group, Stanford University)

166

V.S. Sonwalkar

often trigger hiss emissions that endure up to 10- to 20-s periods and suggested that the lightning may be an embryonic source of plasmaspheric hiss. They recognized that direct multiple path propagation resulting from FAI irregularities may play an important role in the generation of whistler triggered hiss. Subsequently using ray tracing simulations Draganov et al. [35, 36] showed that the combined contribution from whistler rays produced by a single lightning ﬂash entering the magnetosphere at a range of latitudes and with a range of wave normal angles can form a continuous hiss-like signal. It was assumed that at 400 km altitude the initial wave normal angles of the rays are spread around the nominal vertical orientation. The spread in the wave normal angle was presumed to be generated as a result of W-mode propagation through large and small scale FAI in the topside ionosphere. It may be noted that H. C. Koons ﬁrst suggested the idea in 1984 that plasmaspheric hiss might simply be the accumulated waves of many non-ducted whistlers [See p. 19,486 of 126]. Recently, Green et al. [44] found that the longitudinal distribution of the hiss intensity (excluding the enhancement at the equator) is similar to the distribution of lightning: stronger over continents than over the ocean, stronger in the summer than in the winter, and stronger on the dayside than on the nightside. These observations strongly support lightning as the dominant source for plasmaspheric hiss, which, through particle-wave interactions, maintains the slot region in the radiation belts. Ducted Propagation: Observations on the Ground The importance and role of ducts in the propagation of W-mode waves cannot be overemphasized. As discussed in Sect. 6.3, without ducts W-mode waves would not be observable on the ground. The earliest observations of W-mode waves were ducted whistlers observed on the ground. Helliwell [54] in his now classic monograph has given an early history of W-mode wave observations. He gives examples of many types of W-mode waves including lightning-generated whistlers, ground transmitter signals, and VLF emissions of magnetospheric origins. Helliwell et al. [58] were the ﬁrst to suggest that whistlers might become trapped within ﬁeld-aligned density irregularities (ducts) and propagate from one hemisphere to another. The duct hypothesis was prompted by the observation that most whistlers observed on the ground show several discrete cases with varying frequency-time dispersion indicative of propagation over several distinct paths. Figure 6.12 shows examples of ducted propagation of Siple transmitter and lightning-generated whistlers, observed on the ground at Lake Mistissini, Canada and Palmer Station, Antarctica. Figure 6.12a describes the essential idea behind the active experiments performed from Siple Station, Antarctica, during the 1973–1998 period. These experiments provided new results and insight on the eﬀects of cold and hot plasma on W-mode propagation [25, 56, and references therein]. Figure 6.12b shows a 1-s pulse from Siple Station received at Lake Mistassini after propagation along two ducts in which wave

6 Density Irregularities and Whistler Propagation

167

Fig. 6.12. Examples of ducted W-mode propagation observed on the ground: (a) Schematic showing basics of ﬁeld-aligned whistler mode wave injection experiments between Siple Station, Antarctica (SI), and Lake Mistissini, Quebec, Canada (LM). (b) A 1-s pulse (lower right) from SI is received at LM (upper right) after propagation along two ducts in which wave growth and triggering of rising emissions occurs. A VLF receiver located at Palmer (PA) provides data on low L shell whistler mode paths and on subionospheric propagation from Siple Station. (c) Spectrogram of a multi-component ducted whistler received at Palmer Station, Antarctica, and the associated triggered emissions [adapted from Helliwell, 56] (Reprinted with permission of American Geophysical Union)

growth and triggering of rising emissions occur. Emission triggering is very common for ducted signals but rarely observed for non-ducted signals. Figure 6.12c shows an example of multi-component whistler and triggered emissions observed at Palmer station, Antarctica. The spheric originating from the same lightning discharge that caused the whistler was also detected and is marked by an arrow. Spatial and temporal occurrence patterns of ducted and non-ducted whistlers have been a subject of great interest [Helliwell, 54]. Many aspects of the occurrence rates, such as the local time, season, and locations could be related to the occurrence rates of causative lightning, and propagation conditions such as high absorption in the daytime ionosphere and the latitudinal dependance of whistler mode transmission cone. The occurrence of lightning activity reduces at high latitude and the transmission cone is very small at low latitudes. Consequently the whistler activity peaks at nighttime at mid-latitude (∼50◦ ). Many studies show that large and small scale FAI in the topside ionosphere and plasmapause regions play an important role in the propagation and

168

V.S. Sonwalkar

occurrence of ducted whistlers [see, e.g., 26, 27, 65, and references therein]. Carpenter and Colin [26] found that day-to-day variations of W-mode waves at mid-latitude are strongly correlated with the occurrence of small scale FAI in the F-region as observed by HF backscatter. They interpreted these F-region FAI as being the feet of the large scale FAI – whistler ducts. Using wave normal direction data from FR-1 satellite (altitude = 750 km) and ray tracing, James [65], showed that in addition to factors considered above, large scale FAI in the topside ionosphere inﬂuence the rate of occurrence of ˘ c [27] showed that ionospheric processes such as whistlers. Carpenter and Suli´ wave damping, defocusing within mid-latitude trough, focusing within large scale (50–100 km) FAI, and scattering by 10 to 100-m FAI can explain many features of the observed spatial distributions of whistler paths beyond the plasmapause and also the relatively low amplitude of the associated whistlers as compared to that of whistlers found inside the plasmasphere. They also conﬁrmed that ducted whistler propagation outside the plasmasphere occurs at comparatively low rates in comparison to activity within, and tend to occur at certain distances from the plasmapause. W-mode propagation can also be guided by ﬁeld aligned density irregularities other than ducts, such as the plasmapause. Carpenter [23] gives examples of whistler and VLF noises propagating just outside the plasmasphere. These whistlers and emissions exhibited features not observed in usual ducted whistlers, including extension of signal frequencies into the range 0.5−0.8fHeq , where fHeq is the equatorial gyro-frequency of the path, and the echoing or repeated propagation over the path at frequencies above 0.5fHeq . The VLF noise bands and bursts tended to occur within the frequency range 0.4−0.8fHeq . This kind of propagation is not well understood. Using ray tracing, Inan and Bell [61] have shown that rays can be guided along the plasmapause outer surface, but only for the waves leaving with wave normal angles close to relatively large Gendrin angle. Inan and Bell [61] suggest that large scale FAI with strong density gradients can tilt the vertical wave normal angles into Gendrin angle. Cairo and Cerisier [21] have shown that signiﬁcant departures of upgoing wave normals from the vertical are observed within the plasmasphere, apparently due to large scale FAI. There are many aspects of ducted whistler mode propagation that are not well understood. For example, analysis of ground whistler data reveals that a single whistler can contain several hyperﬁne structures [Hamar et al., 49]. These ﬁner structures in whistlers can be interpreted either in terms of electron density ﬂuctuations of the order of ∼ 1% and spatial scale sizes of the order of 50 km, or in terms of excitation of multiple propagating modes within a duct [Hamar et al., 50]. It is possible that the ﬁner details of wave particle interactions that lead to a variety of triggered emissions are related to the details of the ducted propagation.

6 Density Irregularities and Whistler Propagation

169

Ducted Propagation: Observations on the Spacecraft Ducted W-mode waves are rarely observed on satellites in the magnetosphere due to the apparently small volume (0.01%) occupied by the ducts [Burgess and Inan, 20], but are most commonly observed on the ground after they have exited from a duct and propagated in the Earth-ionosphere wave guide to a receiving site [Helliwell, 54]. While the duct hypothesis has been extensively used in interpreting ground observations of VLF waves, only a limited set of experimental data are available that provides for the determination of the properties of magnetospheric ducts based on in situ observations [2, 24, 71, 90, 103, 111, 122]. Detection of ducts using spacecraft observations was based in most cases on measurements of whistler dispersion [2, 24, 90, 103, 111]. In one case it was based on enhancements of whistler mode hiss correlated with density enhancements [Koons, 71]. In one case discussed below, wave normal direction measurement was used to establish ducted propagation and electron density measurement was used to determine duct parameters. These authors provided the following measurements of ducts: ∼ 400 km near L = 3 by Smith and Angerami [111], 223–430 km and 6–22% for L = 4.1 − 4.7 by Angerami [2], 68–850 km and 10–40 % for L = 3.1 − 3.5 by Scarf and Chappell [103], 630– 1260 km and ≤30% between L = 4 and L = 5 by Carpenter et al. [29], and ∼500 km and ∼40% by Koons [71]. Observations listed above, excepting those by Platino et al. [90], were all made in the plasmasphere. Recently, using data from the Cluster spacecraft, Platino et al. [90] have shown that whistlers observed outside the plasmasphere in the low density regions occur only in the presence of large scale irregularities within which the waves are “ducted”. They found that dispersion characteristics of observed whistlers are only matched by ray tracing simulations if the whistlers are ducted. They also propose a possible explanation why whistlers outside the plasmasphere are rarely observed, based on wave conversion from whistler mode to lower hybrid mode [Bell et al., 12]. A direct conﬁrmation of many features of both ducted and non-ducted propagation was obtained in a study involving measurement of the VLF transmitter (15 kHz, 4s ON/4s OFF, 800 kW radiated power) at Khabarovsk, Russia (48◦ N, 135◦ E; λm ∼ 38◦ , φm ∼ −158◦ ) received on the low altitude COSMOS 1809 satellite Sonwalkar et al. [Sonwalkar et al., 121]. Figure 6.13 shows an example of signals from the VLF transmitter at Khabarovsk, Russia (48◦ N, 135◦ E; λm ∼ 38◦ , φm ∼ −158◦ ) received on the low altitude COSMOS 1809 satellite (near-circular orbit at an altitude of ∼960 km and inclination of ∼82.5◦ ). Figure 6.13a top and bottom panel display the DE 1 and COSMOS 1809 satellite orbits in the magnetic meridional and magnetic equatorial planes, respectively, when Khabarovsk signals were observed on the two satellites on 23 Aug 89. Simultaneous observations of ground transmitter signals and electron density on COSMOS 1809 provided unique opportunity to

170

V.S. Sonwalkar (b)

(a)

Fig. 6.13. (a) Orbits of DE 1 and COSMOS 1809 satellites in magnetic meridional and equatorial planes on 23 Aug 1989 during the period when the Khabarovsk transmitter was operating. The thickened parts of the orbits indicate the time period during which the transmitter signals were detected on the satellites. (b) Amplitude of electric (top panel) and magnetic ﬁelds (second panel) of Khabarovsk signals on COSMOS 1809 satellites for 23 Aug 89. Also shown are the mean square root variation in the electron density dNe (third panel) and electron density Ne (fourth panel) [adapted from Sonwalkar et al., 122] (Reprinted with permission of American Geophysical Union)

measure the parameters of density irregularities responsible for direct multiple path propagation and for ducted propagation. Figure 6.13b shows COSMOS 1809 observations of Khabarovsk signals in the 15 kHz-channel of the spectrum analyzer when the satellite passed over the transmitter at ∼1600 UT and when the satellite passed over a region roughly magnetically conjugate to the transmitter location in the southern hemisphere at ∼1622 UT. The ﬁgure presents electric ﬁeld Ey , magnetic ﬁeld Bx , mean square root variation in the electron density dNe , and the total electron density Ne along the satellite orbit. Since the satellite altitude during

6 Density Irregularities and Whistler Propagation

171

this interval was approximately constant, dNe is a measure of horizontal gradients in the ionospheric electron density. The top two panels of Fig. 6.13b show the amplitude fading of Khabarovsk transmitter signals due to multiple path propagation eﬀects and the third panel shows the ﬂuctuations ∆Ne in the electron density believed to be responsible for the multiple path propagation. The typical horizontal extent of these irregularities was ≤100 km and the density ﬂuctuations were of the order of ≤5%, consistent with previous theoretical estimates obtained using wave observations and ray tracing simulations [Sonwalkar et al., 120]. When COSMOS 1809 was in the southern hemisphere, the magnetic ﬁeld of the transmitter signal showed a large value near 1625 UT. The value of the ratio (cBx )/Ey gives an indication of the wave normal angle of the wave. Based on ray tracing simulations, for non-ducted wave propagation of ground transmitter signals injected in the northern hemisphere, a large wave normal angle corresponding to a lower value of the ratio (cBx )/Ey is to be expected in the southern hemisphere. From Fig. 6.14b second and third panel, we ﬁnd this to be the case in general except for a signal observed between 1624:57 and 1625:14 UT when rather large values of Bx and (cBx )/Ey were measured indicating low wave normal angles (< 30◦ ). Sonwalkar et al. [122] interpreted this low wave normal angle signal to be a case of ducted propagation. To further test this assumption, the electron density data were examined. It was found that between 1625:04 and 1625:12 there was a 3000 to 3800 el/cc enhancement over the background electron density of 29500 el/cc, consistent with the duct hypothesis (other variations in electron density seen in Fig. 6.14b third panel are presumed to be local ﬁeld aligned irregularities which are commonly present in the ionosphere). Further, the equatorial gyrofrequency corresponding to L =2.85, the duct location, is 34.4 kHz. Thus the 15 kHz Khabarovsk signal is well below the half gyrofrequency cutoﬀ of fHeq /2 = 17.2 kHz for ducted propagation [54]. The L shell thickness of the duct was 0.06 and corresponds to a duct cross section of 55 km at 981 km altitude and 367 km at the geomagnetic equator. The density enhancement was 10-13% over the background electron density of 29500 el/cc. Only a lower limit of ∆φ ≥ 0.2◦ can be placed on the longitudinal extent of the duct, consistent with the 3-4◦ of width in longitude estimated by Angerami [2]. The eﬃciency of signal coupling between the Earth-ionosphere waveguide and magnetospheric ducts is of importance to studies of VLF propagation and wave ampliﬁcation in the magnetosphere. This coupling is critically dependent on the duct end points in the ionosphere [16]. The above study indicates that duct end points can extend down to at least ∼980 km at ∼0130 local time in the southern hemisphere during austral winter. This result is consistent with theoretical studies of Bernhardt and Park [16] who predict a duct endpoint as low as 300 km at night during winter and equinoxes. The ducted signals were observed over an L shell range of 0.13, about twice the L shell width of the duct. This could be the result of ducted signals leaking from the duct as previously noted in both experimental and theoretical works

172

V.S. Sonwalkar

Fig. 6.14. (a) DE 2 observations of impulsive signals near the geomagnetic equator in the 512- to 1024-Hz channel of the VEFI instrument. (b) Langmuir probe observations of electron density illustrating equatorial (Appleton) anomaly. (c) Ray tracings through a horizontally stratiﬁed ionosphere. Trajectories of rays (750 Hz frequency) injected at 200-km altitudes with vertical wave normal angles. (d) Ray tracings in the presence of an equatorial anomaly. Trajectories of rays (750 Hz frequency) injected at 200-km altitudes with vertical wave normal angles [adapted from Sonwalkar et al., 121] (Reprinted with permission of American Geophysical Union)

[2, 127, 128]. The peak electric and magnetic ﬁeld were detected inside the duct near the duct center. Both the peak electric (>520 µV/m) and magnetic ﬁeld (36 pT) intensities of the ducted signals were comparable to those observed for the non-ducted signals over the transmitter in the northern hemisphere, but were about 20 dB higher than those of the non-ducted signals observed in the southern hemisphere in the vicinity of the duct, consistent with observations of Koons [71]. Inside the duct electric and magnetic ﬁeld intensities show a ﬁne structure, consistent with recent reports of whistler ﬁne structure by Hamar et al. [49, 50]]. Ducted whistlers are rarely observed below L = 2. An unusual observation of lightning energy reaching the low altitude equatorial regions was made by Sonwalkar et al. [Sonwalkar et al., 121]. They show that equatorial anomaly can focus W-mode waves from lightning into a region near the equator. The equatorial anomaly is a tropical ionospheric eﬀect arising from equatorial electrodynamics, which essentially leads to the formation of a ductlike structure with enhanced density along a ﬁeld line near ±20◦ geomagnetic latitude [66]. The Vector Electric Field Instrument (VEFI) on the Dynamic Explorer 2 (DE 2) satellite observed impulsive ELF/VLF electric ﬁeld bursts on almost every crossing of the geomagnetic equator in the evening hours (Fig. 6.14a). These signals were interpreted as originating in lightning discharges. These signals that peak in intensity near the magnetic equator were observed within

6 Density Irregularities and Whistler Propagation

173

5-20◦ latitude of the geomagnetic equator at altitudes of 300–500 km with amplitudes of the order of ∼ millivolts/m in the 512-1024 Hz frequency band of the VEFI instrument. The signals are observed in the same region near the equator where the equatorial anomaly, as detected by Langmuir probe, was found (Fig. 6.14b). Whistler-mode ELF/VLF wave propagation through a horizontally stratiﬁed ionosphere predicts strong attenuation of sub-ionospheric signals reaching the equator at low altitudes. However, ray tracing analysis showed (Figs. 6.14c and 6.14d) that the presence of the equatorial density anomaly, commonly observed in the upper ionosphere during evening hours, leads to the focusing of the wave energy from lightning near the geomagnetic equator at low altitudes, thus accounting for all observed aspects of the observed phenomenon. The observations presented here indicate that during certain hours in the evening, almost all the energy input from lightning discharges entering the ionosphere at < 30◦ latitude remains conﬁned to a small region (in altitude and latitude) near the geomagnetic equator. The net wideband electric ﬁeld, extrapolated from the observed electric ﬁeld values in the 512–1024 Hz band, can be ∼10 mV/m or higher. These strong electric ﬁelds generated in the ionosphere by lightning at local evening times may be important for the equatorial electrodynamics of the ionosphere. 6.4.2 W-Mode Observations When the Source is in the Magnetosphere As shown in Fig. 6.1, a variety of W-mode waves, collectively called VLF emissions, are observed in the magnetosphere. All of them, excepting VLF transmitter signals and lightning generated whistlers, have apparent origins in the magnetosphere. A general discussion of these naturally occurring VLF emissions, beyond the scope of this paper, can be found in several review papers on the subject [1, 48, 53, 54, 74, 101, 115]. This paper brieﬂy reviews those aspects of the propagation of VLF emissions which are aﬀected by the presence of FAI of all types, including plasmapause. The generation and propagation of the VLF emissions is closely related to the properties of the cold and hot plasma distributions found in the regions where these emissions are generated. W-Mode Observations when the Source is Near the Equator: Ducted Propagation It is believed that plasmaspheric hiss, mid-latitude hiss, chorus, periodic and quasi-periodic emissions, and triggered emissions are generated in the equatorial region of the magnetosphere via some kind of gyroresonance related instability [e.g., 1, 115, and references therein]. The energy sources in most of these cases are the energetic electrons (∼1–100 keV) trapped in the magnetosphere. Figure 6.15 shows dynamic spectra illustrating ground observations

174

V.S. Sonwalkar (b)

(a) dB kHz SIPLE STATON 0 4.0

02 MAY 86

17 20 UT

EIGHTS STATON

16 OCT 63

3.0 -10 2.0

2.0 -20 -30

1.0 0

5

10 Time (Sec)

ref 52 dB ∆ f 31 Hz

0

5

10

15

ref 67 dB ∆ f 15 Hz

Time (Sec)

Fig. 6.15. (a) Double banded ELF/VLF chorus observed at Siple, Antarctica (λg = 76◦ S, φg = 84◦ W, λm = 60◦ S). In this case chorus bands are accompanied by weak bands of midlatitude hiss. (b) Periodic emissions observed at Eights, Antarctica (λg = 75◦ S, φg = 77◦ W, λm = 64◦ S) (courtesy of VLF group, Stanford University)

of chorus (Fig. 6.15a, discrete rising tones), mid-latitude hiss (Fig. 6.15a, diffuse bands between 500–1500 Hz and 2–3 kHz), and periodic emissions (Fig. 6.15b). Excepting plasmaspheric hiss, VLF emissions generated near the equator are observed on the ground. As shown in Fig. 6.7a, rays A and B, in general W-mode waves generated near the equator will be reﬂected back into the magnetosphere by either MR reﬂection or by total internal reﬂection. The fact that we routinely observe these emissions on the ground indicates that these emissions are somehow trapped inside one of the nearby ducts and brought down to the ground, as illustrated by rays C and C in Fig. 6.7. As an example of ducted propagation of VLF emissions with the source in the magnetosphere, consider periodic emissions. Figure 6.15b shows a spectrogram of periodic emissions recorded at Eights, Antarctica. A sequence of discrete emissions or clusters of discrete emissions showing regular spacing is called periodic emissions [54, 101]. Usually their period is constant, but on occasion it changes slowly. They generally occur at frequencies below 5–10 kHz with a few kHz bandwidth and a period in the range from two seconds to six seconds. It was established that the period of the emissions is identical to the two-hop whistler transit time [Dowden, 33]. Observations of periodic emissions at conjugate stations Byrd-Hudson Bay showed that the emissions appeared at the two stations with a delay of about 0.8 s [Lokken et al., 77]. Periodic emissions are observed at latitudes that correspond to closed magnetic ﬁeld lines and are rarely observed on satellites [Sazhin et al., 100]. These characteristics indicate that the periodic emissions, like other VLF emissions observed on the ground, have propagated in ﬁeld aligned ducts. The fact that they are rarely observed on the satellite hints that probably they are also generated at small wave normal angles within ducts and thus remain trapped inside the ducts. Thus a satellite has to be inside a duct to record these emissions, the probability of which is very small as discussed earlier.

6 Density Irregularities and Whistler Propagation

175

Chorus is routinely observed both on the ground and on spacecraft and many features, though not all, of chorus observed on the ground and spacecraft are similar [e.g., 53, 101, 115, and references therein]. This indicates that chorus may be propagating in both ducted and non-ducted modes. Recent analysis of chorus observed on the POLAR [Lauben et al., 75] and Cluster spacecraft [Santolik et al., 95, 96] support this viewpoint. Lauben et al. [75] ﬁnd that the upper-band chorus waves (f ≥ 0.5fHeq , where fHeq is the equatorial gyrofrequency) are emitted with wave normal θs 0, where θs is the wave normal angle at the source region near the equator, while the lower-band chorus waves ( f ≤ 0.5fHeq ) are emitted with θs θG , where θG is the Gendrin angle, giving minimum value of refractive index parallel to B0 . For both frequency bands, these respective θs values lead to wave propagation paths which remain naturally parallel to the static magnetic ﬁeld in the source region over a latitude range of typically 3◦ −5◦ , providing an ample opportunity for cumulative wave/particle interaction and thus rapid wave growth, notably in the absence of ﬁeld-aligned cold plasma density enhancements (i.e., ducts). Chorus generated at low wave normal angle can easily be trapped in a nearby duct and brought down to the Earth. Chorus at wave normal angles near θG can be guided to the Earth by the plasmapause [Inan and Bell, 61]. Mid-latitude hiss is frequently accompanied by whistlers echoing in the same path (duct) and ampliﬁed in the same frequency band. Dowden [33] suggested that at least part of the midlatitude hiss may be generated by the superposition of highly dispersed, unresolved, overlapping whistlers continuously echoing from hemisphere to hemisphere along a duct. However to overcome geometric loss of 10–20 dB corresponding to the small fraction of reﬂected energy being re-ducted, a magnetospheric ampliﬁcation of 10–20 dB is required to maintain steady hiss intensity over time scales of tens of minutes [Thompson and Dowden, 132]. On the other hand, ground based direction ﬁnding measurements suggest that midlatitude hiss is generated just inside the plasmapause [Hayakawa, 51]. There are a few examples in the literature where a more direct correlation between W-mode emissions and FAI is made. Koons [71], using data from VLF/MR swept frequency receiver on AMPTE IRM spacecraft, found strong enhancement of W-mode waves (identiﬁed as hiss or chorus) correlated with density enhancement in the outer plasmasphere, in the vicinity of plasmapause, between L = 4 and L = 6, which he identiﬁed as ducts. The ducts consisted of more than 40% density enhancement and had a typical half width of 250 km. The wave intensity inside the duct was an order of magnitude larger than that outside. On Cluster spacecraft Moullard et al. [83] found that electron density ﬂuctuations are regularly observed near the plasmapause together with hiss or chorus emissions at frequencies below the electron cyclotron frequency. Instruments on board Cluster spacecraft often observe two such emission bands with ﬂuctuating wave intensities that suggest wave ducting in density enhancements as well as troughs. During a plasmapause crossing on June 5, 2001 (near the geomagnetic equator, L = 4 − 6,

176

V.S. Sonwalkar

afternoon sector), density ﬂuctuations up to hundreds cm−3 were found while whistler mode waves were observed in two separate frequency bands, at 100– 500 Hz (correlated to the density ﬂuctuations) and 3–6 kHz (anti-correlated). Using the electron density and wave data from Cluster spacecraft, Masson et al. [80] found that in the vicinity of the plasmapause, around the geomagnetic equator, the four Cluster satellites often observe banded midlatitude hiss, typically 1–2 kHz bandwidth and center frequency between 2–10 kHz. They found that the location of occurrence of the hiss was strongly correlated with the position of the plasmapause with no MLT dependence. Plasmaspheric hiss is a broadband and structureless whistler-mode radiation that is almost always present in the Earth’s plasmasphere and is commonly observed by magnetospheric satellites, but not observed on the ground [53, 130]. It is observed in the frequency range extending from a few hundred Hz to 2–3 kHz with a peak below 1 kHz. Plasmaspheric hiss should be distinguished from auroral hiss which is observed at low altitudes in the auroral regions and covers a much wider frequency band (∼100 Hz – 100 kHz) and from mid-latitude hiss that is observed from ground stations and on satellites in 2–10 kHz range. Plasmaspheric hiss is found throughout the plasmasphere and is stronger in the daytime sector compared to the midnight-to-dawn sector, and generally peaks at high (>40◦ ) latitudes; it often shows a sharp cutoﬀ at the plasmapause for frequencies below ∼1 kHz, though it is also observed outside the plasmasphere at higher frequencies. W-mode waves generated in the magnetosphere cannot reach the ground (Fig. 6.7a, rays A and B). VLF emissions such as mid-latitude hiss and chorus, discussed above, are observed on the ground, presumably they are generated at low wave normal angles, and get trapped in a duct and thereby propagate down to the Earth. A question naturally arises: why is plasmaspheric hiss not observed on the ground? The answer to this question may come from studies devoted to determining the wave normal directions of plasmaspheric hiss [e.g., 117, 126, and references therein]. The general conclusion from these studies is that plasmapheric hiss most frequently propagates with large wave normal angles. As a result it cannot be trapped inside a duct and be seen on the ground. On the other hand, recent results from POLAR spacecraft show that plasmaspheric hiss often propagates at wave normals that make small angle with the geomagnetic ﬁeld [Santolik et al., 97]. It thus appears further work is warranted to understand why plasmaspheric hiss is not observed on the ground. It appears that FAI may play an important role in the generation of plasmaspheric hiss. In one generation mechanism, proposed by Thorne et al. [131], plasmapause plays a central role. In this mechanism, hiss propagates along certain cyclic trajectories, made possible by reﬂections from plasmapause. This allows hiss to reach the equatorial region repeatedly, each time with a small wave normal angle, thus permitting maximum possible cyclotron resonance interaction with energetic electrons. Alternatively, hiss may be generated by

6 Density Irregularities and Whistler Propagation

177

lightning-whistlers as discussed in Sect. 6.1.2 [35, 118]. In this mechanism, topside ionospheric FAI play a central role. W-Mode Observations when the Source is in the Auroral Region: Non-Ducted Propagation W-mode emissions commonly observed in the auroral and polar regions include auroral hiss, lower hybrid emissions, and Lion’s roar [74, 115, and references therein]. Auroral hiss (AH), shown in Fig. 6.16, is one of the most intense whistler mode plasma wave phenomenon observed both on the ground at high latitudes and on spacecraft in the auroral zone [e.g., 74, and references therein]. The apparent diﬀerence in the spectral form of continuous AH (Fig. 6.16a) and impulsive AH (Fig. 6.16b) has been attributed to diﬀerences in non-ducted versus ducted propagation [108, 109, 119]. The continuous structure-less spectra of continuous hiss can be explained in terms of non-ducted propagation and the impulsive, or sometimes falling tone spectra, as in Fig. 6.16b, of impulsive hiss in terms of ducted propagation. There has been some diﬃculty in understanding the propagation of AH from its source region to the ground and to a high altitude satellite. There is strong experimental evidence as well as theoretical analysis that indicates that AH is generated at large wave normal angle by Cerenkov resonance mechanism [74, 119]. The standard whistler mode propagation in a smooth magnetosphere, discussed in Sect. 6.3, predicts that auroral hiss generated at large wave normal angle along the auroral ﬁeld lines by Cerenkov resonance cannot penetrate to the ground. Two solutions for this problem have been suggested. Matsuo et al. [81] proposed that large scale FAI refract AH with θ ∼ θRES into AH with small wave normal angle which can fall into the transmission cone. Sonwalkar and Harikumar [119] argue that if the waves start with θ within a degree or so of the resonance cone angle, as the theory demands, then the relatively small tilts of horizontal gradients in the auroral ionosphere

Fig. 6.16. Examples of auroral hiss (AH) observed at South Pole, Antarctica (λg = 90◦ S, λm = 79◦ S). (a) Example of continuous auroral hiss (CAH). (b) Example of impulsive auroral hiss (IAH) showing dispersion [adapted from Sonwalkar and Harikumar, 119] (Reprinted with permission of American Geophysical Union)

178

V.S. Sonwalkar

are insuﬃcient to bend the wave normals enough for them to intercept the transmission cone. They propose that small scale FAI scatter AH propagating close to the resonance cone into small angle W-mode waves with small enough wave normal angles such that they fall into the transmission cone and thus propagate down to the Earth. This mechanism is shown schematically in Fig. 6.7f. There is some experimental evidence supporting this mechanism. Radar observations show a strong correlation between the occurrence of VLF hiss in the 1–10 kHz band and 18 MHz radar echoes from the F-region ﬁeldaligned irregularities [However and Gluth, 60]. Thirty two of the thirty three hiss events observed were associated with 18 MHz radar events. On occasion the hiss and radar events showed very close time correlation. Assuming that 18 MHz radar echoes result from scattering of radar signals from FAI with scale sizes equal to half the radar wavelength, the radar observations indicate presence of ∼15 m scale size FAI. Auroral cavity, a region of reduced electron density, may aﬀect propagation of AH upward from its source region. Snell’s law when applied to a sharp density boundary slanted with respect to the geomagnetic ﬁeld predicts that large wave normal angle W-mode waves can refract into small wave normal angle waves. Morgan et al. [82] suggest that through this mechanism, upward propagating auroral hiss converts from the short wavelengths ( m) observed at low altitudes to longer wavelengths (1–3 km) inferred at high altitudes by undergoing multiple reﬂections from the tilted sides of an auroral cavity which spatially diverges with increasing altitude. Satellite observations have often shown quasi-electrostatic plasma wave bands with a sharp cutoﬀ near the local lower hybrid resonance (fLHR ) frequency [4, 5, 18, 72, 73]. These bands are observed at mid- to high latitudes in the altitude range of ∼1000 km to a few thousand kilometers. These waves are often excited by whistlers, or they could simply be hiss emissions with a cutoﬀ at fLHR . The cold plasma theory predicts a resonance at the lower hybrid frequency for whistler mode waves propagating perpendicular to the static magnetic ﬁeld. At frequencies higher than the local LHR frequency, the resonance takes place at wave normal angles smaller than 90◦ . This explains the sharp lower cutoﬀ at fLHR as well as the quasi-electrostatic nature of the band [Brice and Smith, 18]. Several mechanisms including wave-particle resonances, parametric instabilities, and linear mode conversion have been proposed to explain the generation of lower hybrid waves. Bell and Ngo [7] have proposed that lower hybrid waves are excited when whistler mode waves are scattered by 10–100 m scale plasma density irregularities (Fig. 6.6). Propagation guided by FAI irregularities along open ﬁeld lines at high latitudes has been used to explain Lion’s roar observations at low altitudes in the polar cusp. Lion’s roar are intense sporadic whistler mode wave bursts observed in the Earth’s magnetosheath [1, 115, and references therein]. They are found in the 100–200 Hz frequency range and typically last for ∼10 s, though at times they can last for ≥5 minutes.

6 Density Irregularities and Whistler Propagation

179

Gurnett and Frank [46, 47], using Injun 5 data, detected lion roars in the high latitude magnetosheath near the polar cusp at 70–80◦ geomagnetic latitude and at ≤3000 km propagating parallel to the geomagnetic ﬁeld. They suggested that these waves were generated in the magnetosheath and trapped there in open tubes of force of the geomagnetic ﬁeld. These waves propagate along B0 , being guided by ﬁeld-aligned irregularities, and reach polar zone at ∼3000 km altitude. W-Mode Observations when the Source is at Low Altitude ( fp , presumably Langmuir waves, and in CMA3 in the form of slow-Zmode waves [James, 14]. The explanation of the slow-Z-mode observations

7 Dipole Measurements of Waves in the Ionosphere

197

called upon hot-plasma theory and an intermediate process of radiation by sounder-accelerated electrons to explain the characteristics of the waves at the OC receiver subpayload. 7.2.2 Whistler-Mode Propagation Near the Resonance Cone The two lowest pulse carrier frequencies transmitted in the OC two-point propagation experiment were 25 and 100 kHz, each in a receiver bandwidth of 50 kHz. Figure 7.2 is a detailed spectral display of the whistler-mode waves received in these two bands on the ﬂight downleg [James, 13]. These two greyscaled displays considerably expand the lowest part of Fig. 7.1, and show the spectra of received signals for selected portions of the history. The lower ﬁgure is for the bandwidth centred at 25 kHz and contains evidence of strong signals at that band centre. The same is true for the upper panel showing some of the 100-kHz band history. The two panels have been positioned in the diagram to have both the time after launch (TAL) and frequency scales maintained between the two. It is therefore easy to see that in addition to the spectral strength at the 25- and 100-kHz carriers, there is a broad swath of signal enhancement centred ﬁrst near the carrier in the bottom frame, then running diagonally toward its upper right corner. This same swath is seen to continue diagonally up through the top frame. It was suspected that the enhancement swath was a signature of loweroblique-resonance eﬀects [James, 12]. The “resonance cone frequency” frc is the whistler-mode frequency that puts the group velocity direction, at angle θgr with respect to B, along the transmitter-receiver direction, which is at angle δ with respect to B. Applying an expression for tan θgr as a function of f, fp and fc [Stix, 28] and equating it to tan δ leads to the relation tan2 δ = tan2 θgr =

2 −1 + fp2 /(frc − fc2 ) 2 2 1 − fp /frc

(7.3)

Inverting this equation for frc as a function of all the known variables throughout the total period in Fig. 7.2 produced the thin continuous line in that diagram labelled “frc ”. The frc locus is found to track the centre of the swath, conﬁrming the suspicion about the resonance-cone nature of the waves. This conclusion was supported by the observation of spin modulation of the signal in Fig. 7.2. Seen in both panels, spin modulation is especially clear in the lower frame in 700 < T AL < 750 as a series of pixels at 25 kHz that are alternatively strong and weak. In theory, the electric ﬁeld is polarized linearly along the wave vector direction for waves near the resonance condition. Fortuitously on OC, the major-frame duty cycle of HEX-REX is 3 seconds and the spin period of the receiving subpayload is very close to four times greater, 12 s. During each spin half rotation, HEX-REX executes two duty cycles; on one cycle, the receiving dipole is aligned close to the transmitted wave E; 3 s later, the alignment is perpendicular. Hence the clear 1-on /1-oﬀ pattern emerges.

198

H.G. James

Fig. 7.2. Summary plot of the detailed signal spectra of the 25- and 100-kHz pulses, for those portions of the downleg ﬂight when the resonance-cone frequency frc , as deﬁned in (7.2) and traced with the overlaid line, lies within either of the bandwidths centred on those two frequencies. The two panels have been positioned in the diagram to have both the time and frequency scales maintained between the two

The resonance cone signals at the carrier frequencies in Fig. 7.2 were much stronger than the electromagnetic whistler-mode signals at higher carrier frequencies in Fig. 7.1. These absolute signal strengths have been evaluated in a link calculation by Chugunov et al. [4]. Their detailed computation for the25kHz case took into consideration the geometry of the double-V dipoles on both ends of the link. A radiating theory for the transmitting dipoles that is pertinent to propagation directions near the cone was used [Mareev and Chugunov, 22]. For the receiving dipoles, a new application of the reciprocity principle [Chugunov et al. 3] was employed to estimate the receiving dipole’s Leﬀ , which was found to be 30 times its physical length. Good agreement between observed and theoretical resonance-cone signals was obtained.

7 Dipole Measurements of Waves in the Ionosphere

199

The good agreement obtained has implications for the interpretation of whistler-mode waves in space plasma. Sects. 7.3.2 through 7.3.4 following will illustrate this with a retrospective examination of some past observations of oblique-resonance-cone waves in the ionosphere.

7.3 Retrospective on Past Observations Section 7.1 reviewed analyses of OC data that conﬁrmed the theory for EM waves. The success of the classic short-dipole theory, represented here by the Kuehl [18] CW theory, has motivated a re-examination of reports about dipole Leﬀ for EM cases; this is in Subsect. 7.3.1. The success of the novel Chugunov [3] theory of dipole Leﬀ for quasi-electrostatic resonance-cone waves has prompted a return to published analyses of the amplitudes of such waves. Discussion along this line for both lower-oblique resonance waves (whistler mode) and upper-oblique resonance-cone waves (slow-Z mode) follows in Sects. 7.3.2 through 7.3.4. 7.3.1 Dipole Eﬀective Length for EM Propagation Jordan’s [16, pp. 336] deﬁnition of a generalized Leﬀ is taken as a point of departure. His (10-81) when applied to a cold plasma theory implies that the complex eﬀective-length vector can be written Leﬀ = ek Lkeﬀ + eθ Lθeﬀ + e φLφ eﬀ

(7.4)

This refers to a wave-vector space speciﬁed in spherical coordinates with the directions corresponding to radial direction k, polar-angle direction θ and azimuthal direction φ. The E polarization in magnetoionic theory in general has components in all three directions, as opposed to the vacuum case of Jordan’s (10–81) where the polarization ellipse is perpendicular to the wave vector. Under the reciprocity principle, it is assumed that an expression of the same form as Jordan’s (10–82) can be used to represent the vector ﬁeld E of an active dipole in a magnetoplasma as E=

60π I(0) Leﬀ , rλ0

(7.5)

where I(0) is the current injected at the antenna terminal, r is the distance and 0 is the vacuum wavelength. An expression for E was obtained from the Kuehl [18] theory used in the above-described tests of the OC radiation ﬁeld. His (14) can be rewritten as E=

k02 As e−iφ (eks Bs + eθs Cs + eφ Ds ) , 4π0 r

(7.6)

200

H.G. James

in which k0 is the vacuum wave number, φ is the phase, and we concentrate on a speciﬁc mode and solution of the dispersion relation, labelled with s. Tests of theory for radiated ﬁelds or Leﬀ must be carried out with due regard for two complications: First, the possibility that in some modes like the whistler, a single group-velocity direction can correspond to between one and three diﬀerent propagation directions; second, in space measurements, waves can take diﬀerent paths from a source to a receiver [Sonwalkar et al. 25]. The coeﬃcients As , Bs , Cs , Ds depend on the plasma parameters, the angle θ between the direction of propagation and B, and the dipole moment of the antenna current. When a triangular current distribution is assumed on the antenna, the dipole moment has a magnitude I(0)L/(4πf ). With this and the dipole-moment direction cosines inserted into the coeﬃcients, (7.6) can be rearranged to read E=

60πI(0) L (eks As Bs + eθs As Cs + eφ As Ds ) . rλ0 2

(7.7)

The Kuehl [18] expression is seen to be of the same form as Jordan’s (10– 82) in (7.5). The resulting coeﬃcients of the unit vectors are identiﬁed with the corresponding Leﬀ values in (7.4). It thus appears that the cold-plasma theory is consistent with the Jordan concept of general Leﬀ involving a vector description. For conformity with the principle of reciprocity, the open-circuit voltage induced on a receiving dipole is presumed to be Voc = E · =

L (eks As Bs + eθs As Cs + eφ As Ds ) 2

L (A B Ek + As Cs Eθ + As Ds Eφ ) . 2 s s

(7.8)

The coeﬃcients Bs , Cs , Ds are consistent with the polarization ratios given by the cold-plasma theory for the propagation direction. The coeﬃcients also depend on dipole orientation. Let us now see where the hypothesis of (7.8) leads when its predictions are compared with observations. Sonwalkar and Inan [26] scaled electric and magnetic ﬁelds measured on the DE-1 spacecraft to determine Leﬀ of the 200-m wire dipoles ﬂown on that satellite. Two measurement sets were found to give Leﬀ = 219.8 and 224.6 m. These evaluations were based on their expression (1) for Leﬀ , an assumption that has had wide use in space radio science. Often, Leﬀ as used in Sonwalkar and Inan [26] has been assigned a value of L/2, half the physical length of the dipole, appropriate for vacuum propagation detected by dipoles with L λ, the wavelength. The speciﬁc plasma parameters and geometry reported in Sonwalkar and Inan [26] allow Leﬀ to be evaluated from present (7.8) as L/2 times the multiplier factors As Bs , As Cs , As Ds for the three respective components. In the case of the whistler mode, the k and θ electric ﬁeld components are in phase with each other and in quadrature with the φ component. One axis of the E ellipse is the vector addition of the Ek and Eθ components and the

7 Dipole Measurements of Waves in the Ionosphere

201

other is the Eφ . All three components are generally needed to determine Voc . The magnitude of the eﬀective length of the dipole is !1/2 ≡ (L/2)M . (7.9) |Leﬀ | = (L/2) (As Bs )2 + (As Cs )2 + (As Ds )2 The multiplier M thus deﬁned for the Sonwalkar and Inan [26] cases is plotted as a function of the spin angle ξ in Fig. 7.3 with continuous and broken line. The inset ﬁgure shows that the spin axis of the DE-1 spacecraft is approximated to lie along the y axis. The wave vector direction and the plasma parameters take the values reported by the authors. The results are M values that oscillate between about four and zero. The theory thus conﬁrms that a short dipole receiving EM whistler-mode propagation can have |Leﬀ | ≈ L as reported by SI. Further comparison is not possible since Sonwalkar and Inan [26] assume that Leﬀ is isotropic whereas it is a vector here, depending on the orientations of both the wave vector and the dipole axis. Imachi et al. [8] determined Leﬀ of wire dipoles on the Geotail spacecraft by analyzing electric- and magnetic-ﬁeld amplitudes of chorus emissions. This result has been checked against the present method assuming that these whistler-mode waves propagate along B. The stated plasma and working frequencies and a reasonable electron gyrofrequency, 10 kHz, provide the variation of M (ξ) in the dot-dash line in Fig. 7.3. Although the details of the method of determining Leﬀ are not given by Imachi et al. [8], it is found that the range of M for all ξ brackets 1, the value deduced by those authors. Finally it is noted that (7.5) above simpliﬁes to the familiar expression Eθ =

60πI(0) L cos θ rλ0 2

(7.10)

for short dipoles and for working frequencies much higher than the electron characteristic frequencies, i.e., for the vacuum propagation conditions of (7.1). To check this, the multipliers of (7.8) were calculated for the same plasma characteristic frequencies as for Imachi et al. [8] but for the O mode, at a working frequency of 30 MHz. The result in Fig. 7.3 is the dot-dot-dot-dash curve, labelled “HF”, showing M when the θ term is the only term. The k has been put along the z axis. For the particular case of ξ = 0◦ , θ = 0◦ , for which the propagation direction is at right angles to this vacuum dipole axis, the r, θ, and φ multipliers have values of 0, −0.5i and −0.5, respectively. This same set becomes 0, −0.5i and 0.5 when run for the X-mode. The total Leﬀ magnitudes found from the sum of the two solutions are then (L/2)(0., 1., 0), corresponding to the expected linear polarized ﬁeld at θ = 0◦ from a simple dipole in vacuum, which radiates and detects the θ component only. 7.3.2 Intensity of Auroral Hiss Correct methodology for the dipole measurement of the absolute magnitudes of EM radio emissions in space includes a correct knowledge of the dipole Leﬀ .

202

H.G. James

kuehlef6.for, kuehl2.pro, Kuehl2.PS

Effective length multiplier, M

4 Sonwalkar and Inan (1986)

3

2

Imachi et at (2000) 1

0

0

45

90 135 Spin phase angle, ξ (˚)

180

Fig. 7.3. The multiplier M in (7.9) giving the magnitude of Leﬀ as a function of the dipole spin angle ξ, for comparison for three diﬀerent situations. The continuous and broken lines are the computed M for the two cases reported by Sonwalkar and Inan [26], the dot-dash curve for the Imachi et al. [8] conditions, and the lowest curve (HF) for vacuum conditions

This is illustrated in the history of auroral hiss. In the 1970s, there was interest in understanding the generation of hiss. Earlier work in space and astrophysical contexts had drawn attention to the idea that charged particles moving through magnetoplasma may produce EM radiation through the Cherenkov eﬀect – a particle radiates when its velocity matches the phase speed of EM plasma waves. An important question to be answered was whether a singleparticle theory suﬃced, wherein the total emission of a hiss source region could be obtained by considering all the particles therein as single incoherent radiators. Or were the absolute hiss levels high enough to require a more powerful process, e.g., a coherent beam-plasma interaction? Tests were carried out on diﬀerent satellite data sets. One example was ISIS-I observations of hiss in the dayside cusp at LF and MF [James, 9]. As illustrated in Fig. 7.4, that investigation dealt with passes of the spacecraft through L shells with soft electron precipitation. The observed evolution of hiss spectra was quantitatively consistent with spreading out of radiation

7 Dipole Measurements of Waves in the Ionosphere

203

Fig. 7.4. The trajectory of the ISIS-I satellite takes it through magnetic ﬁeld lines where its soft-particle spectrometer measures soft electron ﬂuxes [James, 9]. Low-frequency hiss is also measured by the sounder receiver as the spacecraft approaches, passes through and recedes from the precipitation. The calculation of total noise power ﬂux spectral density is computed within a bandwidth f ± ∆f by tracing resonance-cone group rays from the spacecraft location to establish 2 three-dimensional group cones that intersect the ﬂux sheet and so determine the total volume of energetic particles irradiating the spacecraft for that location and bandwidth (Reprinted with permission of the American Geophysical Union)

from the source ﬂux tubes on resonance-cone group paths. In one case study applying ISIS-I energetic-particle, ambient-density and wave measurements, incoherent-radiation theory was used to calculate theoretical electric-ﬁeld magnitudes. These were compared with values scaled from receiving-dipole voltages when the dipole was assumed to have an Leﬀ equal to its physical length, L. The ratio of the observed to theoretical ﬁeld strengths was about 20 dB, which led to the conclusion that the test-particle theory was not applicable. Similar discrepancies in the same sense were reported in other work, mostly at very low frequencies in the nightside ionosphere. Jorgensen [17] modelled the hiss generation region in the lower ionosphere and computed resulting

204

H.G. James

power ﬂux densities as high as 10−14 Wm−2 Hz−1 at 10 kHz. Gurnett and Frank [6] and Mosier and Gurnett [23] described the morphology of hiss with respect to other auroral-latitude phenomena observed with the Injun-V spacecraft. Power ﬂux densities for strong events were of the order 10−11 Wm−2 Hz−1 . Lim and Laaspere [20] also computed hiss levels at VLF and LF for typical energetic-ﬂux and ambient parameters. For comparison with the foregoing VLF observations, their computed ﬂux intensity was less than 10−13 Wm−2 Hz−1 . Taylor and Shawhan [29] carried out a case study of VLF hiss and energetic-particle spectra observed on Injun V. They came to similar conclusions, that the shape of the hiss spectrum is as expected based on ray paths and geometry but that the intensity of VLF hiss levels calculated was 2 orders of magnitude below observations. Maeda [21] reﬁned the model of the magnetosphere and calculated dayside hiss levels of 10−14 Wm−2 Hz−1 for strong electron ﬂuxes. For a wider summary of observed hiss intensities, see the Table 1 in LaBelle and Treumann [19]. Table 7.1 summarizes the preceding paragraph. The upshot of several parallel investigations of hiss power ﬂux was that observed values exceeded by at least 100 times values obtained from calculation using the incoherent theory [LaBelle and Treumann, 19]. This held true for a variety of diﬀerent observational circumstances. The power-ﬂux ﬁnding was typically followed by the conclusion that the incoherent test-particle theory for radiation was not applicable and that one should look to more powerful wave-particle interaction theory. However, the recent OC work at 25 kHz has now found that for resonance cone waves Leﬀ = 30 L. The ISIS-I data originally were interpreted assuming Leﬀ = L. This implies that electric ﬁeld strengths deduced were 30 times too large, or that the power ﬂux values were 20 log10 30 = 29.5 dB too high. Since this ﬁgure is similar to the disparity between calculated and observed hiss intensities, the rejection of the incoherent theory in the 1970s should now be reconsidered. Table 7.1. Auroral Hiss Field Strength Measurements and Calculations Observation VLF (0–100 kHz) Gurnett and Frank (1972) 0.5–1 ×10−11 Mosier and Gurnett (1972) >1, 5 × 10−11

LF-MF (100–500 kHz) James (1973) 0.81–4.6 ×10−17

Incoherent Radiation Theory Jorgensen (1968) 10−14 Lim and Laaspere (1972) 10−13 Taylor and Shawhan (1972) 0.5 − 3.2 × 10−13 Maeda (1975) 10−14 Ratio 11–32 dB James (1973) 0.21 − 8.8 × 10−19 Ratio: 17–26 dB

7 Dipole Measurements of Waves in the Ionosphere

205

That auroral hiss generation in the low-altitude auroral oval is correctly described with an incoherent theory may gain some credence when recent work on particle acceleration is consulted. Auroral electron acceleration is thought to result from the action of the parallel electric ﬁeld of shear Alfv´en waves and to occur at altitudes between about 3500 km and a few earth radii [Paschmann et al., 24, , pp. 187-189]. The low-earth-orbit-satellite observations cited above were all made below 3100 km, that is, below the acceleration region. Electron ﬂuxes in that region are expected to have lost their Alfv´en-imparted coherence through beam-plasma interactions. Various instability mechanisms like those producing hiss or auroral kilometric radiation may have the dominant role. In the former case, wave growth and saturation will happen over vertical scales of the order of ten plasma periods multiplied by particle velocities, say, several kilometers. What would then emerge from the bottom of the acceleration region would be turbulent electron ﬂuxes with spatial coherence scales corresponding to the saturation spectrum of the created waves. Paschmann et al. [24] state that the beam-plasma process causes quasilinear plateauing of the beam at the intermediate altitudes studied by James [9] and Taylor and Shawhan [29]. The eﬀect of plateauing is presumed to be one of shifting beam electrons to lower energies. However, at LF and MF, the hiss ray directions at frequencies not far below fp are not far from horizontal. Hence it can be argued that the particles measured by satellite are indeed responsible for the hiss observed on the same spacecraft, especially inside the precipitation L shells. The presence of turbulence would mean that overall the descending ﬂux would radiate incoherently but that the individual radiating entities would not be individual electrons but a three-dimensional ensemble of micro-structures. What might be the dimensions of a hiss-radiating microstructure? First of all, it is noted that an upper-limit on the cross-ﬁeld dimension of the emission source is given by the perpendicular scale size of the Alfv´en wave, which is of the order of the electron inertial length, 1 km. Therefore already the acceleration concept implies incoherence because the observed source regions in the auroral oval must contain a large number of such sources. Further, if thermal damping of whistler-mode waves aﬀects resonance-cone waves at refractive indices >1000, then saturated MF hiss at 300 kHz will have wavelengths of around 1 m. 1 m is probably a lower limit on structure sizes because the Debye length has approximately this order of magnitude in the acceleration region. Phase-space electron holes as small as a Debye length have been observed [Ergun et al., 5]. Assume that the hiss source is a ﬂux of 1-m structures. With a diﬀerential ﬂux density of 107 cm−2 s−1 ster−1 eV−1 in the dayside cusp, the density of particles in the beam is about 107 m−3 . This implies that the electric charge q in the incoherent radiated power expression should be 107 times a single-electron charge and that the number of the structures irradiating any observer must be reduced by 107 . In incoherent emission theory, the spectral power density varies as the square of the charge of the radiating “particle”. Hence to suppose that one observes radiating microstructure leads to a total

206

H.G. James

volume emission of (107 )2 /107 , a +70 dB correction in the absolute emission. The hypothesis of microstructure is discarded because the observation-theory disparity remains as large as before, albeit in the opposite sense. Hence the view is maintained that hiss in the topside ionosphere is incoherent radiation from a downward electron ﬂux below the acceleration region in the auroral magnetosphere. 7.3.3 Intersatellite Whistler-Mode Propagation Near a Resonance Cone Results from ISIS intersatellite propagation studies resemble the strong transmissions in the OC two-point experiment shown in Fig. 7.2. Whistler-mode transmission at 480 kHz was achieved between ISIS-I and ISIS-II 73-m sounder dipoles during a rendezvous campaign [James, 10]. ISIS I was in a 570 × 3520km polar orbit, while ISIS II had a nearly circular polar orbit around 1400 km altitude. The satellites were commanded into the “Alternate Mode” wherein the sounders cycled between conventional swept-frequency sounding and periods of about 20 s when the sounder pulse carrier frequency was held ﬁxed at 480 kHz. The rendezvous occurred at high latitude where the two orbital planes intersected. Transmissions in both directions were obtained on several rendezvous. In Fig. 7.5, a 4-s data excerpt from a rendezvous on 4 April 1974 displays representative pulses received. On the left are pulses emitted by ISIS II and received on ISIS I. The roles are reversed on the right side. Both satellites had 400-W emitters and produced input voltages at the other receiver of about 70 µV. Pulses about 100 µs long upon emission are seen to be dispersed spectacularly by 100 times and more at reception. All pulses in the 20-s interval from which this set is taken exhibit components with delays of the order of 10 ms. The pulses in 28 to 30 s also have sharp rising edges, on account of favourable antenna orientation at that time. Pulse stretching of this order was clearly maintained when the satellite separation lay well inside the group resonance cone for 480 kHz. Ray-optics studies were carried out based on three-dimensional electron density distributions obtained from the the real-height analysis of the sweptfrequency ionograms recorded during the rendezvous. Iterative techniques were used to ﬁnd the delays associated with direct inter-satellite propagation. Good agreement was found for the time of the sharp rising edges of pulses that had such, like those in Fig. 7.5. It was demonstrated that the pulse stretching was not caused by magnetoionic dispersion of the diﬀerent frequency components of the pulse. The signal delays required propagation near the resonance cone. Hence, the conclusion was that wave packets could take a great variety of scattering paths between the transmitter and the receiver, because the stretching was seen preferentially in the high-latitude irregular ionosphere. It was concluded that highly dispersed pulses arose from

7 Dipole Measurements of Waves in the Ionosphere

207

Fig. 7.5. Amplitude scans of whistler-mode pulses received reciprocally during a rendezvous of the ISIS-I and -II satellites [James, 10]. The pulse carrier RF was 480 kHz and the spacecraft separation was about 200 km (Reprinted with permission of the American Geophysical Union)

scattering by density irregularities along and inside the 480-kHz resonance cones through both spacecraft. An additional feature that supported the existence of quasi-electrostatic short-wavelength waves was the appearance of “primary” and “secondary” components of pulses. It was argued that this structure appeared because both the emitting and receiving dipoles had current distributions that favoured the emission or reception of certain wavelengths at a given frequency. These were only observed when each satellite lay near the 480-kHz group resonance cone of the other. The range of refractive indices, up to about 30, needed to explain the observed group delays correspond to wavelengths down to 20 m. This pointed to resonant- and anti-resonant-length relations between the now-ﬁnite-length dipoles and the waves, favouring the transmission of certain wave numbers and blocking others, hence the pulse envelope. 7.3.4 Intersatellite Slow-Z-Mode Propagation Near a Resonance Cone There have been observations of apparently intense waves propagating near the upper-oblique resonance cone in CMA 3. Another two-point experiment during the ISIS rendezvous campaign produced observations of strong, highly dispersed Z-mode pulses that appeared to carry an unusually large total energy [James, 11]. In fact the Z pulses resembled the inter-satellite whistlermode pulses just described, with comparable length and strength. Threedimensional ray-optics was applied to ﬁnd the phase paths of many pulse spectral components at the receiver. The received pulse envelope was then

208

H.G. James

constructed by an inverse Fourier transform, and found to have a very different shape from that measured. The conclusion again was that ionization irregularities eﬃciently quasi-electrostatically scatter waves and that the ISIS 73-m distributed dipole is sensitive to such waves. It was noted that highly dispersed slow-Z pulses were observed in ISIS monostatic ionograms when the satellite passed through high-latitude regions of density irregularities. These Z results are taken as further evidence of the combined eﬀect of the high Leﬀ , eﬀectively sensitizing the receiving dipoles, and of scattering ionospheric irregularities, providing long-delay paths along the group resonance cone. Benson et al. [2] in Chap. 1 of this volume have reviewed the Z mode as investigated using various spaceborne RF sounders. This includes new insights from the Radio Plasma Imager instrument on the IMAGE spacecraft. Pulse elongations similar to those found for whistler and Z- mode pulses during intersatellite propagation experiments between ISIS I and ISIS II were also found in recent whistler and Z-mode sounding experiments on IMAGE [Sonwalkar et al., 27]. These authors have put forward an interpretation similar to that given by [James, 10, 11], i.e. scattering by irregularities.

7.4 Conclusion The success of the cold-plasma CW theory for explaining dipole results from OEDIPUS commends wider use of that theory for supplying the eﬀective lengths of distributed dipoles for space-borne reception. The application of the principle of reciprocity brought over from the vacuum short-dipole theory appears useful for EM propagation. However the particular results about resonance-cone quasi-electrostatic propagation indicate a need for caution in the interpretation of signal levels. The importance in space physics of the correct dipole Leﬀ concept remains. It appears possible that hiss in the topside ionosphere is correctly described, after all, as incoherent radiation from a thermalized ﬂux left over from acceleration in the auroral magnetosphere. However, the interpretation of whistler-mode absolute intensities today is complicated by the clear evidence for scattered resonance-cone propagation. In the computation of power ﬂux densities from sources of emissions like hiss, which can have horizontal dimensions of hundreds of kilometers, there is now the added geometrical complication of dealing with scatter by irregularities. The scattering of EM VLF whistler-mode waves by ﬁeld-aligned small-scale irregularities into waves propagating near the cone has been explained as a consequence of either linear mode coupling or nonlinear mechanisms [Bell et al., 1, and references therein]. Whichever the case, the complete account of radiative transfer between source and observer brings a new challenge in understanding RF emissions of the atmosphere. This may be true not only for whistler-mode hiss but for other phenomena like hiss in the slow-Z mode.

7 Dipole Measurements of Waves in the Ionosphere

209

Given the state of knowledge about dipoles in 1978, the sustained pulses observed with topside sounders seemed to imply that great power was broadcast by the sounders along the oblique resonance cones. This inference is now tempered by the notion of large sensitivity of the receiving dipole to quasi-electrostatic resonance-cone waves, as represented by a big Leﬀ in the distributed dipole theory.

References [1] Bell, T.F., R.A. Helliwell and M.K. Hudson: Lower hybrid waves excited through linear mode coupling and the heating of ions in the auroral and subauroral magnetosphere, J. Geophys. Res. 96, 11,379–11,388 (1991). [2] Benson, R.F., P.A. Webb, J.L. Green, D.L. Carpenter, V.S. Sonwalkar, H.G. James and B.W. Reinisch: Active wave experiments in space plasmas: the Z mode. In: Proceedings Volume for the Ringberg Workshop, edited by J. LaBelle and R. A. Treumann, Springer Lecture Notes in Physics, Springer New YorkHeidelberg (2005). [3] Chugunov, Yu.V.: Receiving antenna in a magnetoplasma in the resonance frequency band, Radiophys. Quantum Electron. 44, 151–160 (2001). [4] Chugunov, Yu.V., E.A. Mareev, V. Fiala and H.G. James: Transmission of waves near the lower oblique resonance using dipoles in the ionosphere, Radio Sci. 38, 1022, doi:10.1029/2001RS002531 (2003). [5] Ergun, R.E., C.W. Carlson, J.P. McFadden, F.S. Mozer, L. Muschietti, I. Roth and R.J. Strangeway: Debye-scale plasma structures associated with magneticﬁeld-aligned ﬁelds, Phys. Rev. Lett. 81, 826–829 (1998). [6] Gurnett, D.A. and L.A. Frank: VLF hiss and related plasma observations in the polar magnetosphere, J. Geophys. Res. 77, 172–190 (1972). [7] Horita, R.E. and H.G. James: Two point studies of fast Z mode waves with dipoles in the ionosphere, Radio Sci. 39, doi:10.1029/2003RS002994 (2004). [8] Imachi, T., I. Nagano, S. Yagitani, M. Tsutsui and H. Matsumoto: Eﬀective lengths of the dipole antennas aboard Geotail spacecraft. In: Proc. 2000 Int. Sympos. Antennas and Propagat, (ISAP2000), IEICE of Japan, Tokyo 819– 822 (2000). [9] James, H.G.: Whistler-mode hiss at low and medium frequencies in the daysidecusp ionosphere, J. Geophys. Res. 78, 4578–4599 (1973). [10] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites, 2, Whistler-Mode pulses, Radio Sci. 13, 543–558 (1978). [11] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites, 3, Z mode pulses, J. Geophys. Res. 84, 499–506 (1979). [12] James, H.G.: Electrostatic resonance-cone waves emitted by a dipole in the ionosphere, IEEE Trans. Antennas Propagat. 48, 1340–1348 (2000). [13] James, H.G.: Electromagnetic whistler-mode radiation from a dipole in the ionosphere, Radio Sci. 38, 1009, doi:10.1029/2002RS002609 (2003). [14] James, H.G.: Slow Z-mode radiation from sounder-accelerated electrons, J. Atmos. Solar-Terr. Phys. 66, 1755–1765 (2004).

210

H.G. James

[15] James, H.G. and W. Calvert: Interference fringes detected by OEDIPUS C, Radio Sci. 33, 617–629 (1998). [16] Jordan, E.C.: Electromagnetic Waves and Radiating Systems, Prentice-Hall, Englewood Cliﬀs NJ (1950). [17] Jørgensen, T.S.: Interpretation of auroral hiss measured on POGO-2 and at Byrd Station in terms of incoherent Cerenkov radiation, J. Geophys. Res. 73, 1055–1069 (1968). [18] Kuehl, H.H.: Electromagnetic radiation from an electric dipole in a cold anisotropic plasma, Phys. Fluids 5, 1095–1103 (1962). [19] LaBelle, J. and R.A. Treumann: Auroral radio emissions, 1. Hisses, roars and bursts, Space Sci. Rev. 101, 295–440 (2002). [20] Lim, T.L. and T. Laaspere: An evaluation of the intensity of Cerenkov radiation from auroral electrons with energies down to 100 eV, J, Geophys. Res. 77, 4145– 4157 (1972). [21] Maeda, K.: A calculation of auroral hiss with improved models for geoplasma and magnetic ﬁeld, Planet. Space. Sci. 23:843–865 (1975). [22] Mareev, E.A. and Yu.V. Chugunov: Excitation of plasma resonance in a magnetoactive plasma by external source, 1, A source in a homogeneous plasma, Radiophys. Quantum Electron. 30, 713–718 (1987). [23] Mosier, S.R. and D.A. Gurnett: Observed correlations betwen auroral and VLF emissions, J. Geophys. Res. 77, 1137–1145 (1972). [24] Paschmann, G., S. Haaland and R.A. Treumann: Auroral plasma physics, Space Sci. Rev. 103, No. 1–4 (2002). [25] Sonwalkar, V.S.: The inﬂuence of plasma density irregularities on whistler mode wave propagation. In: Proceedings Volume for the Ringberg Workshop, edited by J. LaBelle and R. A. Treumann, Springer Lecture Notes in Physics, Springer New York-Heidelberg (2005). [26] Sonwalkar, V.S. and U.S. Inan: Measurements of Siple transmitter signals on the DE 1 satellite: wave normal direction and antenna eﬀective length, J. Geophys. Res. 91, 154–164 (1986). [27] Sonwalkar, V.S., D.L. Carpenter, T.F. Bell, M. Spasojevic, U.S. Inan, J. Li, X. Chen, A. Venkatasubramanian, J. Harikumar, R.F. Benson, W.W.L. Taylor and B.W. Reinisch: Diagnostics of magnetospheric electron density and irregularities at altitudes > k⊥ ), or at the upper-hybrid resonance perpendicular to the background magnetic ﬁeld (k⊥ >> k|| ), or at the oblique high-frequency resonances with fpe < f < fuh . When these excitations are followed by conversion to one of the electromagnetic modes (X or O), the resulting mode conversion radiation can propagate long distances to distant space-borne or ground-based receivers. Rocket Observations Indeed, almost every suitably instrumented experiment penetrating the auroral region records plasma waves in the band between the electron plasma frequency and the upper hybrid frequency [1, 15, 58] as reviewed by LaBelle

214

P.H. Yoon et al.

[34]. Early and recent experiments indicated the bursty nature of these auroral Langmuir waves [see, e.g., Ergun, 9]. Recent experiments include detailed comparisons with electron distribution function measurements. For example, McFadden et al. [43] report on waves near fpe and show evidence that these originate from a Landau resonance with the simultaneously measured electron beams. Similarly, Samara et al. [53] show evidence that waves near fuh result from cyclotron resonance with the measured electrons in which electrons transfer energy into the waves. Ergun et al. [10] and Kletzing et al. [31] use wave-particle correlators to measure directly the interaction giving rise to the auroral Langmuir waves. Recent experiments also show the frequency structure of these high frequency waves and their conversion to electromagnetic modes. Beghin et al. [2] show high frequency waves near the plasma frequency observed in the auroral zone under a variety of conditions using wave receivers on the Aureol-3 satellite. For underdense conditions (fpe < fce ), the wave structure is predominantly at the plasma frequency and below, while for overdense conditions (fce < fpe ) the waves are at and above the plasma frequency. The waves below fpe in the underdense case are attributed to mode conversion of Langmuir waves into whistler waves. McAdams et al. [40] observed similar phenomena during an auroral rocket ﬂight, PHAZE-II in 1997, which penetrated both underdense and overdense plasma conditions. Their wave receiver had unprecedented resolution achieved by full sampling of the waveforms. With this resolution, McAdams et al. [40] observed that in the underdense case, the waves just below the plasma frequency form multiple constant-frequency “bands”. These bands are of longduration, up to tens of seconds (tens of km distance). Sometimes individual band structures are punctuated by an intense burst of emission at the plasma frequency, when the rocket crosses the location where the band frequency matches the plasma frequency. McAdams et al. [40] attribute these bands to mode conversion to whistler mode, as suggested earlier by Beghin et al. [2]. The banded structure is a natural consequence of the bursty intermittent nature of the causative Langmuir waves. In the overdense case, McAdams et al. [41] report completely diﬀerent and more complex wave structure. As reported by Beghin et al. [2] the waves occur at and above fpe in this case. However, the high resolution shows that they consist of multiple discrete features, with bandwidths less than a few hundred Hz, separated by the order of one kilohertz. Sometimes up to four or ﬁve multiplets are observed, but more often doublets. Figure 8.1a shows a spectrogram of the high frequency electric ﬁelds observed during a 3 s time interval of the PHAZE-II experiment. The distinct, rapidly varying wave cutoﬀ corresponds to the plasma frequency as argued by McAdams et al. [41]. Just above the plasma frequency, structured waves with amplitudes of ∼0.7 mV/m occur. Where their frequency nearly matches fpe , their amplitude maximizes, and they are clearly associated and appear trapped in electron density depletions. Further away from these electron density depletions, where their frequency

Frequency (kHz)

8 Mode Conversion Radiation 2200

(a)

2165 2130 2095 2060 140.9

Frequency (kHz)

PHAZE-II

215

141.4

141.8

142.2

142.6

1740

143.0

143.4

RACE (b)

1705 1670 1635 1600 760.7

761.1

761.5

761.9

762.3

762.7

763.2

Seconds After Launch

Fig. 8.1. (a) High frequency electric ﬁelds observed during the PHAZE-II experiment show multiple discrete frequency features, with bandwidths less than a few hundred Hz, separated by the order of one kHz. (b) High frequency electric ﬁelds observed by RACE show similar structures but with diﬀerent frequency-time variations (Reprinted with permission of the American Geophysical Union)

exceeds the plasma frequency by ten percent or so, they appear as descending tones with short durations of order 100 ms. Hundreds of descending tones occur during a 10 s interval. By the time the rocket travelled 10 s (or about 10 km) away from the region with the electron beam and the electron density irregularities, the waves died out to an undetectable level. Since 1997, several further auroral rocket experiments have included fullwaveform measurements of high frequency plasma waves. Figure 8.1b shows a spectrogram of the high frequency electric ﬁelds observed during 10 s interval on one of these experiments, RACE, launched in 2002. The Langmuir waves in the overdense auroral ionospheric plasma show many features in common with those previously observed by McAdams et al. [41]. They are associated with the combination of auroral electron beam and electron density structure evidenced, as before, by variations in the wave cutoﬀ associated with the plasma frequency. Again the waves are most intense where their frequency nearly matches fpe , and there is evidence for trapping of the waves in density depletions in this region. Though not shown in Fig. 8.1b, observations from many kilometers away from this source region, where the wave frequency exceeds the local plasma frequency by ten percent or so, the wave intensity is lower, and again by the time the rocket travels about 10 km from the source region, the waves are undetectable [Samara et al., 52]. The detailed wave

216

P.H. Yoon et al.

structure consists of discrete features separated by about 1–2 kHz. The discrete features often occur in pairs. However, unlike the previous case, features with decreasing frequency are not observed exclusively. Several examples of features with increasing frequency are shown in Fig. 8.1b. In addition, some examples last far more than 100 ms. In one case not shown in Fig. 8.1b, a discrete feature is intermittent but continuously detectable for many seconds. McAdams et al. [41] labelled their observed features “chirps” based on the consistent decreasing frequency and short 100 ms lifetime. However, the recent observations show that this name is overly restrictive, since a far broader variety of frequency variations occurs. As mentioned above, auroral HF waves are detected throughout the band between the plasma and upper hybrid frequencies, and recent experiments include full waveform measurements of the waves near the upper-hybrid frequency, complementing the observations near fpe cited above. Figure 8.2, from Samara et al. [53], shows a spectrogram of waves at and just below fuh in an active aurora penetrated by the HIBAR rocket, launched in 2003. Two bursts of highly structured waves occur where the upper hybrid frequency matches the electron cyclotron harmonic (fuh ≈ 2fce ). The waves lie at and just below fuh , which is indicated by the weak wave cutoﬀ and conﬁrmed by a second wave cutoﬀ at the plasma frequency (not shown). In the ﬁrst wave burst, of duration 0.5 s and total amplitude up to 10 mV/m, the waves occur in 2–4 visible bands with bandwidth of about 1–5 kHz separated by about 8 kHz. Each of these bands is composed of several sub-bands with separation of order 1–2 kHz. The entire band structure decreases in frequency with increasing time and altitude, at a rate of 21 kHz/s, approximately ten times too high to be explained by the decrease of the electron cyclotron harmonic frequency with increasing altitude. The second wave burst is of shorter duration. In the second burst, the emission is also banded, but the separations of the bands are of order 2–4 kHz. It is not clear whether the band separation in the second event should be compared to the larger separation of the principal bands in

Fig. 8.2. A spectrogram highlighting waves just below fuh in an active aurora penetrated by the HIBAR rocket [from 53]. The HIBAR observations suggest trapping of the upper hybrid waves in electron density enhancements, as discussed in Sect. 8.3 (Reprinted with permission of the American Geophysical Union)

8 Mode Conversion Radiation

217

the ﬁrst event, or whether it should be identiﬁed with the sub-band structure in the ﬁrst event, which is about the same size. The electron density observations from HIBAR suggest that these upper hybrid waves are trapped in electron density enhancements: the second wave burst clearly corresponds to an enhancement, whereas the ﬁrst wave burst corresponds to a “shoulder” in the electron density proﬁle, a signature which may result from the one-dimensional sampling of a density enhancement by the rocket trajectory. Earlier lower-resolution rocket observations of auroral upper hybrid waves near fuh also suggested this trapping, supported by careful consideration of the wave dispersion relation by Carlson et al. [6]. Ground-Based Observations Strong evidence suggests that the structured waves excited near the upper hybrid frequency under the matching conditions fuh ≈ 2fce or fuh ≈ 3fce have converted to O mode EM radiation and been detected remotely with satellites overﬂying the auroral zone [e.g., James et al., 24] and with groundbased observatories at northern and southern auroral zone sites [e.g., Kellogg and Monson, 29, 30] and [Weatherwax et al., 59, 61]. Using a topside ionospheric sounder in a passive conﬁguration, James et al. [24] observe LO radiation emerging from sources in the topside ionosphere, and by ray-tracing the observed LO radiation from the satellite to its source, into the topside ionospheric density proﬁle measured with periodic active soundings, they conclude that the radiation is generated near the upper hybrid frequency at the matching conditions. Similarly, Hughes and LaBelle [22] ray-trace the observed radiation from a ground-based direction-ﬁnding antenna array into the bottom-side ionospheric density proﬁle measured with an incoherent scatter radar, and they likewise conclude that the radiation is generated near the upper hybrid frequency at the matching condition. Theoretical work shows that Z-modes at the fuh = N fce matching condition can be excited in auroral conditions [see, Kaufman, 28] and [Yoon et al., 64]. Furthermore, based on such density structures observed at times of auroral roar emissions, Yoon et al. [65, 66] show that for a narrow range of frequencies and initial wave phase angles, trapped Z mode can be converted to O mode via the “Ellis” radio window. Weatherwax et al. [62] also present a detailed maser instability analysis of upper-hybrid, Z, and X mode wave excited along an auroral ﬁeld line, discussing the generation of topside and bottomside radiation within the context of the observations of James et al. [24], Benson and Wong [3], and Hughes and LaBelle [8]. In the case of the ground-level observations, called auroral roar, deployment of high-bandwidth receivers with large data rates reveals that the mode conversion radiation is composed of complex patterns of discrete frequency features as shown by LaBelle et al. [35] and Shepherd et al. [55]. Figure 8.3 shows an example of this ﬁne structure, recorded April 21, 2003, at South Pole Station. The most intense of these emissions have amplitudes up to 500 µV/m.

218

P.H. Yoon et al.

Fig. 8.3. 2595–2685 kHz spectrogram recorded at South Pole Station at 2130 UT on April 21, 2003, showing auroral roar with complex ﬁne structure

As known from previous studies [35, 55] auroral roar consists of multiple ﬁne structures, with width as narrow as 0) can satisfy this criterion. The reverse is true in the overdense case (ωp0 > Ω). Below, we restrict our analysis to the underdense case. Yoon and LaBelle [68] treat the overdense case. The discrete frequency spectrum associated with eigenmodes trapped within the density structure can be derived on the basis of the continuity of the eikonal solutions of (8.14) across the cutoﬀ points, x0 [V (x)]1/2 dx = (n + 1/2) π , (8.15) −x0

where n = 0, 1, 2, . . . Note that the cutoﬀs x ± x0 are real and positive if the condition 2 2 + 3k 2 vT2 /2 < ω 2 < (1 + δ)(ωp0 + 3k 2 vT2 /2) (8.16) ωp0 is satisﬁed. Inserting the speciﬁc expression for V (x) deﬁned in (8.14) to the matching condition (8.15) and carrying out closed-form analytical integration over x, we arrive at 1/2 2 2 v 3k ω p0 T 1+ [K(η) − E(η)] = (2n + 1) π , 4δ 1/2 kL 2 ω ωp0 2 (1 + δ)(ωp0 + 3k 2 vT2 /2) − ω 2 η= 2 1/2 δ 1/2 ωp0 + 3k 2 vT2 /2

!1/2 .

(8.17)

In the above K(η) and E(η) are complete elliptic integrals. Equation (8.17) constitutes a transcendental dispersion equation which supports a discrete

8 Mode Conversion Radiation

227

frequency spectrum as its solution. The simplest approximation is to replace the full elliptic integrals with their small-argument series approximation, K(η) − E(η) ≈ πη 2 /4, which is valid if the normalized cutoﬀs ±x0 /L are suﬃciently small, x20 /L2 1. Adopting such a simpliﬁcation, we obtain an analytical expression for the eigenvalues, ωp0 ω= 2

3k 2 vT2 1+ 2 2ωp0

1/2

$

4(1 + δ) + a2n

!1/2

− an

% ,

(8.18)

(2n + 1) δ 1/2 . kL Applying the above approximate solution for the discrete spectrum requires that the solution ω 2 be conﬁned to the range speciﬁed by (8.16). For typical ionospheric conditions, the condition (8.16) is easily satisﬁed. In the numerical computation of the discrete wave spectrum (8.16), the width of the density structure plays a key role. Taking an example from the auroral ionosphere, if the plasma frequency is fp0 = ωp0 /2π ∼ 2 MHz, and the background electron thermal energy is T ∼ 0.1 eV, then we calculate an =

α ≡ Lωp0 /vT ≈ 105 L[km] . L = 100m implies α ≈ 104 , L =1 km implies α ∼ 105 , and L = 1 m implies α = 102 . The characteristic parallel wave number associated with Langmuir waves is determined by the condition for maximum wave growth rate. For V0 /vT b ≈ 10 and T ≈ Tb , we have kvT /ωp0 ≈ 0.1. In Figure 8.5, we plot the discrete eigenvalue f versus n, for fp0 = 2 MHz, kvT /ωp0 = 0.1, density enhancement factor δ = 0.05, and for two diﬀerent structure widths, L = 100 m and L = 500 m. Note that the wider structure leads to smaller frequency spacing between individual discrete modes. The L = 500 m 2064.8

2064

2064.6

2063

2064.4

f [kHz]

f [kHz]

L = 100 m 2065

2062 2061 2060

2064.2 2064

0

2

4

6 n

8

10

2063.8

0

2

4

6

8

10

n

Fig. 8.5. Discrete Langmuir wave spectrum f [kHz] versus the mode number n, for two diﬀerent density structure widths, L, corresponding to 100 m and 500 m

228

P.H. Yoon et al.

predicted eigenmode spacings are of order 0.5 kHz and 0.1 kHz, respectively. Yoon and LaBelle [68] ﬁnd similar eigenmode spacings in the overdense case, which is appropriate to the observations of McAdams and LaBelle [41]. These spacings are comparable to those observed by McAdams and LaBelle [41] and calculated by McAdams et al. [42]. 8.3.3 Discrete Upper-Hybrid Waves in Density Structures The frequency spacing associated with auroral roar ﬁne structure was ﬁrst addressed by Shepherd et al. [55], who proposed an explanation based upon geometrical considerations in which the excitation of multiplet discrete modes resulted from standing waves within planar ﬁeld-aligned density cavities with vertically converging walls. This approach was similar to an idea originally proposed by Calvert [5] to account for AKR ﬁne structure. This explanation faces several diﬃculties, however, which Shepherd et al. [55] discuss. The principal problem is that it requires the wave-vector to vanish at the reﬂection points, which means that the wave cannot continuously grow as it stands in the cavity. Combined with the relatively low growth rate, it is diﬃcult to envision how suﬃcient gain can be achieved to allow the discrete eigenmodes to grow. Yoon et al. [67] present a model which corrects the primary diﬃculty of the Shepherd et al. [55] approach by considering circulating eigenmodes in a cylindrical geometry. This geometry is similar to that suggested by Carlson et al. [6] to explain the enhancement of upper hybrid waves in an artiﬁcially enhanced density structure, although they did not consider eigenmode structure. As discussed above, recent observations in the auroral ionosphere suggest that waves near the upper hybrid frequency, polarized with wave-number primarily perpendicular rather than parallel to the background magnetic ﬁeld, exhibit ordered frequency spectra with peaks spaced by ∼ 10 kHz, and within those, peaks spaced by ∼1 kHz [Samara et al., 53]. These observations qualitatively conﬁrm the Yoon et al. [67] model of upper hybrid wave eigenmodes in a cylindrical density enhancement. To model these waves, consider the upper-hybrid wave equation (8.12) with the right-hand side ignored, d4 φ(x) ω 2 − ωp2 (x) − Ω 2 d2 φ vT2 ωp2 (x) ω 2 − ≈0. 2Ω 2 Ω 2 ω 2 − Ω 2 dx4 ω2 − Ω 2 dx2

(8.19)

Applying the density model (8.13) to the above equation, we obtain the following equations in dimensionless variables:

8 Mode Conversion Radiation

d2 φ 2 + kuh (X) φ = 0, dX 2 Q=

2 kuh (X) = Q

2 w2 − wn2 , ρ2 α2 w2

wn2 = 1 + α2 ,

X02 =

229

X02 − X 2 , 1 + δ + X2

wx2 − w2 , w2 − wn2

wx2 = 1 + (1 + δ) α2 ,

(8.20)

where

2 ωp0 vT ω x , ρ= , α2 = 2 , w = . (8.21) L LΩ Ω Ω 2 If w2 < 1+α2 , then kuh (X) > 0 everywhere, implying that only continuous 2 2 (X) < 0, and eigenmodes exist. If w > (1 + δ) α2 , on the other hand, then kuh only damped modes exist. Thus, the possibility of trapped eigenmodes exists only in the intermediate range wn2 < w2 < wx2 . One can show that discrete modes cannot exist if δ < 0 (i.e., density depletion), since in this case the internal region (X 2 < X02 ) is characterized by damped solutions while the outer region supports oscillatory solutions. Therefore, the situation is similar to that of discrete Langmuir waves, except that the discrete modes occur for the case of density enhancements rather than density depletions. Asserting the matching condition (8.15) across the cutoﬀ points results in an equation for the eigenfrequencies similar to (8.17),

1 π (2 δ)1/2 (w2 − 1)1/2 [K(ζ) − E(ζ)] = n + , ρα w 2 2

X=

ζ=

[1 + (1 + δ)α2 − w2 ]1/2 , δ 1/2 (w2 − 1)1/2

(8.22)

which can be approximately solved for w2 ,

1/2 1 a2n a2n 2 2 2 2 − an wx (wx − 1) + wx − , w = 1 − a2n 2 4 1/2 δ ρα . an = (2n + 1) 2

(8.23)

In Fig. 8.6 we plot the solution of (8.23) for α = 0.7746, δ = 4, ρ = 10−2 /L, and for two choices for the size of the density structures: L = 100m and L = 500 m. In addition to the above normalized input parameters, we take the electron cyclotron frequency to be 1 MHz, as is approximately the case in the auroral ionosphere. The resulting eigenmode spacings are of order 0.2 kHz for L = 100 m and 0.1 kHz for L = 500 m; both cases fall within the range of frequency spacings observed in ﬁne structures of auroral roar emissions [see Shepherd et al., 55]. These predictions are narrower than the principal band spacing, but not much narrower than the substructure spacing, observed in rocket data of structured upper hybrid waves [Samara et al., 53].

230

P.H. Yoon et al. L = 500 m 2000

1999.5

1999.9 f [kHz]

f [kHz]

L = 100 m 2000

1999 1998.5

1998

1999.8 1999.7 1999.6

0

2

4

6 n

8

10

0

2

4

6

8

10

n

Fig. 8.6. Discrete upper-hybrid wave spectrum f [kHz] versus the mode number n, for two diﬀerent density structure widths, L, corresponding to 100 m and 500 m

8.4 Conclusions As reviewed above, the terrestrial ionosphere and magnetosphere include numerous examples of mode-conversion radiation, which occurs when nonthermal particle distributions excite electrostatic waves which convert to EM waves. In the theory section above, we have presented linear analysis of the normal modes of high-frequency plasma waves in inhomogeneous plasma, where the inhomogeneity, either a density enhancement or depletion, can be described by a Lorentzian form. This generalized treatment contains both the Langmuir wave eigenmodes proposed by McAdams et al. [42] and the upper hybrid wave eigenmodes proposed by Yoon et al. [67]. The frequency spacings derived from these models, for inhomogeneities of scale size 100–1000 m, match to the observed spacings of frequency ﬁne structure observed in auroral Langmuir wave “chirps” [McAdams and LaBelle, 41] and auroral roar ﬁne structure [35, 55]; upper hybrid wave frequency ﬁne structure recently observed directly with a rocket experiment shows somewhat wider frequency spacings [Samara et al., 53]. The theoretical treatment outlined above is independent but does not exclude the possibility that other mechanisms may act to produce wave structure, such as for example the discretization upon wave conversion in inhomogeneous plasma discussed by Willes and Cairns [63]. Perhaps the most signiﬁcant aspect of these studies lies in the potential use of observations of mode conversion radiation to remotely sense the plasma environment of the source region, gaining information about electron densities, magnetic ﬁelds, and depth and spatial scale of irregularities. The solar system is replete with examples of the mode-conversion radiation exhibiting ﬁne frequency structures, ranging from nonthermal emissions of planetary ionospheres and magnetospheres to some types of radio emissions originating in the solar wind. To remotely sense terrestrial ionospheric and magnetospheric plasma using mode conversion radiation has intriguing implications

8 Mode Conversion Radiation

231

for interpretation of the emissions originating in more remote plasmas, where direct measurements are not available and parameters are less well-known.

Acknowledgments This research was supported by the following National Science Foundation grants: ATM-0223764 to the University of Maryland; OPP-0338105, OPP0341470 and ATM-0243645 to Siena College; and OPP-0090545 and ATM0243595 to Dartmouth College. Dartmouth College was further supported by NASA grant NNG04WC27G. We thank Doug Menietti for the Cluster-II data depicted in Fig. 8.4 and Shengyi Ye for the South Pole data shown in Fig. 8.3.

References [1] Bauer, S.J., and R.G. Stone: Satellite observations of radio noise in the magnetosphere, Nature 128, 1145, 1968. [2] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and HF auroral hiss generated at low altitudes, J. Geophys. Res. 94, 1359, 1989. [3] Benson, R.F., and H.K. Wong: Low-altitude ISIS 1 observations of auroral radio emissions and their signiﬁcance to the cyclotron maser instability, J. Geophys. Res. 92, 1218, 1987. [4] Cairns, I. H. and J.D. Menietti: Radiation near 2fp and intensiﬁed emissions near fp in the dayside and nightside auroral region and polar cap, J. Geophys. Res. 102, 4787, 1997 [5] Calvert, W.: A feedback model for the source of auroral kilometric radiation, J. Geophys. Res. 87, 8199, 1982. [6] Carlson, C.W. et al.: unpublished manuscript, 1987. [7] Carpenter, D.L., and R.R. Anderson: An ISEE/whistler model of equatorial electron density in the magnetosphere, J. Geophys. Res. 97, 1097, 1992. [8] Chiu, Y.T., and M. Schulz: Self-consistent particle and parallel electrostatic ﬁeld distributions in the magnetospheric – ionospheric auroral region, J. Geophys. Res. 83, 629, 1978. [9] Ergun, R.E., C.W. Carlson, J.P. McFadden, J.H. Clemmons, and M.H. Boehm: Evidence of transverse Langmuir modulational instability in a space plasma, Geophys. Res. Lett. 18, 1177, 1991a. [10] Ergun, R.E., C.W. Carlson, J.P. McFadden, J.H. Clemmons, and M.H. Boehm: Langmuir wave growth and electron bunching: Results from a wave-particle correlator, J. Geophys. Res. 96, 225, 1991b. [11] Filbert, P.C. and P.J. Kellogg: Electrostatic noise at the plasma frequency beyond the earth’s bow shock, J. Geophys. Res. 84, 1369, 1979. [12] Gough, M.P.: Nonthermal continuum emissions associated with electron injections: Remote plasmapause sounding, Planet. Space Sci. 30, 657, 1982. [13] Green, J.L., B.R. Sandel, S.F. Fung, D.L. Gallagher, and B.W. Reinisch: On the origin of kilometric continuum, J. Geophys. Res. 107, 1105,10.1029/2001JA000193, 2002.

232

P.H. Yoon et al.

[14] Green, J.L., Scott Boardsen, Shing F. Fung, H. Matsumoto, K. Hashimoto, R.R. Anderson, B.R. Sandel, and B.W. Reinisch: Association of kilometric continuum with plasmapsheric structures, J. Geophys. Res. 109, A03203, doi:10.1029/2003JA010093, 2004. [15] Gregory, P.C.: Radio emissions from auroral electron, Nature 221, 350, 1969. [16] Gurnett, D.A. and R.R. Shaw: Electromagnetic radiation trapped in the magnetosphere above the plasma frequency, J. Geophys. Res. 78, 8136, 1973. [17] Gurnett, D.A.: The Earth as a radio source: The nonthermal continuum, J. Geophys. Res. 80, 2751, 1975. [18] Gurnett, D.A., F.L. Scarf, W.S. Kurth, R.R. Shaw, and R.L. Poynter: Determination of Jupiter’s Electron Density Proﬁle from Plasma Wave Observations, J. Geophys. Res. 86, 8199-8212, 1981. [19] Hashimoto, K., W. Calvert, and H. Matsumoto: Kilometric continuum detected by Geotail, J. Geophys. Res. 104, 28645, 1999. [20] Horne, R.B.: Path-integrated growth of electrostatic waves: The generation of terrestrial myriametric radiation, J. Geophys. Res. 94, 8895, 1989. [21] Horne, R.B.: Narrowband structure and amplitude of terrestrial myriametric radiation, J. Geophys. Res. 95, 3925, 1990. [22] Hughes, J.M. and J. LaBelle: Plasma conditions in auroral roar source regions inferred from radio and radar observations, J. Geophys. Res. 106, 21157, 2001. [23] Issautier, K., N. Meyer-Vernet, M. Moncuquet, S. Hoang, and D.J. McComas: Quasi-thermal noise in a drifting plasma: theory and application to solar wind diagnostic on Ulysses, J. Geophys. Res. 104, 6691, 1999. [24] James, H.G., E.L. Hagg, and L.P. Strange: Narrowband radio noise in the topside ionosphere, AGARD Conf. Proc., AGARD-CP-138, 24, 1974. [25] Jones, D.: Source of terrestrial nonthermal radiation, Nature 260, 385, 1976. [26] Jones, D.: Latitudinal beaming of planetary radio emissions, Nature 288, 225, 1980. [27] Kasaba, Y., H. Matsumoto, K. Hashimoto, R.R. Anderson, J.-L. Bougeret, M.L. Kaiser, X.Y. Wu, and I. Nagano: Remote sensing of the plasmapause during substorms: Geotail observation of nonthermal continuum enhancement, J. Geophys. Res. 103, 20389, 1998. [28] Kaufmann, R. L.: Electrostatic wave growth: Secondary peaks in measured auroral electron distribution function, J. Geophys. Res. 85, 1713, 1980. [29] Kellogg, P.J. and S.J. Monson: Radio emissions from aurora, Geophys. Res. Lett. 6, 297, 1979. [30] Kellogg, P.J. and S.J. Monson: Further studies of auroral roar, Radio Sci. 19, 551, 1984. [31] Kletzing, C.A., S.R. Bounds, J. LaBelle and M. Samara: Observation of the reactive component of Langmuir wave phase-bunched electrons, Geophys. Res. Lett., in press, 2005. [32] Kurth, W.S., D.A. Gurnett, and R.R. Anderson: Escaping nonthermal continuum radiation, J. Geophys. Res. 86, 5519, 1981. [33] Kurth, W.S.: Detailed observations of the source of terrestrial narrowband electromagnetic radiation, Geophys. Res. Lett. 9, 1341, 1982. [34] LaBelle, J.: Radio Noise of Auroral Origin: 1968–1988, J. Atmos. Terr. Phys. 51, 197, 1989. [35] LaBelle, J., M.L. Trimpi, R. Brittain, and A.T. Weatherwax: Fine structure of auroral roar emissions, J. Geophys. Res. 100, 21953, 1995.

8 Mode Conversion Radiation

233

[36] LaBelle, J., S.G. Shepherd, and M.L. Trimpi: Observations of auroral medium frequency bursts, J. Geophys. Res. 102, 22221, 1997. [37] LaBelle, J. and R.A. Treumann: Auroral Radio Emissions, 1. Hisses, Roars, and Bursts, Space Sci. Rev. 99, 295, 2002. [38] Lund, E.J., J. LaBelle, and R.A. Treumann: On quasi-thermal ﬂuctuations near the plasma frequency in the outer plasmasphere: A case study, J. Geophys. Res. 99, 23651, 1994. [39] Lund, E.J., R.A. Treumann, and J. LaBelle: Quasi-thermal ﬂuctuations in a beam-plasma system, Phys. Plasmas 3, 1234, 1996. [40] McAdams, K.L., J. LaBelle, M.L. Trimpi, P.M. Kintner, and R.A. Arnoldy: Rocket observations of banded structure in waves near the Langmuir frequency in the auroral ionosphere, J. Geophys. Res. 104, 28109, 1999. [41] McAdams, K.L. and J. LaBelle: Narrowband structure in HF waves above the plasma frequency in the auroral ionosphere, Geophys. Res. Lett., 26, 1825, 1999. [42] McAdams, K.L., R.E. Ergun, and J. LaBelle: HF Chirps: Eigenmode trapping in density deletions, Geophys. Res. Lett. 27, 321, 2000. [43] McFadden, J.P., C.W. Carlson, and M.H. Boehm: High-frequency waves generated by auroral electrons, J. Geophys. Res. 91, 12079, 1986. [44] Melrose, D.B.: A theory for the nonthermal radio continua in the terrestrial and Jovian magnetospheres, J. Geophys. Res. 86, 30, 1981. [45] Menietti, J.D., R.R. Anderson, J.S. Pickett, and D.A. Gurnett: Near-source and remote observations of kilometric continuum radiation from multispacecraft observations, J. Geophys. Res. 108, 1393, doi:10.1029/2003JA009826, 2003. [46] Menietti, J.D., O. Santolik, J.S. Pickett, and D.A. Gurnett: High resolution observations of continuum radiation, Planet. Space Sci., submitted 2004. [47] Meyer-Vernet, N., and C. Perche: Tool kit for antennae and thermal noise near the plasma frequency, J. Geophys. Res. 94, 2405, 1989. [48] Meyer-Vernet, N.: On the thermal noise in an anisotropic plasma, J. Geophys. Res. 21, 397, 1994. [49] Osherovich, V. and J. Fainberg: Dependence of frequency of nonlinear cold plasma cylindrical oscillations, Phys. Plasmas 11, 2314, 2004. [50] Pottelette, R. and R.A. Treumann: Auroral acceleration and radiation, this volume, 2005. [51] R¨ onnmark, K.: Emission of myriametric radiation by coalescence of upper hybrid waves with low frequency waves, Anal. Geophys. 1, 187, 1983. [52] Samara, M., J. LaBelle, C.A. Kletzing, and S.R. Bounds: Rocket Measurements of Polarization of Auroral HF Waves, EOS Trans. Am. Geophys. Union, Fall Meeting, 2002. [53] Samara, M., J. LaBelle, C.A. Kletzing, and S.R. Bounds: Rocket observations of structured upper hybrid waves at fuh = 2fce , Geophys. Res. Lett. 31, L22804, 10.1029/2004GL021043, 2004. [54] Samara, M., J. LaBelle, I.H. Cairns and R.A. Treumann, Statistics of Auroral Langmuir Waves, J. Geophys. Res., submitted, 2005. [55] Shepherd, S.J., J. LaBelle, and M.L. Trimpi: Further investigation of auroral roar ﬁne structure, J. Geophys. Res. 103, 2219, 1998a. [56] Shepherd, S.J., J. LaBelle, and M.L. Trimpi: The polarization of auroral radio emissions, Geophys. Res. Lett. 24, 3161, 1998b.

234

P.H. Yoon et al.

[57] Steinberg, J.-L., S. Hoang, and M.F. Thomsen: Observations of the Earth’s continuum radiation in the distant magnetotail with ISEE-3, J. Geophys. Res. 95, 20781, 1990. [58] Walsh, D., F.T. Haddock, and H.F. Schulte: Cosmic radio intensities at 1.225 and 2.0 Mc measured up to an altitude of 1700 km, in: Space Res., 4, edited by P. Muller, pp. 935–959, North Holland Publishing Company, Amsterdam, 1964. [59] Weatherwax, A.T., J. LaBelle, M.L. Trimpi, and R. Brittain: Ground-based observations of radio emissions near 2fce and 3fce in the auroral zone, Geophys. Res. Lett. 20, 1447, 1993. [60] Weatherwax, A.T., J. LaBelle, and M.L. Trimpi: A new type of auroral radio emission observed at medium frequencies (∼1350–3700 kHz) using groundbased receivers, Geophys. Res. Lett. 21, 2753, 1994. [61] Weatherwax, A. T., J. LaBelle, M. L. Trimpi, R. A. Treumann, J. Minow, and C. Deehr: Statistical and case studies of radio emissions observed near 2fce and 3fce in the auroral zone, J. Geophys. Res. 100, 7745, 1995. [62] Weatherwax, A.T., P.H. Yoon, and J. LaBelle: Interpreting observations of MF/HF radio emissions: Unstable wave modes and possibilities to passively diagnose ionospheric densities, J. Geophys. Res. 107, 1213, doi:10.1029/2001JA000315, 2002. [63] Willes, A.J. and I.H. Cairns: Banded frequency structure from linear mode conversion in inhomogeneous plasmas, Phys. Plasmas 10, 4072, 2003. [64] Yoon, Peter H., A.T. Weatherwax, T.J. Rosenberg, and J. LaBelle, Lower ionospheric cyclotron maser theory: A possible source of 2fce and 3fce auroral radio emissions, J. Geophys. Res. 101, 27,015, 1996. [65] Yoon, P.H., A.T. Weatherwax, and T.J. Rosenberg: On the generation of auroral radio emissions at harmonics of the lower ionospheric electron cyclotron frequency: X, O, and Z mode maser calculations, J. Geophys. Res. 103, 4071, 1998a. [66] Yoon, P.H., A.T. Weatherwax, T.J. Rosenberg, J. LaBelle, and S.G. Shepherd: Propagation of medium frequency (1–4 MHz)auroral radio waves to the ground via the Z-mode radio window, J. Geophys. Res. 103, 29267, 1998b. [67] Yoon, P.H., A.T. Weatherwax, and J. LaBelle: Discrete electrostatic eigenmodes associated with ionospheric density structure: Generation of auroral roar ﬁne frequency structure, J. Geophys. Res. 105, 27580, 2000. [68] Yoon, P.H. and J. LaBelle, Discrete Langmuir waves in density structure, J. Geophys. Res. 110, A11308, doi:10.1029/2005JA011186, 2005.

9 Theoretical Studies of Plasma Wave Coupling: A New Approach D.-H. Lee1 , K. Kim2 , E.-H. Kim1 , and K.-S. Kim1 1

2

Department of Astronomy and Space Science, Kyung Hee University, Yongin, Kyunggi 449-701, Korea, [email protected] Department of Molecular Science and Technology, Ajou University, Suwon, Kyunggi 443-749, Korea

Abstract. New numerical and analytical methods are applied to wave coupling in inhomogeneous plasma. It is found that the X-mode feeds energy into the upper hybrid resonance at plasma inhomogeneities oriented perpendicular to the ambient magnetic ﬁeld. The results are consistent with previous studies using other methods. When a ﬁnite pressure is introduced, the upper hybrid waves are no longer stationary and start propagating. They can form cavity modes and emit a small fraction of O (or X) waves. Limitations of the present study are the neglect of collisions and plasma pressure eﬀects which might limit the growth of the upper hybrid waves; furthermore, this study concentrates on the case for which the density gradient is perpendicular to the magnetic ﬁeld, a condition that is valid near the equator.

Key words: Mode coupling, propagation in inhomogeneous plasma, resonant absorption of X mode waves, upper hybrid resonance, cavity modes

9.1 Introduction Plasma waves become often coupled owing to inhomogeneity in space. A certain mode is reﬂected at the cutoﬀ region, and changes the polarization at the crossover region. At resonances, one mode can be converted into the other resonant mode where wave energy is irreversibly transferred into the resonances. Since mode conversion is often associated with singular solutions, the subject of plasma wave coupling has diﬃculties in both analytical and numerical aspects. For instance, analytical solutions can provide only asymptotic approximations near the resonances or approximate global solutions by adopting the WKB method [4, 5, 16]. When wave equations are strongly coupled, it becomes diﬃcult to treat the coupled equations in an exact manner. Even numerical studies based on eigenmode analysis also meet similar diﬃculties in such coupled systems. Strictly speaking, no pure eigenmodes D.-H. Lee et al.: Theoretical Studies on Plasma Wave Coupling: A New Approach, Lect. Notes Phys. 687, 235–249 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

236

D.-H. Lee et al.

exist owing to the singular behavior in wave coupling, which should limit the application of the eigenmode analysis. To avoid singularities, damping is often introduced and expressed using a complex frequency ω = ωr + iγ. For ﬁnite γ, it is expected that all wave modes will decay via this damping decrement. It should be noted that γ represents an average value resulting from a Fourier transform over inﬁnite time. However, we have interests in ﬁnite time histories in reality and the damping rates among diﬀerent modes can be diﬀerent for a ﬁnite time period. Thus, it is still an important issue to further investigate the process of plasma wave coupling. In this study, we introduce new numerical and analytical techniques, which compensate for such shortcomings mentioned above. One technique is a time-dependent numerical model, which allows multi-ﬂuid components in an arbitrarily inhomogeneous 3-D plasma. This time-dependent model can be very useful in the coupled wave problem [Lee and Lysak 12] since the mode conversion is often associated with singular solutions, which provide only asymptotic approximations. The other technique is an analytical tool, which is called the invariant imbedding method (IIM) [3]. Using the invariant imbedding equations, we are able to calculate the reﬂection and transmission coeﬃcients, and the wave amplitudes for the propagation of arbitrary number of coupled waves in arbitrarily-inhomogeneous stratiﬁed media [10]. In this study, we will show the application of these numerical and analytical techniques to one of the plasma wave coupling problems: the coupling of ordinary (O), extraordinary (X) waves, and upper hybrid resonances (UHR). Linear mode conversion of O and X waves into upper hybrid (UH) waves is investigated numerically by adopting a time-dependent numerical model, and analytically by adopting IIM, respectively.

9.2 Numerical Model We study the wave coupling at the UHR region by using a 3-D multi-ﬂuid numerical model [9, 11]. Our approach diﬀers from previous studies in the sense that plasma waves are studied with time histories of electric and magnetic ﬁeld components. In a cold plasma, the linearized electron waves can be obtained by Maxwell equations, Ohm’s law and the equation of motion. For simplicity, we consider only the motion of electron ﬂuids and assume the cold plasma approximation in this work. ∇×E=−

∂B ∂t

1 ∂E c2 ∂t J = −en0 (x)ve

∇ × B = µo J +

(9.1) (9.2) (9.3)

9 Plasma Wave Coupling

237

Fig. 9.1. (a) The upper hybrid and plasma frequency proﬁles assumed in the model. The solid line represents the local upper hybrid frequency, ωuh (x). The dashed line represents the local plasma frequency, ωp (x). (b) The applied impulse at x = 0, which is assumed on the Ez (or Ey ) component

∂ve = −e(E + ve × B0 ) (9.4) ∂t where E, B, J, v and n represent electric and magnetic ﬁelds, current density, velocity and number density, respectively. In order to solve (9.1)–(9.4) as an initial-valued problem, the ﬁnite diﬀerence method is used in time and space. In our box model, a uniform magnetic ﬁeld B0 is assumed to be parallel to the z-axis and the density gradient is introduced along the x-axis. Details of this numerical model are referred to in Kim and Lee [9] and Kim et al. [11]. The size of box used in this calculation is 102 × 102 × 102 km3 . The proﬁle of plasma frequency (ωp ) and UH frequency (ωuh ) assumed in our model is given in Fig. 9.1a. Frequencies are normalized to the electron cyclotron frequency, ωce (= 6 × 103 rad/sec). Length is normalized to the radial distance, L = 100 km. The impulsive input is assumed on Ez and Ey for O wave and X wave initial inputs, respectively, at x = 0. Fundamental harmonic wave numbers are assumed in both y and z directions. The EM impulse used in the simulations is given by Fig. 9.1b. Time histories of electric and magnetic ﬁeld components at a line of grid points along the x direction are recorded. The boundaries are assumed to become perfect reﬂectors after the impulsive stimulus ends. Thus the total wave energy inside the box model remains constant in time, which enables us to easily examine the energy transfer among diﬀerent wave modes. We start the simulation with the impulsive input on the Ez (or Ey ) component to represent the O wave (or X wave) impulse. me

9.3 Numerical Results The power spectra of electric and magnetic ﬁeld components are obtained through the FFT at each grid point. Figure 9.2 shows the spectra, where the

238

D.-H. Lee et al.

Fig. 9.2. The power spectra of perturbation components E and B when the impulse of O waves is assumed

amplitudes are represented by the degree of brightness after they are scaled logarithmically. In Fig. 9.2, in the spectrum of Ex , one continuous band appears, which corresponds to the local electrostatic (ES) upper hybrid waves in Fig. 9.1a. The Ez component can eﬀectively represent the O wave since the wave vector is almost perpendicular to B0 by assuming relatively small ky and kz . The electromagnetic (EM) waves globally appear in all E and B components since they freely propagate inside the box when an impulsive input excites EM waves. The EM modes show a few cavity modes in Fig. 9.2 owing to the perfect reﬂecting boundary conditions. This feature is well conﬁrmed in the spectra of Bx , By and Bz which are purely EM wave components. To examine the coupling between ES and EM modes at the UHR, we select one point (x1 ) at a certain UH frequency marked by X in Fig. 9.2. Figure 9.3 shows time histories of the electric and magnetic ﬁelds at this resonant point which are obtained by applying the inverse Fourier transform at the given

9 Plasma Wave Coupling

239

Fig. 9.3. The time histories of perturbation components E and B (case of O wave incidence). The electric ﬁelds are normalized to an arbitrary value EA and the magnetic ﬁelds are normalized to BA = EA /c, respectively

frequency. Figure 9.3 indicates that the ES wave grows in time, while the other EM wave damped. Therefore it is found that the EM waves excited by the O wave impulse are mode-converted into UH waves. We now consider an EM impulse of X wave by assuming the impulse above on the Ey component. Figure 9.4 shows the power spectra of E and B. In the spectrum of Ex , one continuous band appears again, which corresponds to the local UH frequency. The spectral feature in Fig. 9.4 is similar to that of Fig. 9.2 except that relatively large power is found in both Ex and Ey compared to Ez , while Fig. 9.2 shows relatively large power in Ez . This diﬀerence arises because the Ez input would produce relatively large O wave power, while the Ey would produce more X wave power inside the box. Figure 9.5 shows the time histories at the resonant point (x2 ) marked by X in Fig. 9.4. Our results show the growth of the ES wave Ex and the decay

240

D.-H. Lee et al.

Fig. 9.4. The power spectra of perturbation components E and B (case of X wave incidence)

of the other EM wave (Bx , By , Bz , Ey and Ez ) at this UH frequency, which are pretty similar to the results of the O wave impulse.

9.4 Invariant Embedding Method In recent work of Kim et al. [10], a new invariant embedding theory was presented for studying the propagation of coupled waves in inhomogeneous stratiﬁed media. We consider N coupled waves propagating in a stratiﬁed medium, where physical parameters depend on only one coordinate. We take this coordinate as the z axis and assume the medium of thickness L lies in 0 ≤ z ≤ L. We also assume that all waves propagate in the xz plane. The x component of the wave vector, q, is then a constant. In a variety of problems, the wave equation of N coupled waves has the form ! d2 ψ dE −1 dψ E (z) + E(z)K 2 M(z) − q 2 I ψ = 0 , − 2 dz dz dz

(9.5)

9 Plasma Wave Coupling

241

Fig. 9.5. The time histories of perturbation components E and B (case of X wave incidence)

where ψ = (ψ1 , · · · , ψN )T is a vector wave function and E and M are N × N matrix functions. We assume that the waves are incident from the vacuum region where z > L and transmitted to another vacuum region where z < 0. I is a unit matrix and K is a diagonal matrix such that Kij = ki δij , where ki is the magnitude of the vacuum wave vector for the i-th wave. By assigning E(z) and M(z) suitably, (9.5) is able to describe various kinds of waves in a number of stratiﬁed media. A wide variety of mode conversion phenomena observed in space and laboratory plasmas can also be studied using this equation [4, 7, 13, 16]. We generalize (9.5) slightly, by replacing the vector wave function ψ by an N × N matrix wave function Ψ , the j-th column vector (Ψ1j , · · · , ΨN j )T of which represents the wave function when the incident wave consists only of the j-th wave. We are interested in the reﬂection and transmission coeﬃcient matrices r = r(L) and t = t(L). Let us introduce a matrix ! +z T exp i z+ dz E(z )P ,! z > z g(z, z ) = ˜ (9.6) z T exp −i z dz E(z )P , z < z

242

D.-H. Lee et al.

where T and T˜ are the time-ordering and anti-time-ordering operators, respectively. When applied to a product of matrices which are functions of z, T (T˜ ) arranges the matrices in the order of increasing (decreasing) z. For instance, T [E(z2 )E(z1 )] = E(z1 )E(z2 ), if z2 > z1 . The matrix P is a diagonal matrix satisfying Pij = pi δij and pi is the negative z component of the vacuum wave vector for the i-th wave. It is easy to prove that g(z, z ) satisﬁes the equations ∂ g(z, z ) = i sgn(z − z ) E(z)P g(z, z ), ∂z ∂ g(z, z ) = −i sgn(z − z ) g(z, z )E(z )P . ∂z

(9.7)

Using (9.6) and (9.7), the wave equation (9.5) is transformed to an integral equation i L Ψ (z, L) = g(z, L) − dz g(z, z ) 2 0 ! × E(z )P − P M(z ) − q 2 P −1 M(z ) + q 2 P −1 E −1 (z ) Ψ (z , L) . (9.8) We take a derivative of this equation with respect to L and obtain ∂Ψ (z, L) = iΨ (z, L)α(L) , ∂L

(9.9)

where 1 α(L) = E(L)P − Ψ (L, L) 2 ! × E(L)P − P M(L) − q 2 P −1 M(L) + q 2 P −1 E −1 (L) .

(9.10)

Taking now the derivative of Ψ (L, L) with respect to L, we obtain dΨ (L, L) = iE(L)P [r(L) − I] + iΨ (L, L)α(L) . dL

(9.11)

Since Ψ (L, L) = I + r(L), we ﬁnd the invariant embedding equation satisﬁed by r(L): dr i = i [r(L)E(L)P + E(L)P r(L)] − [r(L) + I] dL 2 ! × E(L)P − P M(L) − q 2 P −1 M(L) + q 2 P −1 E −1 (L) [r(L) + I] . (9.12) Similarly by setting z = 0 in (9.9), we ﬁnd the invariant embedding equation for t(L): dt i = it(L)E(L)P − t(L) dL 2 ! × E(L)P − P M(L) − q 2 P −1 M(L) + q 2 P −1 E −1 (L) [r(L) + I] . (9.13)

9 Plasma Wave Coupling

243

These equations are supplemented with the initial conditions, r(0) = 0 and t(0) = I. We solve the coupled diﬀerential equations (9.12) and (9.13) numerically using the initial conditions and obtain the reﬂection and transmission coeﬃcient matrices r and t as functions of L. The invariant embedding method can also be used in calculating the ﬁeld amplitude Ψ (z) inside the medium. Rewriting (9.9), we get i ∂Ψ (z, l) = iΨ (z, l)E(l)P − Ψ (z, l) ∂l 2 ! × E(l)P − P M(l) − q 2 P −1 M(l) + q 2 P −1 E −1 (l) [r(l) + I] . (9.14) For a given z (0 < z < L), the ﬁeld amplitude Ψ (z, L) is obtained by integrating this equation from l = z to l = L using the initial condition Ψ (z, z) = I + r(z).

9.5 Application of IEM to the Mode-Conversion of O and X Waves Equations (9.12), (9.13) and (9.14) are the starting point in our exact analysis of a variety of wave coupling and mode conversion phenomena. In the rest of this paper, we demonstrate the utility of our invariant embedding equations by applying them to the high frequency wave propagation and mode conversion in cold, magnetized plasmas. We assume that the plasma density varies only in the z direction and the external magnetic ﬁeld B0 (= (B0 sin θ, 0, B0 cos θ)) is directed parallel to the xz plane and makes an angle θ with the z axis. The cold plasma dielectric tensor, , for high frequency waves in the present geometry is written as 1 + (3 − 1 ) sin2 θ i2 cos θ (3 − 1 ) sin θ cos θ , (9.15) 1 i2 sin θ = −i2 cos θ 2 (3 − 1 ) sin θ cos θ −i2 sin θ 1 + (3 − 1 ) cos θ where 1 = 1 −

2 =

ωp2 (ω + iν) 2

ω (ω + iν) − ωc2 ωp2 ωc 2

ω (ω + iν) − ωc2

3 = 1 −

ωp2 . ω (ω + iν)

,

,

(9.16)

The constant ν is the phenomenological collision frequency. The spatial inhomogeneity of plasmas enters through the z dependence of the electron number density n.

244

D.-H. Lee et al.

For monochromatic waves of frequency ω, the wave equation satisﬁed by the electric ﬁeld in cold magnetized plasmas has the form ∇2 E +

ω2 ·E=0. c2

(9.17)

In this paper, we restrict our interest to the case where plane waves propagate parallel to the z axis. In this situation, we can eliminate Ez from (9.17) and obtain two coupled wave equations satisﬁed by Ex = Ex (z) and Ey = Ey (z), which turn out to have precisely the same form as (9.5) with q = 0 and

ω Ex ψ= , K = I, E = I, Ey c 1 M= 1 + (3 − 1 ) cos2 θ

1 3 i2 3 cos θ ! × . (9.18) −i2 3 cos θ 1 3 − (3 − 1 ) 1 + 22 sin2 θ We use (9.12), (9.13) and (9.18) in calculating the reﬂection and transmission coeﬃcients. In our notation, r11 (r21 ) is the reﬂection coeﬃcient when the incident wave is Ex (that is, linearly polarized in the x direction) and the reﬂected wave is Ex (Ey ). Similarly, r22 (r12 ) is the reﬂection coeﬃcient when the incident wave is Ey (that is, linearly polarized in the y direction) and the reﬂected wave is Ey (Ex ). Similar deﬁnitions are applied to the transmission coeﬃcients. The reﬂectances and transmittances are deﬁned by Rij = |rij |2 and Tij = |tij |2 . When the dielectric permittivities of the incident region and the transmitted region are the same, we can calculate the wave absorption by Aj ≡ 1 − R1j − R2j − T1j − T2j (j = 1, 2). If a mode conversion occurs, this quantity is nonzero even in the limit where the damping constant ν goes to zero. We will call Aj the mode conversion coeﬃcient. For a speciﬁc calculation, we assume that the electron density proﬁle is given by z 2 4 n(z) = 10 9 1 − + 1 m−3 (9.19) L where the plasma thickness L is equal to 105 m. The electromagnetic wave is incident from the vacuum region where z > L and transmitted to the vacuum region where z < 0. We also assume that the electron cyclotron frequency is equal to ωc = 6 × 103 rad/s. In Fig. 9.6, we plot the reﬂectances R11 , R12 (= R21 ), R22 , the transmittances T11 , T12 , T21 , T22 and the mode conversion coeﬃcients A1 , A2 as functions of the normalized frequency ωL/c, when a wave is incident perpendicularly on the stratiﬁed plasma. The external magnetic ﬁeld is directed at θ = 45◦ from the z axis. The resonance associated with the extraordinary wave component occurs for ωe1 ≤ ω ≤ ωe2 where ωe1 and ωe2 are the minimum and maximum values of

9 Plasma Wave Coupling

ω

245

ω

1

ω Fig. 9.6. Reﬂectances, transmittances and the mode conversion coeﬃcients when the external magnetic ﬁeld is directed at 45◦ from the z axis and the wave is incident parallel to the z axis. The mode conversion coeﬃcients are nonzero for ωe1 ≤ ω ≤ ωe2 , where ωe1 and ωe2 are the minimum and maximum values of ωe deﬁned by (9.20) and for ωo1 ≤ ω ≤ ωo2 , where ωo1 and ωo2 are the minimum and maximum values of ωo deﬁned by (9.21)

246

D.-H. Lee et al.

( ωe ≡

1 12 ωp2 + ωc2 4ωp2 ωc2 cos2 θ 2 . 1+ 1− 2 2 ωp2 + ωc2

(9.20)

Both A1 and A2 show a broad and sizable peak in this frequency range. The resonance associated with the ordinary wave component occurs for ωo1 ≤ ω ≤ ωo2 where ωo1 and ωo2 are the minimum and maximum values of ( ωo ≡

ωp2

+ 2

ωc2

12 12 2 2 2 4ωp ωc cos θ . 1− 1− 2 ωp2 + ωc2

(9.21)

A1 and A2 show a tiny peak in this frequency range. In Fig. 9.6, it is evident that ES waves absorb EM wave energy via the resonant absorption at (9.20) and (9.21). These ES waves become the UHR when θ ≈ π/2 and the plasma oscillations θ ≈ 0, respectively. Thus A1 (A2 ) denotes the mode conversion coeﬃcient from linearly polarized EM waves of Ex (Ey ) to the ES resonant modes. It should be noted that these absorptions are not aﬀected by the boundaries between plasma and vacuum since both A1 and A2 disappear when inhomogeneity is excluded and uniform plasmas are assumed in the same region. In Fig. 9.7, we plot the dependence of the mode conversion coeﬃcients on the angle between the external magnetic ﬁeld and the z axis. Both for A1 and A2 , the ordinary wave absorption, the magnitude of which is quite small, is largest when θ = 0 and decreases monotonically as θ increases. The frequency ranges where this absorption occurs agree quite well with ωo1 and ωo2 given by (9.21). The extraordinary wave absorption for A2 is zero when θ = 0 and increases monotonically as θ increases. The frequency ranges where this absorption occurs are designated by square dots, the positions of which agree precisely with ωe1 and ωe2 given by (9.20). For A1 , the extraordinary wave absorption is zero for θ = 90◦ , as well as for θ = 0. In Fig. 9.8, we plot the mode conversion coeﬃcients for two diﬀerent normalized damping coeﬃcients νL/c = 0.01 and 0.0001 when θ = 45◦ .

9.6 Discussion and Summary Our results of both numerical and analytical methods presented above show that both O and X waves are likely to give energy into the UHR when the inhomogeneity lies perpendicular to the ambient magnetic ﬁeld. It is suggested that the EM electron waves are mode-converted to the ES electron waves via the resonant absorption at the UH frequency. Our simulation results are found to be also consistent with previous studies such as investigation of the Budden problem by White and Chen [19], Grebogi et al. [6], Lin and Lin [14], Antani et al. [1, 2], and Ueda et al. [18].

9 Plasma Wave Coupling

θ=0

θ=0

θ=15

θ=15

θ=30

θ=30

θ=45

θ=45

θ=60

θ=60

θ=75

θ=75

θ=90

θ=90

ω

247

ω

Fig. 9.7. Dependence of the mode conversion coeﬃcients on the angle between the external magnetic ﬁeld and the z axis. Square dots represent the positions of ωe1 and ωe2

However, there are several limitations of our calculations. We neglect the eﬀects of collisions and plasma pressure which might limit the growth of UHR amplitude [White and Chen, 19]. In addition, when the ﬁnite pressure is introduced, the ES upper hybrid waves are able to propagate and no longer stationary. Under certain circumstances such as the density striations these ES waves can form the cavity modes and emit a small fraction of O (or X) waves into space [8, 15, 17, 20]. In this study the density gradient is assumed to be perpendicular to the background magnetic ﬁeld. This condition is valid probably only at the equatorial region and it should be extended to the case of arbitrary angles with respect to the Earth’s magnetic ﬁeld in the high-latitude region. Recent

248

D.-H. Lee et al.

ν

ν

ν

ν

ω

ω

Fig. 9.8. Dependence of the mode conversion coeﬃcients on the wave damping parameter ν

analytical methods such as the invariant embedding method of Kim et al. [10] as well as our time-dependent 3-D simulations in Kim et al. [11] enable us to quantitatively investigate such general wave coupling problems. This subject will be left as future work.

Acknowledgments This work was supported by the Korea Science and Engineering Foundation grant R14-2002-043-01000-0 and in part by R14-2002-062-01000-0.

References [1] Antani, S.N., N.N. Rao, and D.J. Kaup: Geophys. Res. Lett. 18, 2285 (1991). [2] Antani, S.N., D.J. Kaup, and N.N. Rao: J. Geophys. Res. 101, 27,035 (1996). [3] Bellman, R. and G.M. Wing: An Introduction to Invariant Imbedding (Wiley, New York, 1976). [4] Budden, K.G.: The Propagation of Radio Waves (Cambridge, Cambridge, 1985). [5] Ginzburg, V.L.: Propagation of electromagnetic waves in plasmas (Pergamon Press, New York, 1964). [6] Grebogi, C., C.S. Liu, and V.K. Tripathi: Phys. Rev. Lett. 39, 338 (1977). [7] Hinkel-Lipsker, D.E., B.D. Fried, and G.J. Morales: Phys. Fluids B 4, 559 (1992). [8] Hughes, J.M. and J. LaBelle: Geophys. Res. Lett, 28, 123 (2001).

9 Plasma Wave Coupling

249

[9] Kim, E.-H. and D.-H. Lee: Geophys. Res. Lett. 30, 2240 (2003). [10] Kim, K., D.-H. Lee, and H. Lim: Europhys. Lett. 69, to appear (2005). [11] Kim, K.-S., D.-H. Lee, E.-H. Kim, and K. Kim: Geophys. Res. Lett., submitted (2005) [12] Lee, D.-H. and R.L. Lysak: J. Geophys. Res. 94, 17,097 (1989). [13] Lee, D.-H., M.K. Hudson, K. Kim, R.L. Lysak, and Y. Song: J. Geophys. Res. 107, 1307 (2002). [14] Lin, A.T. and C.C. Lin: Phys. Fluids 27, 2208 (1984). [15] Shepherd, S.G., J. LaBelle, and M.L. Trimpi: Geophys. Res. Lett. 24, 3161 (1997). [16] Swanson, D.G.: Theory of Mode Conversion and Tunneling in Inhomogeneous Plasmas (Wiley, New York, 1998). [17] Weatherwax, A.T., J. LaBelle, M.L. Trimpi, R.A. Treumann, J. Minow, and C. Deehr: J. Geophys. Res. 100, 7745 (1995). [18] Ueda, H.O., Y. Omura, and H. Matsumoto: Ann. Geophysicae 16, 1251 (1998). [19] White, R.B. and F.F. Chen: Plasma Phys. 16, 565 (1974). [20] Yoon, P.H., A.T., Weatherwax, and T.J. Rosenberg: J. Geophys. Res. 103, 4071 (1998).

10 Plasma Waves Near Reconnection Sites A. Vaivads1 , Yu. Khotyaintsev1 , M. Andr´e1 , and R.A. Treumann2,3 1

2

3

Swedish Institute of Space Physics, Uppsala, Sweden [email protected], [email protected], [email protected] Geophysics Section, The University of Munich, Munich, Germany [email protected] Department of Physics and Astronomy, Dartmouth College, Hanover, NH, USA

Abstract. Reconnection sites are known to be regions of strong wave activity covering a broad range of frequencies from below the ion gyrofrequency to above the electron plasma frequency. Here we explore the observations near the reconnection sites of high frequency waves, frequencies well above the ion gyrofrequency. We concentrate on in situ satellite observations, particularly on recent observations by the Cluster spacecraft and, where possible, compare the observations with numerical simulations, laboratory experiments and theoretical predictions. Several wave modes are found near the reconnection sites: lower hybrid drift waves, whistlers, electron cyclotron waves, Langmuir/upper hybrid waves, and solitary wave structures. We discuss the role of these waves in the reconnection onset and supporting the reconnection, in anomalous resistivity and diﬀusion, as well as a possibility for using these waves as a tool for remote sensing of reconnection sites.

Key words: Reconnection, Hall region, separatrix physics, lower hybrid waves, whistlers, high-frequency waves, electron holes, anomalous resistance, structure of reconnection region, wave signatures

10.1 Background Collisionless magnetic reconnection in space plasma has the two important properties of converting the available free magnetic energy into kinetic energy of charged particles in large regions of space and causing signiﬁcant mass and energy transfer across the boundaries that separate the interacting plasmas. The regions where the energy conversion takes place, e.g. the auroral zone, ionosphere, shocks, etc., emit waves and generate turbulence over a wide frequency range. The occurrence of reconnection is not exceptional. To the contrary, reconnection is abundant in collisionless plasmas taking place everywhere where suﬃciently thin current sheets are generated. Understanding the role of waves and turbulence in the energy conversion, energy transA. Vaivads et al.: Plasma Waves Near Reconnection Sites, Lect. Notes Phys. 687, 251–269 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

252

A. Vaivads et al.

port, and structure formation of the reconnection sites is thus an important and challenging task. Astrophysical environments allow studies of reconnection regions only from remote by observing the emitted electromagnetic radiation. For instance, the electron beams which cause solar and interplanetary type III radio bursts at the local plasma frequency and its harmonics are believed to originate from regions near reconnection sites in the solar corona [see, e.g., Cane et al., 9]. Remote studies of the emission provide solely average information about the reconnection sites with the averaging proceeding over large spatial volumes. They therefore suﬀer from severe limitations on the spatial resolution. As a consequence the information obtained about local conditions and micro processes in the reconnection region is very limited and in most cases no information can be extracted at all. The only places where reconnection sites can be studied in detail are in the laboratory and the Earth’s magnetosphere or other accessible environments in our solar system that have been visited by spacecraft, such as the solar wind, some of the other magnetized planets, comets, and the outer heliosphere. Spacecraft observations give a much more detailed picture of the plasma dynamics than any of the laboratory experiments. This is due mainly to the possibility of resolving the particle distribution functions and electromagnetic ﬁelds at small scales, in some cases down to the smallest electron scales. In the laboratory, in addition, it is practically impossible to reproduce the collisionless and dilute conditions at the temperatures prevailing in space and astrophysical plasmas. Observations in space are thus uniquely suited for the understanding of reconnection. However, since reconnection involves many processes at diﬀerent spatial and temporal scales, numerical simulations serve as a superior tool for understanding the environment and physical processes in the vicinity of reconnection sites. The two main regions in the Earth’s magnetosphere where the processes of reconnection have so far been studied are the subsolar magnetopause and the magnetotail current sheet. Under normal conditions in those regions the plasma is overdense, with fpe fce . Reconnection in the magnetotail proceeds in a relatively symmetric way, in the sense that the plasmas to both sides of the tail current sheet have similar properties. Quite an opposite situation is encountered at the magnetopause. Here reconnection is manifestly asymmetric. Another important diﬀerence between magnetopause and tail reconnection is that the typical spatial scales, e.g. the ion inertial length and the ion gyro radius, are usually a factor of ten smaller at the magnetopause than in the tail. This is important for any in situ studies where the instrumental resolution enters as a limiting factor. Altogether, a signiﬁcant number of studies deal with high frequency waves at the magnetopause and in the magnetotail, but only in a few cases have attempts been made to ﬁnd a direct relation between the observed waves and the reconnection process, even though in some cases the existence of such a relationship has been put forth. In the present paper we summarize in situ observations of high frequency waves under conditions when reconnection signatures are well deﬁned as well

10 Plasma Waves Near Reconnection Sites

253

as observations where one merely speculates about a relation of the observed waves to possible reconnection going on at a distance from the location of the observations. Figure 10.1 shows a sketch of the reconnection site. Two oppositely directed magnetic ﬁelds in the inﬂow regions merge in the diﬀusion region forming an X-line. The magnetic ﬁeld lines connected to the X-line are the separatrices. We call the regions close to the separatrices separatrix regions. Plasma containing reconnected magnetic ﬁelds is ﬂowing away from the X-line in the outﬂow regions, where it escapes in the form of jets. This sketch draws a rather simpliﬁed two-dimensional picture of the reconnection process. In reality, the reconnection site has a considerably more complicated structure, consisting of multiple X-lines and exhibiting a complex and three-dimensional conﬁguration. However, in many cases the simple 2D picture may serve as a lucid and suﬃcient approximation to a reconnection site. A counterexample is the complex structure arising from spontaneous antiparallel reconnection where patchy X-lines form within a narrow current sheet. On the other hand, when a small guide ﬁeld is added, well ordered X-lines develop, and the 2D description can become suﬃciently accurate to serve as an approximate description of the reconnection process [see, e.g., Scholer et al., 33]. Another general property of a single reconnection site is its pronounced inhomogeneity rendering almost all homogeneous plasma models of scales of the order of the spatial extension of the reconnection site invalid. Density and temperature gradients as well as non-Maxwellian particle distribution functions in the vicinity of a reconnection site result in the generation of various plasma wave modes. All of them contribute to the processes of particle acceleration and energy redistribution, generation of transport coeﬃcients, and the wanted diﬀusion of magnetic ﬁeld and plasma near the X-line which is necessary in order to maintain the merging of the oppositely directed magnetic ﬁelds. Space observations allow for the direct in situ observation of the wavegeneration and wave-particle interactions in these merging processes at the reconnection sites. 10.1.1 Observations of Diﬀerent Wave Modes Some of the high frequency wave modes which are usually observed in the vicinity of a reconnection site are sketched in Fig. 10.1. The locations where these modes are observed are rather speculative because there is very limited knowledge so far on the relative locations of the diﬀerent modes. Emissions with the strongest electric ﬁelds are ordinarily detected near or below the lower hybrid frequency fLH , indicating the presence of lower hybrid drift waves (LHD). These emissions have both strong electric and magnetic ﬁeld components. They seem to assume their highest amplitudes just near the steep density gradients like the ones found in the separatrix region. In general the separatrix region seems to be the location of very strong wave emissions of several diﬀerent wave modes in a wide frequency range.

254

A. Vaivads et al.

Fig. 10.1. Sketch of the reconnection site. Separatrices with density gradients are marked. A the top diﬀerent kinds of wave spectra are sketched that are commonly observed near reconnection sites. Some common places to observe those waves are marked in diﬀerent gray shadowing. Typical electron distribution functions in the vicinity of the separatrix are indicated as well

Strong electric ﬁelds are found around the electron plasma frequency fpe . These emissions are usually believed to be Langmuir (L) waves or, if oblique, upper hybrid (UH) waves. Often, electrostatic solitary waves (ESW) can also be associated with reconnection. Whistler emissions (W) are identiﬁed from narrow spectral peaks in the frequency range fLH < f < fce between the lower hybrid frequency and the local electron gyro-frequency fce . Despite the applicability of such waves to astrophysical environments, there have been very few observations of radio emissions from reconnection sites at frequencies around and above the electron plasma frequency even though one expects their presence in the separatrix region where they should be generated by the fast electron beams escaping from the X line. This is probably due to the weakness of the electromagnetic signals in comparison with the electrostatic waves. In the rest of this article we summarize the in situ observations of the above mentioned wave modes and discuss their possible generation mechanisms and relations to reconnection. Before doing so we also summarize some of the results obtained from numerical simulations dealing with high frequency waves or being relevant to the discussion in this paper.

10 Plasma Waves Near Reconnection Sites

255

10.2 Numerical Simulations The generation of high frequency waves involves the dynamics of electron and electron kinetic eﬀects which in most cases are very important. Therefore, fullparticle and Vlasov simulations are best qualiﬁed for this kind of study. Such simulations are also computationally expensive. So far relatively few wave modes have been addressed. An additional diﬃculty is that in some cases, for example in the simulation of lower hybrid drift waves, it is absolutely essential to simulate the phenomena in all three spatial dimensions. The available numerical simulations indicate that the separatrix regions are the most dynamically active regions [see, e.g., Cattell et al., 10]. Intense electron beams are generated in the reconnection process along the separatrices by parallel electric ﬁelds which are distributed along the separatrices as shown by Pritchett [30] and Hoshino et al. [17]. In addition, electron conics and shell distributions can form due to diverging magnetic ﬂux tubes close to the reconnection site. High-frequency wave modes that have been studied in detail by numerical simulations are the following: • Lower hybrid drift waves (LHD). It has been suggested that these waves play a crucial role in the narrowing of the current sheets with the subsequent onset of reconnection [see, e.g., 11, 33, 35]. LHD waves tend to be electrostatic (δE/δB > VA ). Their largest amplitudes are found at the edges of the current sheet. They interact eﬃciently with both electrons and ions and can cause a signiﬁcant anomalous resistivity and corresponding anomalous diﬀusion. It has been realized that these waves possess a signiﬁcant magnetic component in the center of the current sheet that also contributes to anomalous resistivity [Silin et al., 35]. • Solitary waves (SW) and double layers (DL). Solitary wave and double layer generation due to electron beams have been studied in great detail [e.g., Omura et al., 29]. Electron beams form mainly close to the separatrices as shown by Hoshino et al. [17] and Pritchett and Coroniti [31] and particularly under guide ﬁeld conditions. In this case strong double layers can be generated at the reconnection site [Drake et al., 13]. Solitary structures and lower hybrid waves may couple as well, as has been found in the same simulations. High frequency waves that have been studied in numerical simulations much less frequently. The main types investigated are: • Whistlers. Numerical simulations show that the Hall term in the Generalized Ohm’s law is important for the onset of fast reconnection [Birn et al., 8]. The Hall term also introduces whistlers into the system, and it has been speculated that the region close to the reconnection site is some kind of a standing whistler. Observations indicate that electromagnetic modes in the whistler frequency range are observed also at distances far from the reconnection site, but this type of emission has received very little attention in numerical studies.

256

A. Vaivads et al.

• Waves at the plasma frequency. While observations indicate the presence of strong Langmuir/upper hybrid waves in the separatrix regions, these waves have not been studied in any of the reconnection simulations. • Electron cyclotron waves. Similarly, observations indicate the presence of electron cyclotron waves and speculate about their relation to reconnection, but in the numerical simulations they have not been studied yet. • Radio emissions. Free-space modes are usually generated at the local plasma frequency or above. They can freely propagate out into space. They are of primary importance in the astrophysical application of reconnection. There exists a large theoretical eﬀort in studying these waves but the amount of numerical simulations is very limited and has mainly been done for astrophysical plasma conditions.

10.3 Lower Hybrid Drift Waves 10.3.1 Observations Large wave-electric ﬁelds at the magnetopause and in the magnetotail are usually observed at frequencies near the local lower-hybrid frequency fLH [3, ˜ ≥ 1, where 10]. The strongest peak-to-peak amplitudes are of the order δ E ˜ δ E is the normalized root mean square electric ﬁeld. At the magnetopause the waves are strongest on the magnetospheric side where the inﬂow Alfv´enic speed is high. Often, these waves are located in those regions (narrow sheets with spatial scale less than an ion gyroradius) where the DC electric ﬁeld reaches high values, about half of the peak-to-peak amplitude of the waves. At this time there is no statistical analysis available of the occurrence rates of these waves, but it seems that the highest-amplitude lower-hybrid waves are located in the regions of steepest density gradients. Numerical simulations and analytical calculations suggest that the observed waves are lower-hybrid drift (LHD) waves even though the modiﬁed two-stream instability could generate waves with similar properties. An example of LHD wave observations is shown in Fig. 10.2. Particular characteristics of LHD waves are their short wavelengths, kρe ∼ 1, a broadband spectrum extending from frequencies well below to well above fLH , perpendicular wave numbers k⊥ k , a phase velocity of the order of the ion thermal velocity, and a coherence length of the order of one wavelength. The observations also suggest that the wave potential can be close to the electron thermal energy [Bale et al., 6]. Based on spectral and interferometric results, some satellite observations support the LHD-interpretation [Vaivads et al., 38]. However, more detailed studies are required in order to conﬁrm the lower-hybrid drift nature of the observed waves. Observations also show that these waves have a signiﬁcant magnetic component with c δE/δB vA and can have a preferential direction of propagation along the ambient magnetic ﬁeld [Vaivads et al., 38].

10 Plasma Waves Near Reconnection Sites

257

Fig. 10.2. Example of LHD wave observations near the separatrices far from the reconnection X-line. The ﬁgure is adopted from Vaivads et al. [38] and Andr´e et al. [4]. (a), (b): Large-scale density (obtained from the satellite potential) and magnetic ﬁeld observations. The rest of the panels are zoomed-in on the time interval marked bright when the strongest electric ﬁelds were observed. This time is identiﬁed as a possible separatrix region [Andr´e et al., 4]. (c): Electron spectrogram showing electron beams along B. (d): Electric ﬁeld normal and tangential to the magnetopause. (e): Plasma density. (f): Magnetic ﬁeld spectra, and (g): Electric ﬁeld spectra (Reprinted with permission of American Geophysical Union)

10.3.2 Generation Mechanisms In a simpliﬁed picture, the driving force for the LHD waves is a density gradient with relative ﬂow between electrons and ions due to their diﬀerent diamagnetic drifts. In the case of the modiﬁed-two stream instability (MTSI) it is the cross-ﬁeld drift of the electrons with respect to the ions that presents the driving force. Strong DC electric ﬁelds on scales smaller than ion gyroradius are almost always seen in association with LHD waves. From the energetic point of view it has been shown that the observed waves can also be generated

258

A. Vaivads et al.

by the electron beams present at the density gradients [Vaivads et al., 37]. It is well-known that in places like the auroral zone electron beams do indeed generate intense lower-hybrid waves. 10.3.3 Relation to Reconnection Reconnection is one of the mechanisms that is capable of producing strong and narrow density gradients in space. Density gradients are formed along the separatrices which separate the inﬂow from the outﬂow regions. The separatrices contain local density dips which have been observed in numerical simulations [see, e.g., 34] . Numerical simulations and observations also suggest that the separatrices can maintain their steep density gradient structure over distances very far away from the reconnection site (tens of λi ). It has not yet been explained properly why these density dips exist. Such dips can, however, evolve when a static ﬁeld-aligned potential is applied along the separatrix which evacuates part of the plasma locally. The separatrices are also regions where strong electron beams are present [Hoshino et al., 17]. Such beams have a longitudinal pressure anisotropy and therefore are hardly capable of excluding plasma from the separatrix regions. A probable cause is a magnetic-ﬁeld-aligned electric ﬁeld which accelerates the electrons into a beam thereby evaporating the plasma. Nevertheless, it remains unclear which mechanism maintains the required pressure balance. The role of these beams in the generation of waves has not been fully explored. It is not clear, moreover, which other mechanisms besides reconnection could produce such narrow (few λi ) density gradients. There are speculative ideas, like “peeling” or “snow-plowing” due to, e.g., FTEs, but no clear understanding exists as yet of how such a process would work. Thus, while we expect strong LHD waves near the reconnection site along the separatrices it is not yet clear whether all or most of the intense LHD-wave observations are related to ongoing reconnection. Lower hybrid waves can aﬀect reconnection in several important ways: • Through anomalous resistivity: Usually, in simulations and observations the most intense LHD waves generate anomalous collision frequencies for e− ∼ 2πfLH [Silin et al., 35]. electrons of the order of νan • Through electron acceleration: While the phase velocity of lower hybrid waves in the direction perpendicular to the magnetic ﬁeld B is comparable to the ion thermal speed, the phase velocity along the magnetic ﬁeld becomes comparable to the electron thermal velocity thus enabling the LHD waves to resonate with thermal electrons and eﬃciently accelerate these electrons. For the same reason electron beams can be eﬃcient in generating lower hybrid waves by the inverse resonance process. • Through current sheet bifurcation, thinning and reconnection onset: Numerical simulations have shown that LHD waves apparently play a crucial role in the reconnection onset within thin current sheets [11, 33].

10 Plasma Waves Near Reconnection Sites

259

They evolve due to the steep plasma gradients at the current sheet boundary and in fact tend to broaden the current sheet. When propagating into the sheet they contribute to anomalous resistivity, heating and the diﬀusivity necessary for reconnection. This issue is controversial and has not been settled.

10.4 Solitary Waves and Langmuir/Upper Hybrid Waves 10.4.1 Observations Other quasi-stationary structures containing large electric ﬁelds that have been observed close to the reconnection sites are electrostatic solitary waves (ESW), broad-band electrostatic noise (BEN) and Langmuir/upper hybrid waves [e.g., Deng et al., 12]. They all tend to appear approximately in the same region and have similar amplitudes, which are usually several or many times smaller than the amplitudes of the LHD waves. Part of BEN observations are due to ESW passing over the spacecraft thus giving rise to a broadband spectrum. Observations show that the strongest emissions are observed along the separatrices [10, 14, 38]. The emissions change their character rapidly but it seems that narrow-band emissions at the Langmuir frequency and broadband emissions do not appear simultaneously [Khotyaintsev et al., 22]. So far only a rough comparison has been possible between the electron distribution functions and the wave characteristics [Deng et al., 12]. In the magnetotail these waves are usually located at the boundary between lobe and plasmasheet. It has been suggested that they are related to the reconnection process [Kojima et al., 24]. In order to distinguish Langmuir waves from upper hybrid waves one must study the polarization of the waves. It has been found that near the electron plasma frequency the wave electric ﬁelds are often polarized at large angles with respect to the ambient magnetic ﬁeld indicating that the observed waves are upper hybrid waves [12, 14, 21, 23] rather than Langmuir waves. Examples of observations are shown in Fig. 10.3. 10.4.2 Generation Mechanisms It is known from numerical simulations, [e.g. Omura et al., 29], that Langmuir modes are usually driven by the weak-beam instability. ESW can be the result of saturation of the nonlinear bump-on-tail instability or the two-stream instability. Upper hybrid waves can also be generated by beams, however they can also be generated by loss cone and shell distributions. Such distributions form preferentially in diverging magnetic ﬁelds either close to the reconnection site or when magnetic ﬂux tubes approach the Earth.

260

A. Vaivads et al.

Fig. 10.3. Example of wave observations near the separatrices of a reconnection X-line. (a, b) Monochromatic waves which can be interpreted as Langmuir waves when E E⊥ and as upper hybrid waves when E⊥ E . (c) Example of a mixture between electrostatic solitary waves and Langmuir waves. (d) Electrostatic solitary waves [ﬁgure adapted from Deng et al., 12]

10.4.3 Relation to Reconnection One of the major questions of reconnection is how parallel electric ﬁelds are distributed near the reconnection site. These ﬁelds are required to create changes in the magnetic ﬁeld-line topology that is associated with reconnection. ESW with a net potential drop can be one source for parallel electric ﬁelds. At the same time ESW and L/UH waves interact eﬃciently with electrons and generate high energy tails on the electron distribution functions. ESW also contribute to the anomalous resistivity in ﬁeld-aligned currents.

10 Plasma Waves Near Reconnection Sites

261

10.5 Whistlers 10.5.1 Observations There are many observations related to whistler emission in regions that are directly related to reconnection, such as the plasma sheet boundary layer [see, e.g., Gurnett et al., 16] and the magnetopause [LaBelle and Treumann, 26] and more recently [Stenberg et al., 36]. Whistler wave modes can be identiﬁed on the basis of the observed frequency of the waves (in between the electron and ion gyro-frequencies), the presence of a strong magnetic component that can reach ∼0.1 nT [Gurnett et al., 16], or the direct measurement of wave polarization [Zhang et al., 39]. Narrow spectral peaks in a wide region of frequencies between the lower hybrid and electron cyclotron frequencies are typical for these waves. However, broadband spectra extending over whistler frequencies are also often observed [LaBelle and Treumann, 26]. Such waves can consist of whistlers or they can be associated with the magnetic component of lower hybrid drift waves. Figure 10.4 gives an example of wave measurements in the high-altitude magnetopause/cusp region showing indications that the reconnection process proceeds in a high-beta plasma [Khotyaintsev et al., 22]. 10.5.2 Generation Mechanisms In addition to the temperature anisotropy or loss-cone instabilities by which it is conventionally known, whistlers can be excited as a consequence of the perpendicular anisotropy of the electron distribution function, or ion beams [Akimoto et al., 1]. Electron beams [Zhang et al., 39] have also been suggested as possible generators for whistlers under conditions when the loss cone is absent and the plasma is isotropic. This becomes possible since electrons or ions of suﬃciently high speed can undergo resonance with whistlers which not only Landau damps but, under certain circumstances, also excites whistlers. In this spirit it has been shown that whistler waves in the magnetotail are most probably generated by electron beams through the cyclotron resonance and not through the temperature anisotropy or other non-beam instability mechanisms [Zhang et al., 39]. The estimated resonance energy of the electron beam is about 10 keV suggesting reconnection as the most probable source of the beam. The instability mechanism producing broad-band magnetic turbulence is less clear. Laboratory experiments are consistent with the modiﬁed two-stream instability producing the emissions, since the waves propagate in the same direction of the electron ﬂow [Ji et al., 20]. Part of these emissions can be associated with the magnetic component of lower hybrid drift waves. 10.5.3 Relation to Reconnection The most direct evidence that whistlers can play a crucial role in the reconnection process comes from laboratory experiments which show that the reconnection rate correlates with the amplitude of obliquely propagating broad-band

262

A. Vaivads et al.

Fig. 10.4. Example of high frequency wave observations in the high-altitude magnetopause/cusp [adopted from Khotyaintsev et al., 22]. (a, b): The magnetic ﬁeld magnitude and components, (c): convection velocity E × B, (d): plasma beta, (e): spectra of the electric ﬁeld in the 2–80 kHz range, (f): spectra of the magnetic ﬁeld in the 8–4000 Hz range, and electron-cyclotron frequency, (g): polarization of wave magnetic ﬁeld with respect to the ambient magnetic ﬁeld. Whistlers are identiﬁed from right-hand polarization

10 Plasma Waves Near Reconnection Sites

263

whistler waves inside the reconnecting current sheet [cf., Ji et al., 20]. Also, numerical simulations suggest that the magnetic ﬂuctuations in the whistler band can cause a signiﬁcant anomalous resistivity [Silin et al., 35]. Such observations are yet to be conﬁrmed by the observations of reconnection in space. So far, space observations indicate that whistlers are associated with the reconnection processes [Stenberg et al., 36] and most probably are related to electron beams generated during reconnection [Zhang et al., 39], however much more detailed studies are required to conﬁrm that the whistlers are not related secondarily but play an essential role in the process of reconnection itself. An important aspect of whistlers is their ability to propagate over large distances away from the reconnection site without appreciable damping. This property makes whistlers a perfect tool for use in remote sensing of reconnection sites. In addition, whistlers by this particular property may transport information from the reconnection site to other places in the plasma.

10.6 Electron Cyclotron Waves 10.6.1 Observations Electron cyclotron waves are commonly observed in the inner regions of the magnetosphere, i.e. the polar cusp, auroral zone, and plasmasphere. The observations from the outer magnetosphere are not as abundant. Electron cyclotron waves have been observed in association with ﬂux transfer events [2, 25] and the emerging of energetic plasma in the magnetotail [Gurnett et al., 16]. Both electrostatic and electromagnetic cyclotron waves have been observed at the magnetopause (Anderson et al. 1982). Observations in the cusp and close to the magnetopause indicate that electron cyclotron waves tend to be generated on open ﬁeld lines [Menietti et al., 28]. In addition, observations show that there can be a close correspondence between observations of electron cyclotron waves and solitary waves [Menietti et al., 28]. 10.6.2 Generation Mechanisms The simplest way to excite electron cyclotron waves is again by transverse temperature anisotropies or loss cones which excite diﬀuse electron cyclotron waves between the harmonic bands. This has been recognized early. Purely transverse electron cyclotron waves are Bernstein modes which normally are damped and represent merely resonances. However, again, when other kinds of distribution functions are present, the resonance can be inverted and waves can be excited instead of being damped. This is, in particular, the case when electron beams pass the plasma and has been suggested as a possible generation mechanism based on particle observations and close association of electron cyclotron waves with solitary waves [Menietti et al., 28]. Other possible sources of instability as for instance loss-cones, temperature anisotropies,

264

A. Vaivads et al.

and horse-shoe distributions exist as well. The diﬀerent excitation mechanisms have been discussed in LaBelle and Treumann [27]. In case of electron beams, it is crucial that in addition to the beam a cold electron population is present in order for the waves to become unstable. Unfortunately, however, direct measurements of such a component in relation to reconnection are in most cases missing [Menietti et al., 28]. 10.6.3 Relation to Reconnection It has been suggested that electron beams generating electron cyclotron waves originate in the reconnection region. The presence of these waves in association with ﬂux transfer events as well as mainly on open ﬁeld lines indicates that the reconnection process is important in creating unstable electron distribution functions. However, observations of electron cyclotron waves close to the reconnection site are so far missing. It has been suggested that the steep magnetic ﬁeld gradients encountered near the reconnection site should preclude the generation of electron cyclotron waves or inhibit them from developing signiﬁcant amplitudes.

10.7 Free Space Radiation 10.7.1 Observations Electromagnetic radio emissions above the electron plasma frequency can propagate freely throughout the plasma and thus be detected at large distance from the source. This makes the observation of radiation a perfect tool for remote diagnostics of reconnection whenever reconnection sites emit such radiation. Observation of radiation emitted from reconnection is thus of high importance in particular in the astrophysical application, e.g. in the interpretation of the radiation emitted from the solar corona [see, eg., the compilation given in Aschwanden, 5]. In fact, most of the solar radio emission at meter wavelengths is believed to be generated in some coherent emission process that is somehow related to ongoing reconnection in the solar corona [Bastian et al., 7]. These emissions can be classiﬁed into diﬀerent classes of which the most important for our purposes are those which are emitted close to the local plasma frequency fpe and become free space modes as they propagate away from the generation region. Coherent radio emissions from other astronomical objects, such as stellar ﬂares and brown dwarfs, are believed to be generated in similar ways [G¨ udel, 15]. The impossibility of performing in situ measurements in the coronal reconnection regions raises the importance of observations at the accessible magnetospheric reconnection sites. Unfortunately, so far there are no in situ studies of electromagnetic wave generation close to reconnection sites in the magnetosphere. Until now, most attention has been paid to the electromagnetic emission generation near the bow-shock where these waves

10 Plasma Waves Near Reconnection Sites

265

are strongest. However, since reconnection sites are sources of fast particles which are injected into the environment one expects that they are also sources of radio wave emission. 10.7.2 Generation Mechanisms The generation mechanisms of possible free space modes in reconnection are not entirely clear. Several possibilities have been suggested. For example, transverse free space modes (T ) can be generated by mode conversion of Langmuir (L-waves) at steep density gradients, by mode coupling of Langmuir waves with ion sound (S) or other Langmuir (L ) waves according to the relations L + L → T, L + S → T or through direct electron gyro-resonance emission. Of these mechanisms, the latter is the least probable in reconnection as it depends on two conditions. First, the plasma has to be relativistic or at least weakly relativistic. Second, and even more crucial, the plasma has to be underdense with fpe < fce . Close to the reconnection site the involved plasma is, however, overdense as stated in the introduction. In this case cyclotron damping of the free space modes inhibits their excitation. Thus, it seems improbable that reconnection sites would radiate by the gyro-resonance mechanism. In situ measurements close to the reconnection sites are required to distinguish between the remaining possible generation mechanisms. Finally, the large number of energetic electrons generated in reconnection might be another source of nonthermal synchrotron radiation under conditions when the electron energies reached are high. 10.7.3 Relation to Reconnection The localization of electromagnetic radiation generation with respect to the reconnection site has not been investigated. Numerical simulations have studied how electron beams generated in the reconnection process can generate Langmuir waves which, in their turn, can mode convert to the emission of electromagnetic radiation [see, e.g. Sakai et al., 32]. This is the most probable radiation mechanism since electron beams are involved. In addition this mechanism is non-thermal. Depending on the available number of electron beam electrons the emission coming from one single reconnection site might still be below the detection threshold in the magnetosphere or at the magnetopause. For astrophysical applications like the sun and stars the ejected electron beams in type III radiation generating plasma waves are dense and intense enough to provide observable intensities. In remote astrophysical objects, however, synchrotron emission is more important as a nonthermal emission mechanism [Jaroschek et al., 18, 19]. Though it is very weak, the large numbers of particles injected from the reconnection site into a large volume and distributed there increase the emission measure in proportion to the involved volume, making such radiation a good candidate for observation even though it will not provide information about the microscopic scale of the involved astrophysical reconnection sites.

266

A. Vaivads et al.

10.8 Summary and Outlook We have summarized the in situ observations of high frequency waves, at frequencies near the lower hybrid frequency and up to the plasma frequency, generated near the reconnection sites in the Earth magnetosphere. There are many observational studies dealing with the most intense waves, such as lower hybrid drift waves, solitary waves, and Langmuir waves. Some of the observations suggest that these waves are most intense along the separatrices emanating from the reconnection sites. However detailed studies of the wave locations are still missing. Electron beams generated in the reconnection process seem to be a major free energy source that can generate diﬀerent waves, but also density gradients or diﬀerent kinds of distribution functions (e.g., loss-cone or horse-shoe) are important. Electromagnetic waves such as whistlers and radio emissions are important for remote diagnostic possibilities of the reconnection sites. In the case of radio emissions there are direct astrophysical applications. However, in situ studies of both these modes in association with reconnection are very limited. We identify several topics of high importance for further study in the near future: • Wave location. How are diﬀerent wave modes located with respect to the inner structure of the current sheet and the separatrices, and how does this location depend on the reconnection parameters, such as density gradients, velocity shear, plasma beta, temporal evolution? Could some of the wave modes be used to determine the distance to the reconnection X-line? Possible candidates are the intense solitary waves and Langmuir/upper hybrid waves. • Wave-particle interaction. Which wave modes are most important for particle acceleration, heating, and formation of energetic tails on the electron distribution functions? Are electron beams generated in the wave particle interaction or are they generated in prompt electron acceleration in reconnection-generated electric ﬁelds? • Anomalous resistivity. There exist ﬁrst estimates of the anomalous resistivity and anomalous diﬀusion for lower hybrid drift turbulence in connection with reconnection. These results should be conﬁrmed for diﬀerent reconnection conditions. Moreover, the electromagnetic part of the anomalous resistivity needs to be studied more closely both theoretically and experimentally. • Radio emissions. Where and by which mechanism is free space radiation generated near the reconnection sites? How could it be used to remotely sense the reconnection site properties such as the stationarity of reconnection, extension of the reconnection line, etc.?

10 Plasma Waves Near Reconnection Sites

267

References [1] Akimoto, K., S.P. Gary, and N. Omidi: Electron/ion whistler instabilities and magnetic noise bursts, J. Geophys. Res. 92, 11209, 1987. [2] Anderson, R.R., T.E. Eastman, C.C. Harvey, M.M. Hoppe, B.T. Tsurutani, and J. Etcheto: Plasma waves near the magnetopause, J. Geophys. Res. 87, 2087, 1982. [3] Andr´e, M., R. Behlke, J.-E. Wahlund, A. Vaivads, A.-I. Eriksson, A. Tjulin, T.D. Carozzi, C. Cully, G. Gustafsson, D. Sundkvist, Y. Khotyaintsev, N. Cornilleau-Wehrlin, L. Rezeau, M. Maksimovic, E. Lucek, A. Balogh, M. Dunlop, P.-A. Lindqvist, F. Mozer, A. Pedersen, and A. Fazakerley: Multispacecraft observations of broadband waves near the lower hybrid frequency at the earthward edge of the magnetopause, Ann. Geophysicæ 19, 1471, 2001. [4] Andr´e, M., A. Vaivads, S.C. Buchert, A.N. Fazakerley, and A. Lahiﬀ: Thin electron-scale layers at the magnetopause, Geophys. Res. Lett. 31, 3803, 2004. [5] Aschwanden, M.J.: Physics of the Solar Corona: An Introduction, Springer, 2004. [6] Bale, S.D., F.S. Mozer, and T. Phan: Observation of lower hybrid drift instability in the diﬀusion region at a reconnecting magnetopause, Geophys. Res. Lett. 29, 33, 2002. [7] Bastian, T.S., A.O. Benz, and D.E. Gary: Radio Emission from Solar Flares, Ann. Rev. Astron. Astrophys. 36, 131, 1998. [8] Birn, J., J.F. Drake, M.A. Shay, B.N. Rogers, R.E. Denton, M. Hesse, M. Kuznetsova, Z.W. Ma, A. Bhattacharjee, A. Otto, and P.L. Pritchett: Geospace Environmental Modeling (GEM) magnetic reconnection challenge, J. Geophys. Res. 106, 3715, 2001. [9] Cane, H.V., W.C. Erickson, and N.P. Prestage: Solar ﬂares, type III radio bursts, coronal mass ejections, and energetic particles, J. Geophys. Res. 107, 14, 2002. [10] Cattell, C., J. Dombeck, J. Wygant, J.F. Drake, M. Swisdak, M.L. Goldstein, W. Keith, A. Fazakerley, M. Andr´e, E. Lucek, and A. Balogh: Cluster observations of electron holes in association with magnetotail reconnection and comparison to simulations, J. Geophys. Res. 110, 1211, 2005. [11] Daughton, W., G. Lapenta, and P. Ricci: Nonlinear Evolution of the LowerHybrid Drift Instability in a Current Sheet, Phys. Rev. Lett. 93, 105004, 2004. [12] Deng, X.H., H. Matsumoto, H. Kojima, T. Mukai, R.R. Anderson, W. Baumjohann, and R. Nakamura: Geotail encounter with reconnection diﬀusion region in the Earth’s magnetotail: Evidence of multiple X lines collisionless reconnection?, J. Geophys. Res. 109, 5206, 2004. [13] Drake, J.F., M. Swisdak, C. Cattell, M.A. Shay, B.N. Rogers, and A. Zeiler: Formation of electron holes and particle energization during magnetic r, Science 299, 873, 2003. [14] Farrell, W.M., M.D. Desch, M.L. Kaiser, and K. Goetz: The dominance of electron plasma waves near a reconnection X-line region, Geophys. Res. Lett. 29, 8, 2002. [15] G¨ udel, M.: Stellar Radio Astronomy: Probing Stellar Atmospheres from Protostars to Giants, Ann. Rev. Astron. Astrophys. 40, 217, 2002. [16] Gurnett, D.A., L.A. Frank, and R.P. Lepping: Plasma waves at the distant magnetotail, J. Geophys. Res. 81, 6059, 1976.

268

A. Vaivads et al.

[17] Hoshino, M., T. Mukai, T. Terasawa, and I. Shinohara: Suprathermal electron acceleration in magnetic reconnection, J. Geophys. Res. 106, 25979, 2001. [18] Jaroschek, C.H., R.A. Treumann, H. Lesch, and M. Scholer: Fast reconnection in relativistic pair plasmas: Analysis of particle acceleration in self-consistent full-particle simulations, Phys. Plasmas 11, 1151, 2004a. [19] Jaroschek, C.H., H. Lesch, and R.A. Treumann, Relativistic kinetic reconnection as the possible source mechanism for high variability and ﬂat spectra in extragalactic radio sources, Astrophys. J. 605, L9, 2004b. [20] Ji, H., S. Terry, M. Yamada, R. Kulsrud, A. Kuritsyn, and Y. Ren: Electromagnetic Fluctuations during Fast Reconnection in a Laboratory Plasma, Phys. Rev. Lett. 92, 115001, 2004. [21] Kellogg, P.J. and S.D. Bale: Nearly monochromatic waves in the distant tail of the Earth, J. Geophys. Res. 109, 4223, 2004. [22] Khotyaintsev, Y., A. Vaivads, Y. Ogawa, B. Popielawska, M. Andr´e, S. Buchert, P. D´ecr´eau, B. Lavraud, and H.R`eme: Cluster observations of high-frequency waves in the exterior cusp, Ann. Geophysicæ 22, 2403, 2004. [23] Kojima, H., H. Furuya, H. Usui, and H. Matsumoto: Modulated electron plasma waves observed in the tail lobe: Geotail waveform observations, Geophys. Res. Lett. 24, 3049, 1997. [24] Kojima, H., K. Ohtsuka, H. Matsumoto, Y. Omura, R.R. Anderson, Y. Saito, T. Mukai, S. Kokubun, and T. Yamamoto: Plasma waves in slow-mode shocks observed by Geotail Spacecraft, Adv. Space Res. 24, 51, 1999. [25] LaBelle, J., R.A. Treumann, G. Haerendel, O.H. Bauer, and G. Paschmann: AMPTE IKRM obsevations of waves assosicated with ﬂux transfer events in the magnetosphere J. Geophys. Res., 92, 5827, 1987. [26] LaBelle, J. and R.A. Treumann: Plasma waves at the dayside magnetopause, Space Sci. Rev., 47, 175, 1988. [27] LaBelle, J. and R.A. Treumann: Auroral radio emissions, 1. Hisses, roars, and bursts, Space Sci. Rev. 101, 295, 2002. [28] Menietti, J.D., J.S. Pickett, G.B. Hospodarsky, J.D. Scudder, and D.A. Gurnett: Polar observations of plasma waves in and near the dayside magnetopause/magnetosheath, Planet. Space Sci. 52, 1321, 2004. [29] Omura, Y., H. Matsumoto, T. Miyake, and H. Kojima: Electron beam instabilities as generation mechanism of electrostatic solitary waves in the magnetotail, J. Geophys. Res. 101, 2685, 1996. [30] Pritchett, P.L.: Collisionless magnetic reconnection in a three-dimensional open system, J. Geophys. Res., 106, 25961, 2001. [31] Pritchett, P.L. and F.V. Coroniti: Three-dimensional collisionless magnetic reconnection in the presence of a guide ﬁeld, J. Geophys. Res. 109, 1220, 2004. [32] Sakai, J.I., T. Kitamoto, and S. Saito: Simulation of Solar Type III Radio Bursts from a Magnetic Reconnection Region, Astrophys. J. Lett. 622, L157, 2005. [33] Scholer, M., I. Sidorenko, C.H. Jaroschek, R.A. Treumann, and A. Zeiler: Onset of collisionless magnetic reconnection in thin current sheets: Three-dimensional particle simulations, Phys. Plasmas 10, 3521, 2003. [34] Shay, M.A., J.F. Drake, B.N. Rogers, and R.E. Denton: Alfv´enic collisionless magnetic reconnection and the Hall term, J. Geophys. Res. 106, 3759, 2001. [35] Silin, I., J. B¨ uchner, and A. Vaivads: Anomalous resistivity due to nonlinear lower-hybrid drift waves, Phys. Plasmas 12, submitted, 2005.

10 Plasma Waves Near Reconnection Sites

269

[36] Stenberg, G., T. Oscarsson, M. Andr´e, M. Backrud, Y. Khotyaintsev, A. Vaivads, F. Sahraoui, N. Cornilleau-Wehrlin, A. Fazakerley, R. Lundin, and P. D´ecr´eau: Electron-scale structures indicating patchy reconnection at the magnetopause? J. Geophys. Res. 110, submitted, 2005. [37] Vaivads, A., M. Andr´e, S.C. Buchert, J.-E. Wahlund, A.N. Fazakerley, and N. Cornilleau-Wehrlin: Cluster observations of lower hybrid turbulence within thin layers at the magnetopause, Geophys. Res. Lett. 31, 3804, 2004. [38] Vaivads, A., Y. Khotyaintsev, M. Andr´e, A. Retin` o, S.C. Buchert, B.N. Rogers, P. D´ecr´eau, G. Paschmann, and T.D. Phan: Structure of the Magnetic Reconnection Diﬀusion Region from Four-Spacecraft Observations, Phys. Rev. Lett. 93, 105001, 2004. [39] Zhang, Y., H. Matsumoto, and H. Kojima: Whistler mode waves in the magnetotail, J. Geophys. Res. 104, 28633, 1999.

Part III

High-Frequency Analysis Techniques and Wave Instrumentation

11 Tests of Time Evolutions in Deterministic Models, by Random Sampling of Space Plasma Phenomena H.L. P´ecseli1,3 and J. Trulsen2,3 1

2

3

University of Oslo, Institute of Physics, Box 1048 Blindern, 0316 Oslo, Norway [email protected] University of Oslo, Institute of Theoretical Astrophysics, Box 1029 Blindern, 0315 Oslo, Norway [email protected] Centre for Advanced Study, Drammensveien 78, 0271 Oslo, Norway

Abstract. We discuss general ideas, which can be used for estimating models for coherent time-evolutions by random sampling of data. They turn out to be particularly useful for interpreting data from instrumented spacecraft. These “new methods” are applied to examples of localized bursts of lower-hybrid waves and correlated density depletions observed on the FREJA satellite. In particular, lower-hybrid wave collapse is investigated. The statistical arguments are based on three distinct elements. Two are purely geometric, where the chord length distribution is determined for given cavity scales, together with the probability of encountering those scales. The third part of the argument is based on the actual time variation of scales predicted by the collapse model. The cavities are assumed to be uniformly distributed along the spacecraft trajectory, and it is assumed that they are encountered with equal probability at any time during the dynamical evolution. Cylindrical and ellipsoidal cavity models are discussed. It turns out that the collapsing cavity model can safely be ruled out on the basis of disagreement of data with the predicted cavity lengths and evolution time scales. Application to Langmuir wave collapse is suggested in order to check its reality and relevance.

Key words: Wave packet dynamics, lower-hybrid wave collapse, Freja observations, nonlinear dynamics, random data sampling

11.1 Introduction When studying data from rocket or satellite observations, one might often encounter situations where the results invite an interpretation in terms of analytical models for the space-time evolution of some physical phenomena, nonlinear plasma waves for instance. Very often the spacecraft velocity is H.L. P´ ecseli and J. Trulsen: Tests of Time Evolutions in Deterministic Models, by Random Sampling of Space Plasma Phenomena, Lect. Notes Phys. 687, 273–297 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

274

H.L. P´ecseli and J. Trulsen

so large that the observations have to be interpreted as “snap-shots” of the phenomena, and a direct comparison between analytical results for the time evolution and the observations is not feasible. The problem seems almost a dead-lock, by the implication that time evolutions can under no circumstances be observed under such conditions. This situation easily stimulates speculations leading to unsubstantiated claims, in the sense that the interpretation of data has been extended beyond what can be justiﬁed by the observations alone. In the present communication we will describe a general method which allows tests of models for deterministic time evolutions, provided the database covers suﬃciently many observations. The basic idea is that one can predict the probability density of observable quantities analytically, based on the deterministic model together with some plausible assumptions, and that these distributions can then be tested experimentally [1, 2]. The results cannot prove a theory to be correct (logically, this might not be possible anyhow), but can certainly be useful in disproving certain models. We give an outline of the statistical arguments, and illustrate the ideas by an analysis of data for lower-hybrid waves as observed by the Freja satellite. For this case an interpretation in terms of lower-hybrid wave collapse might be tempting and has indeed been suggested. Our analysis indicates that this interpretation is in error. Later on we will present more detailed examples, but as an introduction it might here suﬃce to consider a simple (over-)idealized case, where we assume that spherical voids are forming in a plasma. Such a void is assumed to be embedded in the plasma with density n0 , and have a density depletion of ∆n. Any void forms spontaneously, and its radius has a deterministic time variation as, say,

t − t0 t − t0 R(t) = r0 1− , (11.1) T T for t0 < t < t0 + T and R = 0 otherwise. There is a well deﬁned initial time, t0 , for the formation of individual voids, but for the ensemble of voids, these times are random, and mutually independent. The time evolution of the voids is thus completely deterministic, but the formation time (and consequently the collapse times as well) are statistically distributed. When such an ensemble of voids is sampled by a spacecraft, the sampling process is itself associated with statistically distributed quantities. At a certain time, t∗ , the width of the void is 2R(t∗ ), but if it is determined by analyzing the data obtained by a satellite moving along a straight line orbit, we ﬁnd a chord-length , which is in general shorter. We can argue that a chord length distribution can be determined analytically, with the assumption that we encounter a cavity at any position in its cross section with equal probability: any other assumption will imply that the formation of the void is correlated with the presence of the satellite! It is then a simple matter to obtain the probability density for the observed chord-lengths, .

11 Tests of Time Evolutions in Deterministic Models

275

We ﬁrst consider the impact parameter b as measured from the center of the void. We note that with the given assumptions, the probability for having an impact parameter in the interval {b, b + db}, is obtained as the ratio of the two areas 2πbdb and πR2 (t∗ ), assuming that the void does not change appreciably during the passage of the detector. We have P (b)db being the probability of ﬁnding ' an impact parameter in the interval {b, b + db}, and also the relation b = R2 (t∗ ) − 2 /4. Using P (b)db = P ()d we ﬁnd P (|R) =

, 2R2 (t∗ )

(11.2)

for 0 < ≤ 2R(t∗ ) and P (|R) = 0 otherwise. Since the maximum value of R is Rm = r0 /4, we have the maximum value of being m = r0 /2. Implicit in the argument is, as said, the assumption that the void develops slowly, i.e. it does not change appreciably during the transit time of the satellite. The result (11.2) is conditional, as indicated, in the sense that R is assumed given at the relevant time. The information concerning the density in the void is trivial in the present model, since we assumed ∆n = const. In addition to , also the time of interception, t∗ , is statistically distributed with respect to the formation time t0 . Again, we can safely argue that the satellite intercepts the time-evolving void at a time randomly distributed in the interval {0, T }, i.e. P (t∗ )dt∗ = dt∗ /T for 0 < t∗ < T . For the simple example in (11.1), the time variation is monotonic and symmetric in the two intervals {0, T /2} and {T /2, T }, so we need only be concerned with one of them, say the ﬁrst one. The variation of R is restricted to the interval {0, r0 /4}. We want to determine the statistical distribution of radii in the voids, R, which is a consequence of the random distribution of detection times in the relation (11.1). We have P (R)dR = dt∗ /T . With (11.1) we ﬁnd dR/dt∗ = r0 /T − 2r0 t∗ /T 2 , giving P (R) =

2 1 ' . r0 1 − 4R/r0

(11.3)

For R > r0 /4, we have P (R) = 0. Each realization of the plasma has “blobs” randomly distributed in all different stages of their time-evolution, and we encounter them with a probability depending on their cross section, σ = πR2 (t∗ ) , which gives P () =

1 C

(11.4)

r0 /4

σ(R)P (|R)P (R)dR ,

/2

(11.5)

276

H.L. P´ecseli and J. Trulsen

where C is a normalization constant, introduced because the probability density obtained by introducing σ is not automatically normalized. The expression (11.5) is readily solved for this case to give ( 15 1−2 , (11.6) P () = r0 r0 r0 and P () = 0 for > r0 /2. The result (11.6) is the probability density which can be tested experimentally. An estimate for the probability density (11.5) could be obtained experimentally, and we might then hope to ﬁnd support (if not proof) for the proposed model, including the time variation it implies. The model described here is oversimpliﬁed in many respects, in particular also by assuming all structures to develop identically. As a minimum requirement to make the model at least slightly convincing, we have to relax the assumption of identical r0 ’s, and allow the reference magnitude, r0 , of the “blobs” to be randomly distributed, in general over the interval {0; ∞}. We can assume that this distribution is also random, and assign a probability density P (r0 ). The result for the probability density (11.5) or (11.6) of is then conditional, P (|r0 ) and we ﬁnd ∞ P (|r0 )P (r0 )dr0 , (11.7) P () = 2

for a given P (r0 ). Unfortunately we may not know P (r0 ) a priori. The problem can, however, be “inverted”, and we might see whether it is possible to propose a physically realistic P (r0 ), which gives agreement with the observed probability densities. If not, we have a contradiction, and have to reject the model, and may be a little wiser, at least in this respect. If reasonable agreement is found, we might proceed with the model, by making it more detailed and explore its limitations. Finally, before entering into discussions of more realistic cases and their associated models, we might draw attention to a “practical” problem. Even if the phenomena we consider are persistent, we might be in need of data. Thus, we would like to make a data-basis consisting of statistically independent observations, and might be tempted to retain only one observation from each satellite pass, for instance. This seems a safe and sound idea, but it leaves many observations redundant, and the estimate of the probability densities may become uncertain. In addition, the plasma parameters are likely to change from one pass to the next, and having to model statistically distributed plasma as well may become troublesome. It is preferable to use as many data as possible from consistent plasma parameters. If the observations are abundant in some orbits (fortunately, they are, sometimes), we might like to know to what extent, if at all, these observations form a set of independent data? This question is not easily answered, but we can test at least one hypothesis against the observations. By randomly distributed events in space, we mean that the position of one is statistically independent of the position

11 Tests of Time Evolutions in Deterministic Models

277

of all the others. The probability of ﬁnding a structure in a short interval dx measured along the satellite orbit is taken to be µdx, with µ being the density of structures along this line (the dimension of µ is length−1 ). This assumption, we know, leads to a Poisson distribution of the number, N , of structures in an interval L, i.e. P (N, L) = (µL)N exp(−µL)/N !. This distribution can be tested, usually by simply checking ( N 2 − N 2 )/ N = 1, as valid for the Poisson distribution. More extensive tests can also be carried out [3]. In the following Sect. 11.2 we extend the model discussions, by considering the consequences of having more general density variations than the simple “inverted top hat” density depletion model used for illustrations in this Introduction. Then, in Sect. 11.3, we consider one particular problem, namely collapse of lower hybrid waves, and apply some of the ideas developed here.

11.2 Model Discussions Assume that we have an analytical model for the time evolution of a physical phenomenon, predicting a quantity Ψ (r, t), where Ψ might denote the space-time varying plasma density, or the potential or something else. As an example, we might have Ψ representing the space time evolution of a sound wave pulse, phase space vortex, a shock, or similar. Often, we can assume that the phenomenon is associated with one or more characteristic lengths. We can take the width at half-amplitude, for instance. Let such a length scale be Lp (t), which is in general a function of time. We predict Lp (t), but measure Lp (t = t0 ), where t0 is the time we encounter the structure. We do not know what time this is in relation to the beginning of the time evolution of the phenomenon, and this is basically one of the roots of our problem. Another is that we predict a characteristic length as the width of a wave pulse, but the trajectory of the spacecraft need not cross the structure at its maximum diameter. If we later cross a similar object in precisely the same stage of its development we will measure a diﬀerent width, simply because we are likely to cross the object along a diﬀerent trajectory in diﬀerent trials. The problem thus has two basic statistically distributed quantities, one associated with the temporal, and another with the spatial variable. There is a further statistical distribution associated with the fact that diﬀerent events are not identical, but originate from statistically distributed initial conditions. 11.2.1 Spatial Sampling with One Probe Available Let us ﬁrst discuss the distribution in a characteristic quantity as a length scale, due to the random distribution of sampling trajectories. In general, we may not have any a priori knowledge about the actual shape of the structures, i.e. n = n(x, y) in this case. A model can be proposed, however, and subsequently be tested against the data. This will be feasible in particular when the detecting spacecraft is equipped with two or more spatially separated probes.

278

H.L. P´ecseli and J. Trulsen

As a suﬃciently general generic form, useful as an illustration, we consider a density pulse in a simple two dimensional model n(x, y) ≡ n0 − n & = n0 − ∆n e− 2 (x 1

2

+y 2 )m /R2m

.

(11.8)

This model contains an inverted top-hat model as the limiting case m → ∞, and a rotationally symmetric Gaussian model corresponding to m = 1. We might let ∆n have either sign, corresponding to density “humps” or depletions. We can use the same exponent on both x and y if a symmetry condition is satisﬁed, i.e. in the absence of a preferred direction. The chord length, , corresponding to the 1/e-width of the density depletion, as obtained along the satellite orbit, is easily obtained from (11.8). For m > 1 an increase in the impact parameter implies a decrease in measured chord length. For m < 1 the opposite impact parameter variation is obtained; a somewhat counter-intuitive result. The depth of cavities as detected along the sector determined by the satellite trajectory can also be obtained from (11.8), giving an expression varying with impact parameter y. A straight forward parametric representation of the normalized depth – chord length relation can be obtained as 0 1 2m (11.9) {& n/∆n, /R} = e− 2 ξ , 2 (2 + ξ 2m )1/m − ξ 2 , with ξ ≡ y/R being the normalized impact ' parameter. In particular, for m → 2m ∞, we have ξ → 0, so that /R = 2 1 − ξ 2 . The expression (11.9) gives the peak cavity depth and chord length for a given normalized impact parameter y for cavities deﬁned by (11.8), assuming ∆n, R and m to be known. When examining an actual record, the impact parameter is in general varying from cavity to cavity, with y being statistically distributed. The only acceptable assumption concerning its probability density is a uniform distribution, P (|y|) = 1/LM for 0 < |y| < LM and zero otherwise. Here LM is an assumed maximum impact parameter from the satellite to a cavity. Eventually we let LM → ∞. With this assumption we can in principle obtain the probability densities P () and P (& n). For instance, using P (& n)d& n= P (|y|)d|y| we readily ﬁnd P

n & ∆n

=

−1+1/(2m) n & R ∆n 2 ln , mLM n & ∆n

(11.10)

&/∆n ≤ 1. The probability density (11.10) is normalized for e− 2 (LM /R) < n for all ﬁnite LM , but the expression may for some values of m give trivial results in the limit of LM → ∞. The corresponding results for P () are not so easily obtained in a closed form, but can be determined for selected values of m. 1

2m

11 Tests of Time Evolutions in Deterministic Models

279

1) For m = 1/2, we ﬁnd, using P ()d = P (y)dy P () =

, 8RLM

(11.11)

' for 4 < /R < 4 1 + LM /R, see the second half of the expression (11.9). 2) For m = 1 we have √ (11.12) P () = δ − 2R 2 . 3) For m = 2 we have (y/R)2 = [32 − (/R)4 ]/[8(/R)2 ] giving P () =

1 √ LM 2 2

R

2

32 + (/R)4 ' , 32 − (/R)4

(11.13)

' with 2 (2 + (LM /R)4 )1/2 − (LM /R)2 < /R < (32)1/4 . Models with other values of m can be analyzed similarly. It is evident that the results depend sensitively on the model, here the exponent m, and a statistical analysis for determining this and other parameter values is worthwhile. All these foregoing results are conditional in the sense that they assume all cavities to be identical, i.e. given ∆n and R all cavities are of the same constant form (11.8). Given a priori knowledge, or at least a qualiﬁed guess, of the probability density of the characteristic parameters, we can then use Bayes’ theorem to obtain unconditioned distributions. 4) For the case where m → ∞ the analysis becomes more lengthy. We consider two cases, a cylindrical and an ellipsoidal model. Cylindrical Model As an illustration assume ﬁrst that the cavity has a cylindrical shape with a circular cross section of radius L⊥ (t), while L is for the moment considered inﬁnite. The observations are then essentially restricted to a plane. As a simple case, we here assume that we have an “inverted top-hat” density depletion, i.e. n = n0 outside the cavities, and n = n0 − ∆n inside. The geometrical cross-section for the cavity is thus σ = 2 L⊥ with dimension “length” in this planar approximation. The probability for actually encountering a cavity in an interval of length dx along the spacecraft trajectory is proportional to µ(L⊥ ) 2L⊥ dL⊥ dx, where, again, µ(L⊥ )dL⊥ is the + ∞density of cavities with perpendicular radii within L⊥ , L⊥ +dL⊥ . We have 0 µ(L⊥ )dL⊥ = µc , where µc is the cavity density irrespective of diameter. For the present model, the dimensions of µc are length−2 , i.e. the dimensions of µ(L⊥ ) are length−3 . Assuming the cavities to appear and disappear at random, the dynamics can be assumed to be time-stationary in a statistical sense, with the density of cavities µ(L⊥ ) being constant, even if L⊥ for the individual cavity varies with time. In a given realization there are many diﬀerent scale sizes present

280

H.L. P´ecseli and J. Trulsen

at the same time. The probability for encountering one particular value for L⊥ depends on the relative density of cavities with that particular diameter. Assuming that the spacecraft has encountered a cavity it is evident that the probability of its radius being L⊥ is proportional to the density of cavities with that particular radius as well as to the corresponding geometrical crosssection. We can, more generally, derive the probability density for the radii of observed cavities P(L⊥ ), as illustrated later on in (11.25). Let the angle between the spacecraft trajectory and the magnetic ﬁeld be θ. A straight-line trajectory intercepts cavities of radius L⊥ along chords at a constant y-value in the ellipse x2 y2 = 1, + L2⊥ / sin2 θ L2⊥ all |y|-values in the interval {0, L⊥ } being equally probable. With the foregoing assumption, the distribution of chord lengths, , is readily obtained as P (|L⊥ ) =

sin2 θ 1 0 2L⊥ 4L2 − 2 sin2 θ

(11.14)

⊥

for 0 < < 2L⊥ / sin θ for a given ﬁxed L⊥ . The angle θ is considered as a constant. It was also here assumed that the satellite speed is so large that the cavity does not change appreciably during its passage. Evidently (11.14) predicts that there is a large probability of ﬁnding ∼ 2L⊥ / sin θ. For distributed cavity radius L⊥ we have ∞ P () = P (|L⊥ )P(L⊥ )dL⊥ (11.15) 0

= sin2 θ

L⊥max

sin θ/2

1 P(L⊥ ) 0 dL⊥ 2L⊥ 4L2 − 2 sin2 θ ⊥

with the actual probability density P(L⊥ ) for the cavity radii to be inserted. Ellipsoidal Model The foregoing result for P (|L⊥ ), the probability density of observed chord lengths, was derived for a cylindrical form of the cavity. For a spherical cavity we have the simple expression (11.2) already derived in the introduction. More generally a rotationally symmetric ellipsoidal model (i.e. a “cigar”), with half axes L⊥ perpendicular to B and L parallel to B, can be used for the density cavities. We can generalize (11.8) to n = n0 − ∆n e− 2 [(x 1

2

+y 2 +(z/L )2 L2⊥ )m /L2m ⊥ ]

,

(11.16)

Again, the case m = 1 is relatively simple, so we consider the simple “inverted-top-hat” model, with m → ∞.

11 Tests of Time Evolutions in Deterministic Models

281

Because of the rotational symmetry of the problem with respect to the magnetic ﬁeld, (11.16) is presumably an adequate model. The centers of the ellipsoids are assumed to be randomly and uniformly distributed in three dimensions. The cross-section of a given cavity for a spacecraft moving at an angle θ to the major axis is 0 σ(L , L⊥ ) = πL⊥ L2⊥ cos2 θ + L2 sin2 θ. For θ = 0, i.e. for satellite propagations along the major axis of the ellipsoid, which will generally be the magnetic ﬁeld direction, we ﬁnd σ = πL2⊥ , as expected. The probability density P(L⊥ , L ) of encountering a certain cavity speciﬁed by (L , L⊥ ), is proportional to σ(L , L⊥ ) and to the appropriate density of cavities, µ(L , L⊥ ), as discussed before. We have µ(L , L⊥ )dL dL⊥ = +∞+∞ µc P (L⊥ , L )dL dL⊥ , with 0 0 µ(L , L⊥ )dL dL⊥ = µc being the density of cavity centers, irrespective of cavity widths, as before. The probability density P(L⊥ , L ) can not oﬀ-hand be equated to P(L⊥ )P(L ) since the time evolutions of L and L⊥ are in general not statistically independent. As an illustrative model, which will be useful later on with (11.24), we assume L = βL2⊥ , where the constant β is determined by initial conditions. Assuming the probability density P (L⊥ ) to be known (an example will be analytically determined in (11.25), for a speciﬁc model) as well as the cross section obtained previously, we obtain P(L⊥ , L ) = δ(L − βL2⊥ ) P (L⊥ )σ(L , L⊥ ) −1 L⊥max Lmax 2 × δ(L − βL⊥ )P (L⊥ )σ(L , L⊥ ) , 0

0

where Lmax = βL2⊥max . The normalizing factor in the angular parentheses can in many cases be calculated analytically [1], but might become a rather lengthy expression. We now consider the conditional probability density for observed chord lengths for a given cavity speciﬁed by (L , L⊥ ). After some calculations we obtain L2 P ( | L⊥ , L ) = 2 2 , (11.17) 2L L⊥ for 0 < < 2L L⊥ /L with L2 = L2⊥ cos2 θ + L2 sin2 θ. For propagation across the major axis of the ellipsoid, i.e. θ = 90◦ , we have P ( | L⊥ , L ) = /(2L2⊥ ) as a particular case of (11.17), independent of L . This result is identical to the one obtained for a sphere with radius L⊥ , and can be understood by simple geometrical arguments. For distributed values of L⊥ and L we ﬁnd P () as in (11.14) 2 cos θ sin2 θ + (11.18) P(L⊥ , L )dL⊥ dL , P () = 2 L2 L2⊥

282

H.L. P´ecseli and J. Trulsen

with the integration restricted to that part of the L⊥ ,L -plane where 2L2 L2⊥ > 2 (L2⊥ cos2 θ + L2 sin2 θ) and 0 < L⊥ < L⊥max , 0 < L < Lmax . The parenthesis in (11.16) originates from P ( | L⊥ , L ). The maximum values Lmax and L⊥max , are also here assumed to be given. In the actual situation they will diﬀer from case to case, and the corresponding probability density must be included in the analysis at a later stage. 11.2.2 Spatial Sampling with Two Probes Available The models can be tested to a greater accuracy in cases where the spacecraft is equipped with two or more probes for measuring ﬂuctuations in the plasma density. We can then have two cross sections for the same structure, and be able to determine, or at least estimate, parameters in (11.8). Assume that the two probes are separated by a distance ∆s in the direction perpendicular to the velocity vector of the satellite. Referring to the model (11.8) we let again ξ be the normalized impact parameter for probe ' 1, so the normalized chord length detected by this probe is 1 /R = 2 (2 + ξ 2m )1/m − ξ 2 . The normalized chord length detected'by probe 2 crossing the same cavity will be one of the two values 2± /R = 2 (2 + (ξ ± ∆)2m )1/m − (ξ ± ∆)2 with equal probability, since the satellite can impact the structure on either side with equal probability. We introduced the normalized probe separation ∆ = ∆s /R. If it is possible to eliminate ξ from the expressions for 1 and 2± , we can obtain a distribution for the two measurable chord-lengths. If, in particular, we have m = 1, we ﬁnd 1 = 2± irrespective of ξ, emphasizing the special properties of structures with Gaussian shape. For m = 1/2 we have ξ = (1 /4R)2 − 1 to give 0 2 2 2± = 2 (2 + (1 /4R)2 − 1 ± ∆) − ((1 /4R)2 − 1 ± ∆) ' = 4 1 + (1 /4R)2 − 1 ± ∆ . (11.19) For the limiting case m → ∞ we have 1 0

2 2± = 2 1 − 1 − 21 /(4R2 ) ± ∆ ,

(11.20)

although this latter model is complicated by the possibility that one probe crosses the structure, while the other one does not, i.e. ξ < 1 but ξ + ∆ > 1, for instance. If we take an expression like (11.19) or (11.20) to be a hypothesis, it can be directly tested against available data. 11.2.3 Temporal Sampling Independent observations happen at diﬀerent times of the evolution of the structures. If we, for instance, were to analyze whistler wave-packets as they

11 Tests of Time Evolutions in Deterministic Models

283

are often excited in the ionosphere by lightning strokes, we might detect the individual whistler wave-packets at diﬀerent times after the lightning, this time diﬀerence being statistically distributed. Since the arrival of the spacecraft is safely assumed to be independent of the generation mechanism, we can assume the time of observation to be uniformly distributed in a (large) time interval {0, T }, where we might eventually let T → ∞. We have P (t) = 1/T , which gives an immense simpliﬁcation of the analysis, as already illustrated in the Introduction. This assumption is consistent with our other basic assumptions, namely that spatial structures are traversed at positions which are uniformly distributed in the plane perpendicular to the spacecraft orbit.

11.3 Nonlinear Lower-Hybrid Wave Models A number of spacecraft observations demonstrated the existence of small density depletions in the Earth’s ionosphere [4, 5, 6], see Fig. 11.1. The structures were associated with a localized enhanced wave activity, where the waves were identiﬁed as electrostatic lower-hybrid waves [7]. The observations had a close resemblance to what could be expected from an intermediate stage of wave-collapse, and the observations evidently received considerable attention. However, it is evident that a simple visual inspection is not suﬃcient for identifying a collapse phenomenon, and a more detailed analysis had to be performed [1, 3, 8]. As the ﬁrst part of such an analysis it is necessary to ﬁnd a model for the (deterministic) time evolution of observable quantities. A set of nonlinear model equations for the lower-hybrid wave dynamics has been proposed [9, 10, 11]. They are based on a set of nonlinear equations for the wave potential φ and the slowly varying density perturbation n −2i

1 ωLH

2 2 ωpe ∂ 2 M ωLH ∇ φ − Λ2 ∇4 φ + φ ∂t m c2 Ωce

M ∂2 n M ωLH ˆ + φ = −i ∇φ × ∇ ·z m ∂z 2 m Ωce n0

with Λ2 =

2 ωpe Te +2 2 3 2 2 + Ω2 ωLH M Ωce m ωpe ce

Ti

(11.21)

,

2 2 2 2 2 where ωLH = Ωci + ωpi /(1 + ωpe /Ωce ). The time evolution of n is, in one limiting case of the model, given by 2

2 ε0 ωpe ∂ n 2 2 ˆ. − C ∇ = i ∇2 (∇φ∗ × ∇φ) · z (11.22) s ∂t2 n0 4n0 M Ωce ωLH

A driving term in (11.21) is not included as the waves are frequently observed in regions where such a mechanism cannot be identiﬁed. The magnetic ﬁeld is

284

H.L. P´ecseli and J. Trulsen

Fig. 11.1. Example of lower hybrid wave cavity detected by instruments on the FREJA satellite. The lower hybrid wave electric ﬁeld is shown in the upper frame, giving the medium frequency band of the detecting probe circuits. The signal has a bandwidth of 0–16 kHz and a sampling rate of 32 · 103 samples/s. The bottom frame shows the low frequency relative plasma density variations in percent as obtained from two Langmuir probes with an 11.0 m separation. The middle frame shows a wavelet transform of the upper frame [1]. The spin-period of the satellite is approximately 6 s, so its spin phase can be considered constant during a given data sequence

11 Tests of Time Evolutions in Deterministic Models

285

ignored in (11.22) for the slow plasma variation because it is anticipated that the appropriate time scale is shorter than the ion gyro-periods, approximately 2.5 ms for H+ and 35–40 ms for O+ for the present conditions. The assumption has to be veriﬁed a posteriori, but in case the assumption is violated, the equations are readily modiﬁed. For time-stationary conditions this equation is reduced to 2 ε0 ωpe n ˆ. =i (∇φ × ∇φ∗ ) · z (11.23) n0 4n0 Te Ωce ωLH Contrary to the case of Langmuir oscillations, the lower-hybrid waves can be localized within density wells, n < 0, as well as density “humps”, n > 0 [11]. This apparent symmetry is broken by small terms left out in the analysis. The properties of the model equations (11.21)–(11.22) have been studied in great detail [11]. Most important is the observation of collapsing solutions. For suﬃciently large wave intensities the nonlinearities form wave-ﬁlled plasma cavities that collapse into singularities within a ﬁnite time with time variations of their perpendicular and parallel diameters given by L⊥ ∼ (tc − t)1/2 and L ∼ (tc − t) ,

(11.24)

where tc is the collapse time and t0 an arbitrarily chosen initial time for the process, t0 < t < tc . At the same time the electric ﬁeld amplitude at the center of the collapsing cavity increases without bounds. The result in (11.24) is based on a sub-sonic model ' for the collapse. The wave cavity is strongly Bﬁeld aligned with L > L⊥ M/m. There is no natural propagation velocity associated with the cavity. We introduce the initial maximum length scales as L⊥ = L⊥max and L = Lmax at t = t0 . Ultimately, for very large ﬁeld amplitudes and small cavity sizes, the model equations (11.21)–(11.22) become unphysical also because they lack a proper dissipative mechanism. With linear damping being small or negligible, we expect transit time damping to be the dominating mechanism [12, 13]. Here, particles from the surrounding plasma pass through the collapsing cavity and interact with the intensiﬁed electric ﬁelds there. A particle passing through this region in less than one period of the oscillating electric ﬁeld will on average gain energy, giving rise to a corresponding damping of the electric ﬁeld. For parameters Ωe > ωpe discussed here, the collapse will be damped by electrons with small pitch angles passing through the collapse region in the direction parallel to the magnetic ﬁeld. Signiﬁcant damping and eventual arrest of the collapse process are expected when the parallel length of the density depletion becomes comparable to the distance travelled by a thermal electron during one period of the lower-hybrid wave. For the present parameters we ﬁnd a corresponding minimum length scale of Lmin ∼ 500 m. Electrons gain energy (on average) at a single pass when propagating along the magnetic ﬁeld. Fast, light, ions moving with a large Larmor radius from the surrounding plasma into the cavity can also contribute to the damping by being accelerated in that fraction of their gyroperiod they spend inside the cavity. This contribution to

286

H.L. P´ecseli and J. Trulsen

the wave damping is somewhat more complicated to describe analytically, and it is eﬀective only for perpendicular cavity widths smaller than or comparable to the ion Larmor radius, L⊥ ≤ ρL . Due to their gyration, ions interact many times with an intensiﬁed ﬁeld in a long ﬁeld aligned cavity. Since ωLH > Ωci we can assume that the wave phase is randomly varying at each ion encounter with the cavity. This multiple interaction gives a multiplication factor of the order of L /(v τi ), where v is the B-parallel ion velocity and τi = 2π/Ωci . Thus, even for cases where ions (on average) obtain only a small energy increment at a single pass through a cavity, the net result can be signiﬁcant. This mechanism clearly favors transverse ion energization where v is small. In agreement with the analytical results based on (11.21) and (11.22), we would expect lower-hybrid wave ﬁelds to build up by some mechanism, e.g. a linear instability. When a suﬃciently large amplitude has been reached, a modulational instability sets in, breaking the wavetrain up into wave cavities which collapse and ultimately dissipate the wave energy at small scales. In the case where the waves are generated at short wavelengths the model outlined here may have to be modiﬁed by inclusion of parametric decay processes. Two basically diﬀerent scenarios can be envisaged; one where the lowerhybrid waves are continuously maintained, essentially rendering the cavity formation a statistically time stationary process. Alternatively, we can imagine a situation where lower-hybrid waves are excited in a large region of space as a burst, which eventually breaks up into cavitons and ﬁnally dissipates. It is not possible to discriminate between these two scenarios on the basis of the available data. However, large amplitude lower-hybrid waves are frequently observed at medium frequencies, and it is likely that they constitute a background wave component for extended periods of time.

11.4 Probability Densities for Observables In this section we discuss the observations with the a priori assumption that their explanation is to be sought in a collapse of lower-hybrid waves, without reference to the wave generation mechanism. As it is not possible to study the full space-time evolution of individual events in detail, we attempt to derive probability densities for observable quantities within such a model. These can then be compared with those obtained from the data. In the early investigations of Freja observations, when data were sparse, a simple “inverted-top-hat” density depletion was used to model the density cavities [1]. Later, more detailed, studies [8] demonstrated that a more simpler Gaussian depletion model was the most accurate. In particular, we discuss here the probability density for the cavity width determined from the data, as this is one of the quantities which is most easily and accurately obtained. This width coincides with the width of the localized wave packets, so these need not be analyzed separately in this context. The

11 Tests of Time Evolutions in Deterministic Models

287

measurements of the electric ﬁeld intensities are uncertain and give underestimates, and a statistical analysis of the lower-hybrid wave amplitudes is not really meaningful. At early stages a cavity is presumably large and irregular and will hardly be distinguished from the background of low frequency density ﬂuctuations. Assume that after reaching a signiﬁcant amplitude, a cavity can be eﬀectively recognized in a time interval {t0 , tc }, and that the time variation here can be described by (11.24) with tc being the collapse time. Evidently, we can assume that the spacecraft intercepts a cavity at times uniformly distributed in {t0 , tc }, i.e. the probability density for the time of interception is P (t) = 1/(tc − t0 ) for t0 < t < tc . The distribution of length scales is obtained from P (L)dL = P (t)dt. By use of (11.24) the results are P (L⊥ ) =

2 L⊥ L2⊥max

and

P (L ) =

1 , Lmax

(11.25)

for 0 < L⊥ < L⊥max and 0 < L < Lmax , respectively. For a uniform random temporal distribution of collapse times tc , the density of cavities having a transverse scale in the interval {L⊥ , L⊥ + dL⊥ } is given by µ(L⊥ )dL⊥ = µc P (L⊥ )dL⊥ , where µc is the spatial density of collapse centers. It was here implicitly assumed that all cavities start out with essentially the same value for Lmax . Later we relax this condition. As the satellite moves (essentially in the direction perpendicular to B) we expect predominantly large scales to be detected, observations of the actual collapse being statistically improbable. Although the collapse itself is thus unlikely to be detected (its “cross-section” is too small), the entire time evolution preceding it will in principle be reﬂected in the probability distribution of length scales. The cavity spends a comparatively large time in a state with large diameters, where it also has the largest cross-section for observation. For the simple cylindrical Gaussian model, with m = 1 in (11.8), we have L⊥max √ √ (11.26) δ( − 2L⊥ 2)P (L⊥ )dL⊥ = P (|L⊥max ) = 2 2L⊥max 2 0 √ using (11.25). For > 2L⊥max 2, we have P (|L⊥max ) = 0. If we have a statistical distribution of L⊥max for the ensemble of cavities, we evidently have P (L⊥max ) ∞ P () = dL⊥max . σ(L⊥max ) C /2 L2⊥max The integral, and the normalizing quantity C, can readily be solved for realistic choices of P (L⊥max ), using that σ(L⊥max ) ∼ 2 L⊥max for the present cylindrical model. Within an “inverted top hat” cylindrical density depletion model, m → ∞ in (11.8), and the time evolution for L⊥ given by (11.24) and (11.25) as appropriate for a lower hybrid collapse, the probability density of observed chord lengths becomes

H.L. P´ecseli and J. Trulsen 1.5

1.5

a

P2Lmax

P2Lmax sinΘ

288

1 0.5 0

0.5 1 sinΘ2Lmax

b

1 0.5 0

1.5

0.5 1 2Lmax

1.5

Fig. 11.2. Probability densities for chord lengths, , in the case of randomly distributed time-stationary cylindrical cavities in (a), and ellipsoidal ones in (b). Solid lines show cases where all cavities are identical, the dashed line illustrates the eﬀect of distributed diameters, according to a model distribution. The results are obtained by explicitly using the properties of an “inverted top-hat” density depletion, m → ∞ in (11.8)

3 sin2 θ P (|L⊥max ) = 4 L2⊥max

1

1−

sin θ 2L⊥max

2 (11.27)

for sin θ < 2L⊥max and P (|L⊥max ) = 0 otherwise. This result is shown in Fig. 11.2a. In case there are reasons to expect that cavities have a significant distribution in scale sizes, L⊥max , the averaging over the appropriate probability density is easily included in (11.27). Still with an “inverted top hat” density depletion model and the time evolution for L⊥ given by (11.24), we consider a three dimensional case with randomly distributed ellipsoids. For simplicity we consider the case θ ∼ 90◦ , i.e. the satellite propagates essentially perpendicular to B. Inserting the previously obtained P(L⊥ , L ) we obtain

2 1− P () = 2 . (11.28) L⊥max 2L⊥max This result is shown in Fig. 11.2b with a full line. This result, as well as that in Fig. 11.2a, is applicable for the case where the spread in the values of L⊥max is small. The dashed line on the ﬁgure indicates a corresponding result for a large spread, where for the sake of argument, we assumed a probability density of the form

4 64 L⊥max √ P (L⊥max ) = exp(−2(L⊥max /L0 )2 ) , L0 L0 3 2π where L0 is a typical scale length for the conditions.√(For the form of P (L⊥max ) used here, we actually ﬁnd L⊥max = 8/(3 2π)L0 ≈ 1.06L0 .) We have no a priori physical arguments for choosing the probability density given here, but ﬁnd it suﬃciently general to accommodate realistic problems. Now, (11.27) as well as (11.28) assigns a ﬁnite probability for measuring

11 Tests of Time Evolutions in Deterministic Models

289

very small chord lengths, corresponding to cases when the satellite trajectory crosses almost at the boundary of the cavity. In reality such cases are unlikely to be properly recognized in the background noise level and these contributions will be under represented in the experimental estimate for the probability density. It is interesting that (11.27) as well as (11.28) gives a ﬂat maximum for P () for in the range ∼ (1 − 2)L⊥max , see Fig. 11.2. Also, we ﬁnd it comforting that the cylindrical model and the ellipsoidal one in Figs. 11.2 a) and b) are similar, this means that the result is relatively robust and does not depend on ﬁne details in the models. In the foregoing discussion, the length scale L⊥max entered simply as a parameter. It can be related to other basic parameters for the wave-ﬁeld. Thus, the typical length-scale for lower-hybrid cavities formed by the modulational instability can be estimated by balancing the nonlinear term with the dispersive term [11], using (11.23) in (11.21). The estimate for the resulting maximum, B-transverse length scale is given by L⊥max ∼ Λ

2 4n0 Te m Ωce 2 ε0 | ∇φ |20 M ωpe

1/2 ,

(11.29)

where | ∇φ |20 is the square of the electric ﬁeld in the center of the cavity at the initial stage. In case the high-frequency ﬁeld is turbulent with a broad wavenumber range, we expect that the scale-length obtained here have to be shorter than the correlation length of the ﬂuctuations. The B-parallel scale length becomes ( L2⊥max M . (11.30) Lmax ∼ Λ m Also a characteristic time-scale for the collapse process can be estimated by balancing the time derivative with the nonlinear term in (11.21) and using (11.23), giving 2 −1 m Ωce 4n0 Te tLH ∼ ωLH . (11.31) 2 ε | ∇φ |2 M ωpe 0 0 The time scale obtained in this way may, of course, be a somewhat crude estimate. A more accurate expression can be obtained from the growth rate γLH for the modulational instability which can be obtained with some algebra. −1 , at least for order of magnitude estimates. A It was found [11] that tLH ∼ γLH characteristic contraction rate for the cavity evolution based on the analytical estimates is given by the ratio L⊥max /tLH . This contraction rate is below the ion-sound speed consistent with sub-sonic collapse. Assuming that we have determined a characteristic shape for the structures, for instance by estimating an optimum value for m in (11.8), we can attempt to obtain a distribution of the peak absolute value (with positive or negative sign) of the density detected by a satellite pass of a structure. For instance for m = 1 we ﬁnd distributions illustrated in Fig. 11.3 by using (11.10).

10 8 6 4 2

Pn ΣLM R

H.L. P´ecseli and J. Trulsen Pn nLM R

290

a 0 0

0.5

n n

1

5 4 3 2 1 0

b 0

0.1

0.2

n Σ

Fig. 11.3. The conditional probability density for normalized density depletions, n ˜ /∆n is shown in (a), obtained for the preferred model with m = 1 in (11.10), with a given ∆n. Assuming a Rayleigh distribution for the peak densities ∆n we obtain the result shown in (b)

Assuming that we deal with density depletions we have in Fig. 11.3a the distribution of detected density minima, assuming that all cavities are identical, i.e. ∆n is the same for all. The two singularities in P (& n/∆n) are easily understood by simple geometrical arguments. In Fig. 11.3b we illustrate the eﬀect of Rayleigh distributed values for ∆n.

11.5 Observations As already mentioned, our emphasis will be on the statistical properties of the observed density depletions associated with lower hybrid wave cavitation, as detected by the Freja satellite. First we want to investigate to what extent we might assume these cavities to be randomly distributed along the spacecraft orbit. The experimentally obtained distribution of the number of cavities in intervals of given length are shown in Fig. 11.4, together with the probability density for the relative distance between cavities. For the present case we estimate the density of cavities to be approximately µ = 2.5 × 10−3 m−1 . The distributions are very reproducible, also when data are combined from orbits obtained with more than a year time separation. Very large distances are under-represented in the data because of the ﬁnite sample duration (usually 0.75 s). The dots show an exponential ﬁt, exp(−x/ζ), where ζ corresponds to approximately 200 m. The exponential ﬁt indicates that the probability of ﬁnding a cavity in a small interval dx is proportional to dx itself, with a constant of proportionality given by the density, µ, of cavities along the spacecraft trajectory. In particular, we should like to point out that a model with spatially uniformly distributed, statistically independent density depletions results in a Poisson distribution, P (N ) = (µL)N exp(−µL)/N !, for the number of cavities, N , in an interval, L, along the spacecraft trajectory, as discussed in Sect. 11.1. In order to give an estimate for the shape of the cavities, we note that very often we have two probes for density measurements active on Freja. It

25

25

20

20 Occurrence (%)

Occurrence (%)

11 Tests of Time Evolutions in Deterministic Models

15 10

15 10 5

5 0

291

0 0

5

10 Number of cavities

15

0

500

1000

1500

2000

2500

3000

Distance between cavities (m.)

Fig. 11.4. Left: Experimental estimate for the probability density for the number of cavities in a segment of duration 0.375 s (corresponding to a distance of 2330 m along the spacecraft trajectory). Right: distribution of distance between cavities. A slight data gap appears for cavity separations smaller than a typical cavity width. The small ﬁlled circles indicate the appropriate results for a Poisson distribution in both ﬁgures

Fig. 11.5. Scatter plot of the two corresponding chord lengths ( 1 , 2 ) as detected by the two Langmuir probes crossing a cavity at two separated positions. The ﬁgure contains observations of 130 cavities. All points are essentially located on the center line, indicating that the chord-lengths detected by the two probes are independent of the probe separation, giving an indication for the validity of the Gaussian model with m = 1 in (11.8).

is therefore feasible to follow the ideas outlined in Sect. 11.2.2. In Fig. 11.5 we thus show the two dimensional distribution of chord-lengths, as obtained by simultaneous observations of the same cavity by both probes, irrespective of the spin-phase of the satellite, i.e. independent of the component of the probe separation in the direction of the satellite’s velocity vector. We note basically that all observation points are located on the line 1 = 2 , indicating

H.L. P´ecseli and J. Trulsen 25

25

20

20

Occurrence (%)

Occurrence (%)

292

15 10

15 10 5

5

0

0 0

25

50

75

100

Chord lengths (m)

125

150

0

0.05

0.10

0.15

Depth of cavities (relative amplitude dn/n)

Fig. 11.6. Left: Distribution of chord lengths (not to be confused with cavity widths if m = 1 in (11.8)) obtained from the data. Right: Distribution of cavity depths as detected along the spacecraft orbit

an excellent ﬁt to the model assuming m = 1 in (11.10). We can thus argue that the depletions have a Gaussian shape to a good approximation [8]. The distribution of cavity widths obtained on the basis of data from one selected orbit is shown in Fig. 11.6. Here chord lengths are identiﬁed as the separation between the two points of maximum curvature at the baseline of the signal. The distribution shown in Fig. 11.6 is very robust; it is well reproduced for data obtained with a time separation of more than a year. We can also obtain the distribution of the maximum depth of the density depletion along the probe trajectories, with results shown in Fig. 11.6. We note a convincing similarity with the results shown in Fig. 11.3b as obtained for the Gaussian model-shape, which is supported by Fig. 11.5, provided that we include a wide statistical distribution of the peak values of the density depletions over the ensemble of cavity realizations. In Fig. 11.7, we show a scatter-plot of the distribution of density depletions and chord lengths, as detected along the spacecraft orbit. As already demonstrated, we have solid evidence for the Gaussian model, with m = 1 in (11.8) to be representative for the density depletions associated with the cavities [8]. Then the abscissa on Fig. 11.7 in eﬀect gives the cavity width, just as Fig. 11.6.

11.6 Discussions By comparing the observed probability densities for chord lengths with those derived on the basis of a simple collapse process we ﬁnd that the shape of the theoretical probability density are not readily made to agree with the observed distributions. The disagreement is most conspicuous in Fig. 11.7, where we readily note that the deepest cavities are associated with the intermediate scales, and not with the most narrow ones, as one would expect by a collapse model. This observation can be supported even more strongly by data from other passes [8]. We note also nontrivial disagreements when

11 Tests of Time Evolutions in Deterministic Models

293

Fig. 11.7. Left: Scatter plot for distribution of density depletions and chord lengths (which for the m = 1 model equals cavity widths), as detected along the spacecraft orbit. The vertical “stripes” in the ﬁgure are due to the temporal sampling of the data. Right: scatter plot of corresponding values for peak observed relative density variations, n ˜ /n0 in percent, and the available electric ﬁeld components obtained from the peak electric ﬁeld detected at medium frequencies by the antennas within the cavities. The electric ﬁelds are measured as the half peak-to-peak value (pp/2), measured at maximum. For a given electric ﬁeld component, we usually have two measurements for the density depletion, one from each probe. We plot both points, shown with small open and ﬁlled circles, respectively

characteristic times, tLH , and scale sizes, L⊥max , from theory are compared with observations [8]. The seeming disagreement with an interpretation based on collapsing lower-hybrid waves can be substantiated also by use of the analytical expressions (11.21) and (11.22). 11.6.1 Length Scales The observational results speak in favor of a situation where the lower-hybrid waves are of a somewhat “bursty” nature, and it is the distance between the bursts which would determine the separation, while the amplitude and spatial extent of the bursts set the value for L⊥max in the individual cavitations. This interpretation is physically quite plausible and is in itself not contradictory to the collapse model. Concerning individual cavities, the data are not able to provide a direct estimate of L , so we are here only concerned with the characteristic value for the perpendicular length scale. From Figs. 11.6 and 11.7 we estimate L⊥ ≈ 60 m, while L⊥max ∼ 150−200 m. A typical relative density depletion is approximately 25%. It should be brought in mind that the analytical estimate (11.29) can be used at any stage of the collapse [10], i.e. the instantaneous value of the cavity !1/2 2 2 width can be estimated as L⊥est ∼ Λ 4n0 Te mΩce /(ε0 | ∇φ |2 M ωpe ) as an order of magnitude, with ∇φ now being the actual value of the electric ﬁeld in the center of the cavity. Taking the value of 40 mV/m from Fig. 11.1, we ﬁnd L⊥est ∼ 2 m. The electric ﬁeld is a lower limit and L⊥est derived from it is then an upper limit for the cavity diameter. This value cannot by any

294

H.L. P´ecseli and J. Trulsen

means be accommodated within the results of Fig. 11.1. The disagreement between estimates from the collapse model and the observations become even more pronounced by taking | ∇φ | ∼ 70–100 mV/m which are, after all, also being observed, see Fig. 11.7. For these cases, the observed chord lengths are similar to those in Fig. 11.6, while the analytically estimated maximum cavity width should be less than 1 m. Such narrow deep cavities are never observed, although this is, with a slight margin, within the instrumental capability of the detecting circuits. Evidently, the arguments in these sections are based on the assumption that the cavity parameters L⊥ and L do not change appreciably during the time it takes the spacecraft to traverse a cavity. The slow time variation in the density signals from the two probes conﬁrm that this requirement is fully satisﬁed. It is, however, not so for an arbitrary collapse scenario predicted from the model equations (11.21) and (11.22). It could be argued that the transit time damping arrests the contraction of cavities by damping out the lower-hybrid waves at a certain small scale. This argument would, however, imply that predominantly the smallest cavities should be void of wave activity, in disagreement with observations. 11.6.2 Time Scales From the observations (Fig. 11.2) a typical size of the observed cavities perpendicular to the magnetic ﬁeld is ∼60 m. The satellite (with velocity ∼ 6 km/s in the B-perpendicular direction) is traversing this distance in approximately 10 ms. Cavities are frequently traversed without indications of any deformation during the time of passage, as judged from the two density probe signals, see for instance Fig. 11.1. In the case of a signiﬁcant cavity contraction during the passage of the satellite, the time record of the density depletions and corresponding wave-envelopes should appear skew, or signiﬁcantly non-symmetric. Signiﬁcantly skew density variations are observed only very rarely, and when they occur their skewness can have both signs. On the basis of the experimental data, we thus argue that the characteristic time for the cavity evolution must be signiﬁcantly larger than 10 ms, a time-scale of 100 ms is probably an underestimate. This means that at least the H + -component is magnetized on relevant time scales. The characteristic time scale, tLH , for a wave-collapse process estimated from the analytical result (11.31) is, on the other hand, comparable to or smaller than 10 ms and therefore too short to be in agreement with the experimental results. The probability densities derived on the basis of a collapse model are giving unfair representation of the smallest scales, i.e. no collapse model is expected to remain valid when the length scales approaching scales where transit time damping becomes important. This argument is not signiﬁcant for our interpretation because such small scales will escape observation due to their small cross section. The smallest scales thus have negligible weight in our results. The essential element of the collapsing time-variation of the cavity

11 Tests of Time Evolutions in Deterministic Models

295

widths, as given by (11.24), is that the time variation of large scales is slow, while small scales change rapidly. Signiﬁcant modiﬁcations of the obtained statistical distributions will only be obtained if these conditions are reversed, thus making the small scales being those most likely to be observed. This type of time evolution can, however, not be accommodated within a description based on wave collapse.

11.7 Conclusions In this paper we discussed general ideas which can be used for estimating models for coherent time-evolutions by random sampling of data. The ideas may turn out to be particularly useful for interpreting data from instrumented spacecrafts, and were here illustrated by discussions of localized bursts of lower-hybrid waves and correlated density depletions observed on the FREJA satellite [5, 6]. Particular attention was given to an explanation in terms of wave-collapse. The statistical arguments are based on three distinct elements. Two are purely geometric, where the chord length distribution is determined for given cavity scales, together with the probability of encountering those scales. The third part of the argument is based on the actual time variation of scales predicted by the collapse model. The statistical assumption is basically that the cavities are uniformly distributed along the spacecraft trajectory and that they are encountered with equal probability at any time during the dynamic evolution. We believe that the cylindrical and ellipsoidal models discussed are suﬃciently general to accommodate actual forms of a collapsing cavity. The time variation given by (11.24) is only an approximation at early times of the evolution of large cavities. This uncertainty cannot be of importance as the large scales seem not to be signiﬁcantly represented in the data, in spite of their large cross-section. Since the measured electric ﬁelds are somewhat uncertain [1, 3, 8], we have not discussed the statistics of this quantity in any greater detail. We concluded that the interpretation in terms of wave collapse in its simplest form can be ruled out on the basis of a pronounced disagreement between the length and time scales predicted by the collapse-model and those observed in the data. The experimentally observed distribution of chord lengths in the cavities was explained best by assuming a large number of B-elongated cavities, uniformly distributed in space, with values of 40–80 m for the diameters in the direction perpendicular to the ambient magnetic ﬁeld [14]. We proposed [1] a mechanism where the cavities start out with a B-perpendicular scale size very close to the one they end up with. It is assumed that the waves give up energy to ions as well as electrons. This seems to be the best candidate for explaining this characteristic length scale, which is then a consequence of the thermal expansion of the plasma, with electrons streaming along B while ions expand across the magnetic ﬁeld lines for transverse structures smaller than or comparable to twice the ion Larmor diameter of ions. It was demonstrated

296

H.L. P´ecseli and J. Trulsen

[1] that based on this model, a probability density for chord-lengths can be derived which agrees well with observations. With minor modiﬁcations, the statistical analysis presented in the present work can be generalized also for studies of the possible evidence of Langmuir wave collapse in rocket or satellite data. In particular, the ellipsoid approximation discussed here contains also the “pancake” model which is relevant for the Langmuir problem in weak magnetic ﬁelds [15]. It is self evident that the analysis summarized in the present communication refers to coherent phenomena. In case the structures of interest are embedded into a turbulent background, a ﬁltering of the data might be advantageous. Several such methods have been discussed in the literature, conditional sampling for instance [16], but also matched ﬁlters might be useful.

Acknowledgments The present work was carried out as a part of the project “Turbulence in Fluids and Plasmas,” conducted at the Centre for Advanced Study (CAS) in Oslo in 2004/05.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

P´ecseli, H. L. et al.: J. Geophys. Res. 101, 5299, 1996. Kofoed-Hansen, O., H. L. P´ecseli, and J. Trulsen: Phys. Scr. 40, 280, 1989. Kjus, S. H. et al.: J. Geophys. Res. 103, 26633, 1998. Vago, J. L. et al.: J. Geophys. Res. 97, 16935, 1992. Dovner, P. O., A. I. Eriksson, R. Bostr¨ om, and B. Holback: Geophys. Res. Lett. 21, 1827, 1994. Eriksson, A. I. et al.: Geophys. Res. Lett. 21, 1843, 1994. Schuck, P. W., J. W. Bonnell, and P. M. J. Kintner: IEEE Trans. Plasma Sci. 31, 1, 2003. Høymork, S. H. et al.: J. Geophys. Res. 105, 18519, 2000. Musher, S. L. and B. I. Sturman: Pis’ma Zh. Eksp. Teor. Fiz. 22, 537, 1975, english translation in: JETP lett. 22, 265 1975. Sotnikov, V. I., V. D. Shapiro, and V. I. Shevchenko:, Fiz. Plazmy 4, 450, 1978. Shapiro, V. D. et al.: Phys. Fluids B5, 3148, 1993. Robinson, P. A.: Phys. Fluids B3, 545, 1991. Skjæraasen, O. et al.: Phys. Plasmas 6, 1072, 1999. McBride, J. B., E. Ott, J. P. Boris, and J. H. Orens: Phys. Fluids 15, 2367, 1972. Krasnosel’skikh, V. V. and V. I. Sotnikov: Fiz. Plazmy 3, 872, 1977, see also Sov. J. Plasma Phys. 3, 491, 1977. Johnsen, H., H. L. P´ecseli, and J. Trulsen: Phys. Fluids 30, 2239, 1987.

12 Propagation Analysis of Electromagnetic Waves: Application to Auroral Kilometric Radiation O. Santol´ık1 and M. Parrot2 1

2

Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic; also at IAP/CAS, Prague, Czech Republic. [email protected] LPCE/CNRS, Orl´eans, France. [email protected]

Abstract. We give a brief tutorial description of techniques for determination of wave modes and propagation directions in geospace, based on multi-component measurements of the magnetic and electric ﬁeld ﬂuctuations. One class of analysis methods is based on the assumption of the presence of a single plane wave and can be used to determine the direction of the wave vector. If the wave ﬁeld is more complex, containing waves which simultaneously propagate in diﬀerent directions and/or wave modes, the concept of the wave distribution function becomes important. It is based on estimation of a continuous distribution of wave energy with respect to the wave-vector direction. This concept can furthermore be generalized to the distribution of energy in diﬀerent wave modes. As an example of analysis of satellite data, our emphasis is on application of these techniques to the high-frequency waves, for example to Auroral Kilometric Radiation (AKR). We analyze multicomponent data of the MEMO instrument obtained using multiple magnetic and electric antennas onboard the Interball 2 spacecraft. Results of diﬀerent analysis techniques are compared. The intense structured AKR emission is found to propagate predominantly in the R-X mode with wave energy distributed in relatively wide peaks at oblique angles with respect to the terrestrial magnetic ﬁeld. As expected, the AKR sources correspond to multiple active regions on the auroral oval.

Key words: Electromagnetic waves in plasmas, direction ﬁnding, wave distribution function, auroral kilometric radiation

12.1 Introduction Waves in space plasmas can often simultaneously propagate in diﬀerent modes (with diﬀerent wavelengths) at a given frequency. To trace these waves back to their original source regions and to estimate their source mechanisms, recognition of their modes and propagation directions in the anisotropic plasma O. Santol´ık and M. Parrot: Propagation Analysis of Electromagnetic Waves: Application to Auroral Kilometric Radiation, Lect. Notes Phys. 687, 297–312 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

298

O. Santol´ık and M. Parrot

medium is often crucial. Experimental analysis of propagation modes and directions of waves in plasmas is easier and more reliable if we measure several components of the ﬂuctuating magnetic and electric ﬁelds at the same time. Such measurements using multiple antennas on spacecraft have been ﬁrst proposed by Grard [12] and Shawhan [66]. The complete set of three Cartesian components of the vector of magnetic ﬁeld ﬂuctuations and three Cartesian components of the electric ﬁeld ﬂuctuations (and, in some cases, also the density ﬂuctuations) would be the basis of an ideal data set. However, most often some of those measurements are missing for various, predominantly technical, reasons. For example, it is technically diﬃcult to place a long electric antenna in the direction of the spin axis of a spin-stabilized spacecraft. Several analysis methods applicable to the multi-component measurements have been ﬁrst developed for the ground-based geophysical data [e.g. 2, 35, 36, 54, 55]. These methods and other newly developed techniques [e.g. 8, 25, 58, 64, 71, 72] have been later used for analysis of data of spacecraft missions carrying instruments for multi-component measurements of the wave magnetic and electric ﬁelds in various frequency ranges, such as GEOS, Aureol 3, Freja, Polar, Interball 2, Cluster, Double Star, and DEMETER. Although these missions were designed to investigate diﬀerent regions of the geospace, similar analysis methods have been used for their wave measurements [e.g. 29, 30, 31, 32, 33, 45, 46, 47, 48, 56, 57, 58, 59, 60, 61, 62, 73]. Analysis methods which are described in this chapter rely on this heritage. We will concentrate on their application to the high-frequency waves, namely to Auroral Kilometric Radiation (AKR). AKR is a strong emission of radio waves at frequencies from a few tens of kHz up to 600–800 kHz. It was ﬁrst observed more than 30 years ago by Benediktov et al. [5] and Dunckel et al. [9] but still is a subject of active research. The ﬁrst proper interpretation was done by Gurnett [15], who demonstrated the electromagnetic nature of the waves and found the origin of the radiation in the terrestrial auroral zone at altitudes of a few Earth radii (RE ). A widely accepted generation mechanism is based on the relativistic cyclotron maser interaction with a “horse-shoe” or “shell” electron distribution function [34, 51, 53, 74] connected to active auroral regions [15, 20, 39]. The AKR emissions consist of many narrow-band components with varying center frequencies [see, e.g., Gurnett et al., 16]. The bandwidth of these components is typically 1 kHz but it could be as low as 5–10 Hz [see, e.g., Baumback and Calvert, 3]. This ﬁne structure could be explained by a wave ampliﬁcation in a resonator represented by small ﬁeld-aligned structures in the auroral region [6]. Measurements inside the AKR source have shown that the radiation originates in ﬁeld-aligned density depletions with transverse dimensions of the order of 100 km, ﬁlled by hot and tenuous plasmas [7, 10, 11, 19]. Theory considering relativistic eﬀects on the wave dispersion relation [Pritchett, 51, 52] and small scale gradients [see, e.g., Le Qu´eau and Louarn, 26] appears to be necessary to explain all the details of the wave generation and subsequent propagation from the source cavity.

12 Propagation Analysis of Electromagnetic Waves

299

Since AKR propagates at frequencies higher than all the characteristic frequencies of the plasma medium (above the plasma frequency and the electron cyclotron frequency) it can leave the source region in one of the two basic propagation modes: R-X (right-handed/extraordinary) and L-O (lefthanded/ordinary) [67]. The two modes can be recognized using measurements of phase shifts between the diﬀerent components of the electric and/or magnetic ﬁeld ﬂuctuations. Kaiser et al. [22] analyzed direct measurements of the polarization sense made by multiple electric antennae of Voyager 1 and 2. Supposing that the satellites are always in the northern magnetic hemisphere they found R-X mode in more than 80% of cases. Analysis of wave propagation directions was used in many AKR studies to localize the source region or to verify the generation mechanism. Gurnett [15], using Imp 6 and Imp 8 data, ﬁrst estimated the direction of AKR propagation analyzing the spin pattern of a single electric antenna. At a distance of about 30 RE , he found the wave vector within 6◦ from the Earthward direction. Kurth et al. [24] conﬁrmed his results by a detailed study based on large Hawkeye-1 and Imp-6 data sets. To ﬁnd the propagation direction they extended the spin pattern method using the least-squares ﬁt of the modulation envelope of the electric antenna. However, they had to work with 1-hour averages to achieve suﬃcient angular resolution. Similar results were also obtained by Alexander and Kaiser [1] who analyzed the RAE-2 recordings of lunar occultations of the Earth at radio frequencies to locate the emission region in three dimensions. Gurnett et al. [16] proposed to use a phase interferometry method using two-point measurements of AKR waveforms onboard ISEE-1 and ISEE-2 satellites, and Baumback et al. [4] found by this method that the AKR source has a diameter between 10 and 20 km. James [21] applied the spin-pattern method to the data of ISIS-1 sounder receiver. He measured wave normals declined by 90◦ –140◦ from the terrestrial magnetic ﬁeld, and in a ray-tracing study he inferred originally downward propagation and a subsequent reﬂection. Calvert [6] developed a direction ﬁnding method based on DE-1 onboard correlation measurements. He used signals from two orthogonal electric dipoles, and with a model of their phase diﬀerence with respect to the spin phase he obtained the wave-normal direction. This method was also used by Mellott et al. [37] who studied diﬀerent propagation of R-X and L-O mode emissions. Several studies using the Viking data have again used a simple spin-pattern method. For instance, de Feraudy et al. [7] found that the spin-pattern phase abruptly changes by about 80◦ at the boundary of the AKR source region. Morioka et al. [38] used ﬁve-component measurement of the Akebono satellite to calculate the Poynting ﬂux vector in the satellite frame. With onboard narrow-band receivers they studied phase relations between the ﬁeld components in a band of 100 Hz inside an AKR emission. Gurnett et al. [13] and Mutel et al. [39, 40] implemented the very long baseline interferometer (VLBI) technique to determine locations of individual AKR bursts. They used a triangulation method based on diﬀerential delays from cross-correlated wide-band electric ﬁeld waveforms recorded by

300

O. Santol´ık and M. Parrot

the WBD instruments on board the four Cluster spacecraft and conﬁrmed that the AKR bursts are generally located above the auroral zone with a strong preference for the evening sector. Multi-component measurements of the MEMO (Mesures Multicomposantes des Ondes) instrument onboard the Interball 2 spacecraft have been analyzed by Lefeuvre et al. [32] and Parrot et al. [46], showing that the wave normals of the right-hand quasi-circularly polarized (R-X mode) waves have wave vectors inclined by approximately 30◦ from the direction of the terrestrial magnetic ﬁeld. Santol´ık et al. [58] showed for another AKR case measured by the same instrument at an altitude of 3 RE that the analysis of wave propagation indicates a source region located at an altitude of 1.2 RE above the northern auroral zone. Schreiber et al. [65] performed, based on similar Interball-2 measurements of the MEMO and POLRAD [17] instruments, a three-dimensional ray tracing study using the analysis of wave-normal directions at the spacecraft position. Rays traced back toward the sources imply the existence of two diﬀerent large active regions, as seen at the same time by the ultra-violet camera on board the Polar spacecraft. Analysis presented in this chapter is also based on the data set of the MEMO instrument. The chapter is organized as follows. A short tutorial description of the analysis methods is given in Sect. 12.2. Subsection 12.2.1 introduces methods based on the approximation of the single wave-normal direction. Subsection 12.2.2 then brieﬂy describes methods of estimation of the wave distribution function. We show that this method can be naturally extended to estimate also a distribution of energy between diﬀerent wave modes. Section 12.3 follows showing results of analysis of AKR measured by multiple antennas onboard the Interball 2 spacecraft.

12.2 Analysis Methods The multi-component measurements of the wave magnetic and electric ﬁelds allow us to determine, for example, the average Poynting ﬂux. Supposing the presence of a single plane wave, the direction of the wave vector can be determined. For a more complex wave ﬁeld we can estimate a continuous distribution of wave energy with respect to the wave-vector direction (wave distribution function). These results are useful for the localization of sources of observed emissions. 12.2.1 Plane Wave Methods Supposing the presence of a single plane wave at a frequency f with a wave vector k, the magnetic ﬁeld B as a function of time t and position x can be written ! B(t, x) = B0 + B(f, k)exp i(2πft − k · x) , (12.1)

12 Propagation Analysis of Electromagnetic Waves

301

where B0 is the ambient stationary magnetic ﬁeld (terrestrial magnetic ﬁeld in the magnetosphere), means the real part, and B is the “magnetic vector complex spectral amplitude” for a given frequency f and wave vector k. Under the same circumstances, similar expression can be used for the ﬂuctuating electric ﬁeld E using the “electric vector complex spectral amplitude” E, and similarly also for the “current density complex spectral amplitude” J and √ “charge density complex spectral amplitude” . Using the SI units with 1/ ε0 µ0 = c (speed of light), the Maxwell’s equations can be written for the complex spectral amplitudes (12.2) k × B = i µ0 J − 2πf E/c2 , k × E = 2πf B , k · E = /ε0 ,

(12.3) (12.4)

k · B = 0.

(12.5)

Equation (12.3) (Faraday’s law) implies that B is always perpendicular to both wave vector k and E, k·B =0 (12.6) E ·B =0

(12.7)

Note that the ﬁrst of these conditions (12.6) is equivalent also to the fourth Maxwell’s equation (12.5). If we now write it in Cartesian coordinates, where B = (B1 , B2 , B3 ) ,

(12.8)

and multiply (12.6) successively by three Cartesian components of the complex conjugate B∗ we obtain a set of six mutually dependent real equations S11 S12 S13 S12 S22 S23 k1 S13 S23 S33 · k2 = 0 , A·k = (12.9) 0 −S12 −S13 k 3 S12 0 −S23 S13 S23 0 where means the real part, means the imaginary part, and the components of the Hermitian magnetic spectral matrix Sij are obtained from the three Cartesian components of the magnetic vector complex amplitude using the relation i, j = 1 . . . 3 . (12.10) Sij = Bi Bj∗ , Note that the homogeneous set of equations (12.9) can be multiplied by any real coeﬃcient. Consequently, this set cannot be used to determine the modulus of the unknown vector k. It only can determine the direction of this vector. Note also that the set of equations (12.9) naturally contains only two independent real equations corresponding to the single complex equation

302

O. Santol´ık and M. Parrot

(12.6). The importance of this expansion, however, becomes evident when it is used with experimental data. Using the experimentally measured multi-component signals of the wave magnetic ﬁeld ˆ = (B ˆ2 , B ˆ3 ) ˆ1 , B (12.11) B we can use spectral analysis methods (for example, fast Fourier transform or wavelet analysis) to estimate, at a given frequency, the components of ˆ and, subsequently, the the “magnetic vector complex spectral amplitude” B magnetic spectral matrix, Sˆij = Bˆi Bˆj∗ ,

i, j = 1 . . . 3 ,

(12.12)

where means average value. The homogeneous set of six equations (12.9) can be then rewritten, ˆ ·κ ˆ =0, A

(12.13)

ˆ is composed of the superposed imaginary and real parts of where the matrix A the experimental spectral matrix Sˆij (instead of the idealized spectral matrix ˆ is an unknown unit vector deﬁning the estimate of the wave vector Sij ), and κ direction, ˆ k| ˆ . ˆ = k/| κ (12.14) Using Cartesian coordinates connected to the principal axis of symmetry of the plasma medium where the waves propagate (the direction of the ambient stationary magnetic ﬁeld B 0 ), the wave vector direction can be deﬁned by two angles θ and φ, where θ is the deviation from the B 0 direction and φ is an azimuth centered, for instance, to the plane of the local magnetic meridian (see Fig. 12.1). The wave vector direction then reads ˆ = (sin θ cos φ, sin θ sin φ, cos θ) . κ

(12.15)

ˆ thus reduces to two real unknowns θ and The unknown unit vector κ φ. As a consequence, the system (12.13) is over-determined, containing six

Fig. 12.1. Cartesian coordinate system for the determination of the wave vector ˆ = k/|k| direction κ

12 Propagation Analysis of Electromagnetic Waves

303

equations for two unknowns. Generally, these 6 equations are independent. This is diﬀerent compared to the case of the ideal set of equations (12.9). ˆ is composed of experimental data which The reason is that the matrix A can contain natural and/or experimental noise and do not necessarily exactly correspond to an ideal plane wave. Since we only have two unknowns, a subset of any two independent equations picked up from the set (12.9) is suﬃcient to obtain a unique solution for θ and φ. This is the basis of several analysis methods. The method of Means [36] is based on imaginary parts of three cross-spectra and the procedure is equivalent to solving any two of the last three equations in (12.13). The method of Samson and Olson [55] (their equation 11) is equivalent to ﬁnding a unique solution from another subset of equations selected in (12.13). The method of McPherron et al. [35] uses the ﬁrst three equations and ﬁnds a unique solution using the eigenanalysis of the real part of the spectral matrix. However, the choice of the subset of equations is rather arbitrary and for a diﬀerent subset we do not generally obtain the same result from given experimental data. Other methods thus attempt to estimate an “average” solution of the entire set (12.9) using diﬀerent techniques. Samson [54], again using the eigenanalysis, presented methods of decomposition of the entire complex spectral matrix. Santol´ık et al. [64] used a singular value decomposition (SVD) technique to estimate a solution of the entire set of equations (12.13) in the ˆ “least-squares” sense, decomposing the matrix A ˆ = U · W · VT , A

(12.16)

where U is a matrix 6 × 3 with orthonormal columns, W is a diagonal matrix 3 × 3 of three non-negative singular values, and VT is a matrix 3 × 3 with orthonormal rows. Note that the SVD algorithm can often be found in nuˆ is then directly merical libraries [e.g., 50]. The “least-squares estimate” for κ found as the row of VT corresponding to the minimum singular value at the diagonal of W. The results of the plane-wave analysis often allow a straightforward interpretation of results. This was found useful, for example, in the analysis of sub-auroral ELF hiss emissions from the measurements of the Aureol 3 and Freja spacecraft [33, 62]. These diﬀerent methods often provide us also with estimates of the validity of the initial assumption of the presence of a single plane wave. Diﬀerent deﬁnitions of such an estimator (“degree of polarization”, “polarization percentage” or “planarity”) have been introduced, based on diﬀerent descriptions of the coherence of the magnetic components and their conﬁnement to a single polarization plane [49, 54, 64]. Similar techniques can also allow us to estimate the sense of the magnetic polarization with respect to the ambient stationary magnetic ﬁeld B 0 . This has been used, for example, to analyze electromagnetic emissions in the auroral region by Lefeuvre et al. [30, 31], Santol´ık et al. [59], and Santol´ık and Gurnett [56], using the data of the Aureol 3, Interball 2, and Polar spacecraft, respectively.

304

O. Santol´ık and M. Parrot

The above mentioned SVD technique can also be used with both the measured magnetic and electric components. In that case, an “average” solution to an over-determined set of 36 equations derived from equation (12.3) is esˆ , i.e., to distinguish timated. This allows us to determine also the sign of κ between the two antiparallel wave vector directions [for more details, see 64]. This technique also allows us to estimate the validity of the plane-wave assumption, but this time it is deﬁned as a measure of closeness of the observed wave ﬁelds to equation (12.3). This determination of the “electromagnetic planarity” was, for example, used in [57] to estimate the dimension of the source of chorus emissions from the data of the four Cluster spacecraft. 12.2.2 Wave Distribution Function The concept of wave distribution function is necessary when the wave ﬁeld is more complex, for example when waves from multiple distant sources are simultaneously detected. The wave distribution function (WDF) is deﬁned as a continuous distribution of wave energy with respect to the wave-vector direction; see the review by Storey [69]. It was ﬁrst introduced by Storey and Lefeuvre [70]. The theoretical relation of the WDF to the experimentally measurable spectral matrix has been called the WDF direct problem. Supposing a continuous distribution of elementary plane waves at a frequency f having no mutual coherence and a narrow bandwidth ∆f , the relationship between the spectral matrix Sij (f ) and the WDF Gm (f, θ, φ) is given by 4 Sij (f ) = (12.17) amij (f, θ, φ) Gm (f, θ, φ) d2 κ , m

where m represents the diﬀerent simultaneously present wave modes. The integration is carried out over the full solid angle of wave-normal directions κ, and for a given wave mode m the integration kernels amij are calculated from ∗ (f, θ, φ) ξmi (f, θ, φ) ξmj , (12.18) amij (f, θ, φ) = ∆f um (f, θ, φ) where ξmi , ξmj are complex spectral amplitudes of the ith and jth elementary signals of the wave electric or magnetic ﬁelds, and um is the energy density. All these quantities correspond to an elementary plane wave propagating in a mode m with a normal direction deﬁned by θ and φ. f represents the Dopplershifted frequency in the spacecraft frame. The complex spectral amplitudes of the wave electric or magnetic ﬁelds can be calculated by considering the physical properties of the medium. This calculation requires the knowledge of the theoretical solutions to the wave dispersion relation. Characteristics of the particular wave experiment should also be taken into account. The theory of the WDF direct problem for the cold-plasma approximation has been developed by Storey and Lefeuvre [71, 72], and revisited by Storey [68]. This basic theory has been used by Lefeuvre [27], Lefeuvre and

12 Propagation Analysis of Electromagnetic Waves

305

Delannoy [28] and Delannoy and Lefeuvre [8] to develop practical methods for estimation of the WDF from the spacecraft measurements (the WDF inverse problem). Using a slightly diﬀerent deﬁnition of the WDF and abandoning the explicit dependence of the WDF on the wave frequency, Oscarsson and R¨onnmark [42, 43] and Oscarsson [41] introduced the hot plasma theory into the WDF reconstruction techniques. With the wave-vector dependent WDF they also introduced the Doppler eﬀect in a natural way. Santol´ık and Parrot [60] used the hot plasma theory for the frequency-dependent WDF and further investigated the inﬂuence of the Doppler eﬀect in this more complex situation. Santol´ık and Parrot [63] compared diﬀerent techniques for resolution of the WDF inverse problem, mainly based on the minimization of the least-squares type merit function, in the context of the plane wave estimates. The WDF techniques have been used in numerous studies with both ground based and spacecraft data. For instance, Lefeuvre and Helliwell [29], Parrot and Lefeuvre [45], Hayakawa et al. [18], and Storey et al. [73] used the multi-component measurements of the GEOS spacecraft to characterize the WDF of the ELF chorus and hiss emissions on the equatorial region. Based on the data of the Aureol 3, Akebono, and Freja spacecraft, Lefeuvre et al. [33], Kasahara et al. [23] and Santol´ık and Parrot [61, 63], respectively, estimated the WDF of the down-coming ELF hiss in the sub-auroral and auroral regions. Oscarsson et al. [44] compared the diﬀerent reconstruction schemes for the data of the Freja spacecraft. The up-going funnel-shaped auroral hiss has been investigated by Santol´ık and Gurnett [56] using the WDF analysis of measurements of the Polar spacecraft. Parrot et al. [46] estimated the WDF for auroral kilometric radiation observed on board the Interball spacecraft, compensating at the same time for the a priori unknown experimentally induced phase between the electric and magnetic signals. Simultaneous WDF estimation of the Z-mode and the whistler mode in the auroral region has been done by Santol´ık et al. [59], based on the data of the Interball 2 spacecraft.

12.3 Analysis of Auroral Kilometric Radiation The MEMO instrument on board the Interball 2 spacecraft [32] had two basic modes of measurement: burst mode and survey mode. During the survey mode, low-resolution overview spectrograms were recorded. In its burst mode, the instrument measured waveforms of several components of the electric and magnetic ﬁeld ﬂuctuations in three frequency bands from 50 Hz to 200 kHz. In the high-frequency band (30–200 kHz) which is relevant to this study, the device recorded waveforms measured at the same time by three magnetic antennas and one electric antenna. On 28 January 1997, between 1950 and 2125 UT, the Interball 2 spacecraft was located on the night side and moved over the northern auroral region at altitudes of 2.6-3 RE . During this time interval, the MEMO instrument recorded a highly structured emission of auroral kilometric radiation at frequencies

306

O. Santol´ık and M. Parrot

2125 Polar 2100 2030 2000 1952

1952 Interball 2 2125

60o 30o 0o h 14h 16h 18h 20h 22h 0h2

Fig. 12.2. Comparison of AKR observed by the Interball 2 and Polar spacecraft on 28 January 1997 between 1950 and 2125 UT. Left: Time-frequency spectrograms recorded by the MEMO instrument onboard Interball 2 (bottom) and by the SFR analyzer of the PWI instrument onboard Polar. The same time interval and frequency range is used for both spectrograms. Coordinates of the two spacecraft are given on the bottom of the spectrograms. Over-plotted white lines show the local electron cyclotron frequency. Right: Portions of orbits of the two spacecraft in the corresponding time interval and their projections along the magnetic ﬁeld lines onto the Earth’s surface. The arrows pointing out of the Interball-2 orbit show the average wave-vector directions for two burst-mode intervals at a frequency of 80 kHz (see Fig. 12.5)

between 30 kHz and more than 270 kHz, as reported in [46]. At the same time, the Polar spacecraft also moved over the northern auroral zone but at approximately twice the altitude and on the evening side. The PWI instrument on board Polar [for details of the instrument see, Gurnett et al., 14] observed the same AKR emission. Comparison of observations of the Interball 2 and Polar spacecraft is given in Fig. 12.2 on the left. The strong AKR emission between 2030 and 2125 UT is clearly seen on both spectrograms, demonstrating the global character of AKR. Some detailed time-frequency

12 Propagation Analysis of Electromagnetic Waves

307

Fig. 12.3. Spectrograms calculated from one snapshot of burst-mode MEMO waveform data recorded on 28 January 1997 after 2107:15.48 UT. The spectrograms show the electric component in the frequency interval 0–220 kHz with a frequency resolution of 260 Hz (left) and detailed analysis of pronounced structures around 32 kHz (right) with a frequency resolution of 52 Hz. Black rectangle represents data used later for wave propagation analysis (Fig. 12.4). The local electron cyclotron frequency is 23.5 kHz

features are, however, diﬀerent. Some of those diﬀerences may be attributed to diﬀerent sensitivities of the two instruments as a function of frequency, but most of them are probably owing to diﬀerent positions of the two spacecraft. As their local time is diﬀerent, the two spacecraft can observe AKR emitted from diﬀerent portions of the auroral oval. This also concerns slight timing diﬀerences for the onset of intense AKR around 2030 UT. During this orbit, several intervals of burst-mode data were recorded by the MEMO instrument on board the Interball 2 spacecraft. One electric and three magnetic components were always sampled at 533.3 kHz during 0.32 sec. Figure 12.3 shows the power spectrograms calculated from the electric-ﬁeld data measured during one of those burst-mode intervals. We can see that, at short time scales, the emission still has a complex time-frequency structure with many spectral lines observed at slowly drifting frequencies. Detailed analysis of the middle portion of this burst-mode interval has been done using the available data of one electric and three magnetic ﬁeld components. The wave distribution function (WDF), representing the distribution of the wave energy for diﬀerent wave-vector directions and wave-propagation

308

O. Santol´ık and M. Parrot

Fig. 12.4. Detailed analysis of the multicomponent MEMO data in the frequency band of 0.4 kHz around 32.1 kHz, between oﬀsets 0.19 and 0.22 sec (see Fig. 12.3). The wave distribution function (WDF) is plotted in the coordinate system connected to the Earth’s magnetic ﬁeld B0 and to the local magnetic meridian (Fig. 12.1). The polar diagrams represent the energy density as a function of angles θ and φ, on the left for downgoing waves, and on the right for upgoing waves, on the top for the R-X mode and on the bottom for the L-O mode

modes, is shown in Fig. 12.4. We use a narrow frequency band around the maximum of the power-spectral density of the intense spectral line observed around 32 kHz. A cold plasma model is used to calculate theoretical wave spectral densities, necessary for simultaneous WDF estimation in both R-X and L-O propagation modes and for both hemispheres of wave-normal directions (upgoing and downgoing with respect to the direction of the ambient magnetic ﬁeld). The method of “discrete regions” was used to estimate the WDF [63]. It is based on least-squares optimization of the non-negative WDF values in the total number of 824 discrete regions, homogeneously covering

Fig. 12.5. Wave distribution function for the upgoing waves in the R-X mode propagating in the frequency band of 2.2 kHz centered at 80.1 kHz. The same coordinate system as in Fig. 12.4 is used. Two 0.32-s burst-mode snapshots have been used: (a) time interval starting at 2037:23.58 UT; (b) time interval starting at 2107:15.48 UT. Overplotted are plane wave estimates: triangles for the method of McPherron et al. [35], squares for the method of Means [36], diamonds for the SVD method [64]

the 4π solid angle of wave normal direction in the two propagation modes. The results show that nearly all the wave energy is found in the upgoing R-X mode. Within statistical uncertainties of our analysis, we can neither exclude nor conﬁrm the presence of a small fraction of the L-X mode. Figure 12.5 shows a comparison of obtained results of wave propagation analysis for two diﬀerent time intervals separated by 30 minutes. The second of these time intervals is the same as in Fig. 12.3, and both of them were analyzed using diﬀerent methods by Parrot et al. [46]. For the WDF estimation in Fig. 12.5 we use the model of Gaussian peaks [63]. This model describes concentration of wave energy near one or several wave-normal directions. Information about the WDF can then be reduced to a set of the peak directions and accompanied by the respective energy densities and parameters describing the degree of concentration of the wave energy near the respective directions (peak widths). In this case we model the WDF G as a single peak represented by a Gaussian function in 2D space of wave-normal directions (or, equivalently, on the surface of a unit sphere),

2 [1 − cos θ0 cos θ − sin θ0 sin θ cos(φ0 − φ)] Γ exp − , G(θ, φ) = π∆2 ∆2 (12.19) where Γ is the peak energy density, ∆ is the peak width, and θ0 and φ0 deﬁne the central direction. These free parameters are optimized by a nonlinear least-

310

O. Santol´ık and M. Parrot

squares method to obtain the best model (12.17) for the given experimental spectral matrix. The results show that the emission propagates at oblique angles from the Earth’s magnetic ﬁeld, at diﬀerent azimuth angles φ for the two time intervals. The obtained WDF estimates show large peak widths, (∆ = 20◦ for Fig. 12.5a and ∆ = 28◦ for Fig. 12.5b). The plane wave results are localized within the 75% intensity level compared to the peak maximum. The diﬀerences between the diﬀerent plane wave estimates are larger for a larger width of the WDF peak. The corresponding angular sizes of the global sources of observed AKR have been investigated by Schreiber et al. [65] leading to the suggestion that we observe superposed waves from many elementary sources. The positions of these sources roughly corresponded to active auroral regions remotely sensed by the Ultraviolet Imager (UVI) instrument onboard the Polar spacecraft.

12.4 Conclusions We have shown that wave propagation analysis based on multicomponent spacecraft data can be a useful tool for investigation of high-frequency wave phenomena. Several analysis methods have been described, some using the plane wave approach and others based on a general continuous distribution of wave energy for diﬀerent propagation directions. This second approach has been shown to be useful also for recognition of the propagation mode of observed waves. Observations of AKR by the MEMO instrument on board the Interball-2 spacecraft have been analyzed using these methods. We have shown that the observed highly structured AKR emission propagates predominantly in the R-X mode. The propagation analysis has allowed us to trace the waves back to the source regions which, consistently with previously published results, corresponded to large active regions on the auroral oval. These examples demonstrate the value of using the described analysis methods in future in situ investigations of high frequency waves in geospace and in the solar wind.

Acknowledgments We thank D. A. Gurnett of the University of Iowa, PI of the Polar PWI instrument for the SFR data used in Fig. 12.2. We thank J. D. Menietti of the University of Iowa for useful discussions. This work was supported by the ESA PECS contract No. 98025 and by the GACR grant 202/03/0832.

References [1] Alexander, J.K. and M.L. Kaiser: J. Geophys. Res. 81, 5948, 1976. [2] Arthur, C.W., R.L. McPherron, and J.D. Means: Radio Sci. 11, 833, 1976.¨ o

12 Propagation Analysis of Electromagnetic Waves

311

[3] Baumback, M.M. and W. Calvert: Geophys. Res. Lett. 5, 857, 1978. [4] Baumback, M.M., D.A. Gurnett, W. Calvert, et al.: Geophys. Res. Lett. 13, 1105, 1986. [5] Benediktov, E.A., et al.: Kossm. Issled. 3, 614, 1965. [6] Calvert, W.: Geophys. Res. Lett. 12, 381, 1985. [7] de Feraudy, H., et al.: Geophys. Res. Lett. 14, 511, 1987. [8] Delannoy, C. and F. Lefeuvre: Comp. Phys. Comm. 40, 389, 1986. [9] Dunckel, N., et al.: J. Geophys. Res. 75, 1854, 1970. [10] Ergun, R.E., et al.: Astrophys. J. 538, 456, 2000. [11] Ergun, R.E., et al.: Geophys. Res. Lett. 25, 2061, 1998. [12] Grard, R.: Ann. Geophys. 24, 955, 1968. [13] Gurnett, D.A., et al.: Ann. Geophys. 19, 1259, 2001. [14] Gurnett, D.A., et al.: Space Sci. Rev. 71, 597, 1995. [15] Gurnett, D.A.: J. Geophys. Res. 79, 4227, 1974. [16] Gurnett, D.A., et al.: Space Sci. Rev. 23, 103, 1979. [17] Hanasz, J., et al.: J. Geophys. Res. 106, 3859, 2001. [18] Hayakawa, M., M. Parrot, and F. Lefeuvre: J. Geophys. Res. 91, 7989, 1986. [19] Hilgers, A.: Geophys. Res. Lett. 19, 237, 1992. [20] Huﬀ, R.L., et al.: J. Geophys. Res. 93, 11445, 1988. [21] James, H.G.: J. Geophys. Res. 85, 3367, 1980. [22] Kaiser, M.L., et al.: Geophys. Res. Lett. 5, 857, 1978. [23] Kasahara, Y., et al.: J. Geomag. Geoelectr. 47, 509, 1995. [24] Kurth, W.S., M.M. Baumback, and D.A. Gurnett: J. Geophys. Res. 80, 2764, 1975. [25] LaBelle, J. and R.A. Treumann: J. Geophys. Res. 97, 13789, 1992. [26] Le Qu´eau, D. and P. Louarn: Planet. Space Sci. 44, 211, 1996. [27] Lefeuvre, F.: Analyse de champs d’ondes ´ electromagn´etiques al´eatoires observ´ees dans la magn´etosph` ere, ` a partir de la mesure simultan´ ee de leurs six composantes. Doctoral thesis, Univ. of Orl´eans, Orl´eans, France, 1977. [28] Lefeuvre, F. and C. Delannoy: Ann. Telecommun. 34, 204, 1979. [29] Lefeuvre, F. and R.A. Helliwell: J. Geophys. Res. 90, 6419, 1985. [30] Lefeuvre, F., et al.: Ann. Geophys. 4, 457, 1986. [31] Lefeuvre, F., et al.: Ann. Geophys. 5, 251, 1987. [32] Lefeuvre, F., et al.: Ann. Geophys. 16, 1117, 1998. [33] Lefeuvre, F., et al.: J. Geophys. Res. 97, 10601, 1992. [34] Louarn, P., et al.: J. Geophys. Res. 95, 5983, 1990. [35] McPherron, R.L., C.T. Russel, and P.J. Coleman, Jr.: Space Sci. Rev. 13, 411, 1972. [36] Means, J.D.: J. Geophys. Res. 77, 5551, 1972. [37] Mellott, M.M., R.L. Huﬀ, and D.A. Gurnett: Geophys. Res. Lett. 12, 479, 1985. [38] Morioka, A., H. Oya, and A. Kobayashi: J. Geomagn. Geoelectr. 42, 443, 1990. [39] Mutel, R.L., D.A. Gurnett, and I.W. Christopher: Ann. Geophys. 22, 2625, 2004. [40] Mutel, R.L., et al.: J. Geophys. Res. 108, 1398, doi: 10.1029/2003JA010011, 2003. [41] Oscarsson, T.: J. Comput. Phys. 110, 221, 1994. [42] Oscarsson, T. and K.R¨ onnmark: J. Geophys. Res. 94, 2417, 1989. [43] Oscarsson, T. and K. R¨ onnmark: J. Geophys. Res. 95, 21,187, 1990.

312

O. Santol´ık and M. Parrot

[44] Oscarsson, T., G. Sternberg, and O. Santol´ık: Phys. Chem. Earth (C) 26, 229, 2001. [45] Parrot, M. and F. Lefeuvre: Ann. Geophys. 4, 363, 1986. [46] Parrot, M., et al.: J. Geophys. Res. 106, 315, 2001. [47] Parrot, M., et al.: Ann. Geophys. 21, 473, 2003. [48] Parrot, M., et al.: Ann. Geophys. 22, 2597, 2004. [49] Pin¸con, J.L., Y. Marouan, and F. Lefeuvre: Ann. Geophys. 10, 82, 1992. [50] Press, W.H., et al.: Numerical Recipes. Cambridge Univ. Press, New York, 1992 [51] Pritchett, P.L., et al.: J. Geophys. Res. 107, 1437, doi: 10.1029/2002JA009403, 2002. [52] Pritchett, P.L.: J. Geophys. Res. 89, 8957, 1984. [53] Roux, A., et al.: J. Geophys. Res. 98, 11657, 1993. [54] J.C. Samson: Geophys. J. R. Astron. Soc. 34, 403, 1973. [55] Samson, J.C. and J.V. Olson: Geophys. J. R. Astron. Soc. 61, 115, 1980. [56] Santol´ık, O. and D.A. Gurnett: Geophys. Res. Lett. 29, 1481, doi: 10.1029/ 2001GL013666, 2002. [57] Santol´ık, O., et al.: Geophys. Res. Lett. 31, L02801, doi: 10.1029/ 2003GL018757, 2004. [58] Santol´ık, O., et al.: J. Geophys. Res. 106, 13191, 2001. [59] Santol´ık, O., et al.: J. Geophys. Res. 106, 21137, 2001. [60] Santol´ık, O. and M. Parrot: J. Geophys. Res. 101, 10639, 1996. [61] Santol´ık, O. and M. Parrot: J. Geophys. Res. 103, 20469, 1998. [62] Santol´ık, O. and M. Parrot: J. Geophys. Res. 104, 2459, 1999. [63] Santol´ık, O. and M. Parrot: J. Geophys. Res. 105, 18885, 2000. [64] Santol´ık, O., M. Parrot, and F. Lefeuvre: Radio. Sci. 38, 1010, doi: 10.1029/ 2000RS002523, 2003. [65] Schreiber, R., et al.: J. Geophys. Res. 107, 1381, doi: 10.1029\-/\-2001\ -JA009061, 2002. [66] Shawhan, S.D.: Space Sci. Rev. 10, 689, 1970. [67] Stix, T.H.: Waves in Plasmas. Am. Inst. of Phys., New York, 1992. [68] Storey, L.R.O.: Ann. Geophys. 16, 651, 1998. [69] Storey, L.R.O.: The measurement of wave distribution functions. In: M.A. Stuchly, editor, Modern Radio Science 1999, page 249, Oxford University Press, Oxford, 1999. [70] Storey, L.R.O. and F. Lefeuvre: Theory for the interpretation of measurements of a random electromagnetic wave ﬁeld in space. In M.J. Rycroft and R.D. Reasenberg (Eds.), Space Research XIV, page 381, Akademie-Verlag, Berlin, 1974. [71] Storey, L.R.O. and F. Lefeuvre: Geophys. J. R. Astron. Soc. 56, 255, 1979. [72] Storey, L.R.O. and F. Lefeuvre: Geophys. J. R. Astron. Soc. 62, 173, 1980. [73] Storey, L.R.O., et al.: J. Geophys. Res. 96, 19469, 1991. [74] Wu, C.S. and L.C. Lee: Astrophys. J. 230, 621, 1979.

13 Phase Correlation of Electrons and Langmuir Waves C.A. Kletzing1 and L. Muschietti2 1

2

Department of Physics and Astronomy, University of Iowa [email protected] Space Sciences Lab., University of California [email protected]

Abstract. Multiple spacecraft observations have conﬁrmed the ubiquitous nature of Langmuir waves in the presence of auroral electrons. The electrons show variations consistent with bunching at or near the plasma frequency. Linear analysis of the interaction of a ﬁnite Gaussian packet of Langmuir waves shows that there are two components to the perturbation to the electron distribution function, one in-phase (or 180◦ out-of-phase) with respect to the wave electric ﬁeld called the resistive component and one which is 90◦ (or 270◦ ) out-of-phase with respect to the electric ﬁeld. For small wave packets, the resistive perturbation dominates. For longer wave packets, a non-linear analysis is appropriate which suggests that the electrons become trapped and the reactive phase dominates. Rocket observations have measured both components. The UI observations diﬀer from those of the UC Berkeley observations in that a purely reactive phase bunching was observed as compared to a predominantly resistive perturbation. The resistive phase results of the UC Berkeley group were interpreted as arising from a short wave packet. The UI observations of the reactive phase can be explained by either a long, coherent train of Langmuir waves or that the narrower velocity response of the UI detectors made it possible to capture only one side of the reactive component of the perturbed distribution function for a short wave packet in the linear regime. Future wave-particle correlator experiments should be able to resolve these questions by providing more examples with better velocity space coverage.

Key words: Electron bunching, Langmuir waves, resistive and reactive bunching phases, phase correlators

13.1 Introduction The precipitation of auroral electrons provides an example of a beam-plasma interaction which generates Langmuir waves from the free energy in the electrons. The resulting waves play several important roles in the Earth’s auroral ionosphere. First, the waves Landau damp on the thermal electron population C.A. Kletzing and L. Muschietti: Phase Correlation of Electrons and Langmuir Waves, Lect. Notes Phys. 687, 313–337 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

314

C.A. Kletzing and L. Muschietti

and thereby form a direct conduit for energy exchange between the auroral electron beam and the thermal electrons. Several authors have speculated that Langmuir waves play a signiﬁcant role in establishing the electron temperature in the auroral ionosphere [3, 15]. In addition to heating electrons, the Langmuir/upper hybrid waves radiate away some of their energy into electromagnetic radiation, which can serve for remote sensing of auroral plasma processes from ground level and from satellites. For example, auroral roar is an EM emission observed near 2–3 and 4–4.5 MHz at ground level [14, 30] and from satellites [1, 12]. Understanding these auroral wave emissions is important not only to fully understand terrestrial aurora and related phenomena, but also because they shed light on analogous emission processes elsewhere in the solar system and beyond. For example, the generation of auroral roar is similar to that of terrestrial continuum radiation, which is generated via mode conversion of upper hybrid waves at the plasmapause, and possibly continuum radiations at other planets as well. Solar type III radiation results from mode conversion of Langmuir waves in the solar wind, and recent observations of structured type III emission [Reiner et al., 23, 24] indicate the signiﬁcance of frequency structure in the causative Langmuir waves. High frequency (HF) electric ﬁeld observations in the topside auroral ionosphere, for which the plasma frequency is typically greater than the electron cyclotron frequency, have revealed plasma waves in the range fpe ≤ f ≤ fuh ever since the earliest measurements [2, 29]. Simultaneous wave and electron distribution measurements have shown that the waves are excited by Landau resonance but that temporal variation of the distribution function or wave refraction from vertical density gradients can limit wave amplitudes as shown by McFadden et al. [19]. Many examples of waves near fpe were observed using Aureol/ARCAD 3 satellite wave receivers and the free-energy source was identiﬁed as the electrons [Beghin et al., 3]. Although rocket-borne receivers have detected HF waves at E-region altitudes attributed to generation by secondary electrons [Kelley and Earle, 13], most observations pertain to altitudes from 300 km and up, where the waves have amplitudes ranging from less than 1 mV/m to as large as 1 V/m [4, 6, 8, 17, 18, 28] These waves are highly bursty in time, sometimes lasting as little as 1 ms but sometimes appearing continuous for ∼1s. Recent high-resolution experiments reveal that they can have complex frequency structure [17, 18, 25]. Other investigators have detected plasma waves through particle-particle correlator techniques. Using a rocket-borne electron detector of large geometric factor, ∼5% modulation at frequencies 4.2–5.6 MHz was found during a 7-second interval when the rocket passed the boundary of two oppositely directed Birkeland sheet currents [27]. Strong modulation (∼30%) was observed at 2.65 MHz of 4–5 keV electrons, corresponding to energies at which a positive slope was observed in the perpendicular and parallel velocity distribution functions [Gough and Urban, 11]. Rocket measurements in which 1.4 MHz ﬂuctuations were detected in the 7.5 keV electrons, just below the electron beam energy, were reported to

13 Langmuir Wave-Electron Phase Correlation

315

occur simultaneously with a positive slope in the electron distribution function [Gough et al., 10]. Linear theory explains the existence and parallel polarization of auroral Langmuir waves [see, e.g., Nicholson, 22]. Theory and simulations have shown that Langmuir waves in the auroral plasma most likely do not develop into strong turbulence but that observed non-linear features in amplitudes and modulations of the waves are consistent with nonlinear interactions between the Langmuir waves and ion acoustic waves [Newman et al., 21]. More recently, the diﬀerence between Langmuir wave-electron interactions both at rocket altitudes and at the altitude of the Freja spacecraft have been studied, ﬁnding that standard quasi-linear diﬀusion theory does not hold for large amplitude Langmuir waves at Freja altitudes, but should hold at lower altitudes [Sanbonmatsu et al., 26]. Correlating waves and particles directly probes the physics of their interaction by providing a superior picture of the microphysics compared to statistically associating an unstable feature of the distribution function with the presence of waves. Elementary theory implies that if the electrons and the waves are exchanging energy, the electrons will have an oscillatory component at a velocity equal to the phase velocity of the waves. Identifying the velocity at which the electron distribution function has this oscillatory component determines the wave phase velocity and therefore the wave number. Measuring the phase of this oscillatory component of the electron distribution function relative to the wave electric ﬁeld yields further information. The phase bunching splits into two pieces: 1) the resistive component, in phase with the electric ﬁeld indicating wave-electron energy exchange; and 2) the reactive component, 90◦ out of phase with the electric ﬁeld indicating trapping [Nicholson, 22]. Detailed treatment of phase relationships by Muschietti et al. [20] for Gaussian Langmuir wave packets, show that the linear perturbation in the distribution function may be considered as the sum of a resistive and reactive components. For both components, the perturbation narrows and increases in magnitude as the wave packet length increases. Electron detectors with broad energy resolution can only detect the resistive component because it has only positive polarity and adds over the entire energy range. Detecting the bipolar nature of the reactive component requires narrow energy response (∆v/v ≤ 5 − 6%) detectors. Only three high frequency (MHz) wave-particle correlation experiments have been reported in the literature [7, 9, 16]. Very similar experiments have been tried on FREJA by Boehm et al. [5] and on FAST by Ergun et al. [7] but were limited in phase resolution. The ﬁrst two of these experiments used a correlator that worked by binning individual detected particles according to the phase of the strongest wave detected by a broadband (0.2–5 MHz) wave receiver [Ergun et al., 7, 9]. The most recent experiment used a correlator which had higher phase resolution and detected electrons bunched 90◦ outof-phase with respect to the electric ﬁeld, suggesting non-linear evolution and trapping of the electrons.

316

C.A. Kletzing and L. Muschietti

In what follows, we discuss the theory of linear perturbations to the distribution by a Gaussian wave packet and also present the theory of extended wave packets via a BGK analysis. We then describe in detail two of the three high frequency wave-particle correlator measurements referred to above. We then conclude with a discussion of these results in the context of theory.

13.2 Finite-Size Wave Packet in a Vlasov Plasma Consider a coherent, Langmuir packet propagating in the x direction with a phase ψ = kx − ωt where the frequency ω is close to the plasma frequency ωp . The wave packet is assumed to be localized, which we specify by a form factor η(x, t) which is piecewise continuous, bounded, and vanishing for x → ±∞. Its slow time dependence describes the drift due to the group velocity of the Langmuir wave or the growth (or damping) due to the interaction with the electrons. Explicitly, the wave electric ﬁeld is written as E(x, t) = E0 η(x, t)eiψ + c.c.

(13.1)

where c.c. is the complex conjugate. The electrons are assumed to consist of two populations: a dense background and energetic, streaming particles. In the problem that we treat, the density of the energetic electrons is orders of magnitude smaller than the background density, so that only the latter determines the real part of the dispersion relation. The electrons are described by a homogeneous distribution function F (v) which includes velocities around the phase velocity of the Langmuir wave, vp ≡ ω/k. Under the inﬂuence of the wave ﬁeld, the streaming population develops a time-dependent, inhomogeneous component f (x, v, t) which satisﬁes the Vlasov equation

∂ ∂ d eE ∂ f≡ +v (F + f ) (13.2) f= dt ∂t ∂x m ∂v where −e and m are the electron charge and mass, respectively. Our goal is to ﬁnd explicit expressions for f (x, v, t) and to analyze their phases versus the wave phase ψ in view of correlator applications. In the linear approximation, one neglects the perturbation ∂f /∂v on the right-hand side of (13.2) and formally integrates the equation, eE0 fL (x, v, t) = m

t

−∞

η(x , t )eiψ(x ,t )

∂ F (v ) dt + c.c. ∂v

(13.3)

The integration is carried out along the trajectories x (t ), v (t ) that have x and v for end points at t = t, namely x (t) = x and v (t) = v. The perturbation at large negative times when x → ±∞ is ignored. From (13.3) one sees that the linear perturbation fL oscillates in time with the same

13 Langmuir Wave-Electron Phase Correlation

317

frequency as the electric ﬁeld. One also notices that the perturbed distribution depends upon the structure of the form factor η along the past trajectories of the particles, which leads to a phase shift relative to the present wave phase ψ(x, t). 13.2.1 Linear Perturbation of the Electrons Taking as characteristics the straight trajectories v (t ) = v and x (t ) = x + v(t − t), where v is constant and positive, one rewrites (13.3) as fL (x, v, t) = A0 ei(kx−ωt)

∂F ∂v

0

η(x + vτ, τ + t)ei(kv−ω)τ dτ + c.c. (13.4) −∞

where A ≡ eE0 /m. The integral explicitly relates the distribution fL to the motion of the particles at earlier times, so that the eﬀect of spatial gradients in the electric ﬁeld amplitude will be directly seen in the resulting phase relation. Let us for instance consider the simple proﬁle of a square window with a width L, 0 : if x < 0 η(x) = 1 : if 0 < x < L (13.5) 0 : if x > L From (13.4), one readily obtains iAeiψ ∂F : if 0 < x < L 1 − βx + c.c. (13.6) fL (x, v, t) = ω − kv ∂v β (x−L) − β x : if L < x where β ≡ exp i(ω/v −k). In the square bracket to the right, the exponents reﬂect the presence of the boundaries and modify the phase of the perturbation relative to the wave phase ψ. They also yield a ballistic term β (x−L) −β x downstream of the interaction region. An important point is that these boundary terms keep the perturbed distribution function of resonant electrons bounded. The usual expression for a plane wave [(13.6) with the square bracket equal to unity] shows the perturbation to have a singularity at resonance, where the assumed linear solution, therefore, breaks down. Instead, the right-hand side for 0 < x < L can be expanded for velocities close to the phase velocity vp . The resulting expression is ﬁnite. One can then diﬀerentiate it with respect to v and thus evaluate at v = vp the derivative we neglected in the linearization of Vlasov equation (13.2). Using the notation F ≡ ∂F/∂v, the result can be written as ∂fL ∂lnF 2A 2 |v = F (vp ) − 1)kx cos ψ + (kx) sin ψ (13.7) (vp ∂v p ωvp ∂v The term proportional to sin ψ (thus out of phase with the electric ﬁeld) is seen to grow quadratically with the distance into the packet. Therefore, the packet must have a ﬁnite extent to assure the validity of the linear perturbation,

318

C.A. Kletzing and L. Muschietti

which we write (kL)2 ωvp /(2A). This inequality introduces an important parameter to the problem at hand 2 2eE0 k kL . (13.8) µ≡ m ω This quantity measures the eﬀect of the localized electric ﬁeld on the electrons and can be used as a small expansion parameter. We can think of it as the square of the bounce frequency times the transit duration of a resonant particle. 13.2.2 Case of a Gaussian Packet A more realistic envelope of wavepacket is provided by a Gaussian. In fact, since the dispersion relation of Langmuir waves is quadratic, a Gaussian packet moving at the group velocity describes well a propagating Langmuir packet. Its dispersion time is given by td ≈ (L/λd )2 ωp−1 with λd the Debye length. This time is very long compared to, for example, the transit duration of resonant electrons, tt ≈ kL ωp−1 . For simplicity, we choose here a static form factor η(x) = exp[−x2 /(2L)2 ] ,

(13.9)

which is justiﬁed since for many applications the group velocity u is much smaller than the velocity of resonant electrons, u/vp = 3 (kλd )2 1. After substituting (13.9) into (13.4) and some algebra, one obtains fL (x, v, t) = A η(x)ei(kx−ωt)

∂F L (−i)Z(ξ) + c.c. ∂v v

(13.10)

where Z is the usual plasma dispersion function, yet has here a completely diﬀerent argument: L x ξ = (ω − kv) − i . (13.11) v 2L The real part of the argument is the Doppler-shifted frequency seen by a traversing electron times its transit duration through the wavepacket. It determines the proximity to resonance during the interaction. The imaginary part of the argument describes the position with respect to the center of the packet. Let us evaluate now (13.10) at the center of the packet. We can deﬁne a resonance function that represents the response of the electron distribution to the wave ﬁeld: ξr 2 2 L √ −ξr2 −iL Z(ξr ) = πe − i 2e−ξr ey dy (13.12) R≡ v v 0 The real part of R, associated with the Gaussian e−ξr , represents the resistive contribution (in phase with the electric ﬁeld). The imaginary part, associated 2

13 Langmuir Wave-Electron Phase Correlation

319

Resistive component

150. kL=80

100.

50.

kL=30 kL=15

0. -0.1

0.0

0.1

Reactive component

100. kL=80

50.

kL=30 kL=15

0. -50. -100. -0.1

0.0

0.1

Fig. 13.1. Resonance function deﬁned by (13.12). Real and imaginary parts of R are shown for various packet sizes (Reprinted with permission from American Geophysical Union)

with the Dawson integral, represents the reactive contribution (out of phase with the electric ﬁeld). When the packet is large so that ωL/v → ∞, Re(R) tends to the usual delta function of the Plemelj formula, πδ(kv−ω), and Im(R) tends to the principal part of 1/(kv − ω). Plots of real and imaginary parts of R are displayed in Fig. 13.1 for various sizes of packets. One clearly sees the tendency to a delta function for large kL and, conversely, the broadening of the resonance for a more localized wavepacket. The perturbed oscillating distribution has components in and out of phase with respect to the electric ﬁeld. Let us split fL of (13.10) into resistive and reactive terms, in a way that emulates the procedure performed by the wave correlator on rocket data. One obtains fL =

µ vp2 F (v) η(x) [Zi cos(kx − ωt) + Zr sin(kx − ωt)] kL v

(13.13)

where the acceleration factor A = eE0 /m has been rewritten in terms of µ using (13.8). The perturbation is thus either rather resistive or reactive depending upon the relative weight of Zi and Zr . Its amplitude is proportional to the wave amplitude, to the packet size, and to the slope of the distribution F . Figure 13.2 displays the perturbation in phase space by means of contours. Solid contours with gray shadings indicate a positive value, i.e. an

320

C.A. Kletzing and L. Muschietti e

n

Te

dis tance kx

30.

30.

20.

20.

10.

10.

0.

0.

-10.

-10.

-20.

-20.

-30.

-30.

-40.

-40.

-50.

-50.

0.90

0.95

1.00

1.05

1.10

-0.01

0.01

velocity v/vp

Fig. 13.2. Linear perturbed distribution as described by (13.13) with power-law model F (v) ∼ v −4 . Contours with gray shadings indicate enhancements at levels 0.006 (light) and 0.018 (dark). Dashed contours denote depletions at levels −0.006 and −0.018. The perturbation (normalized to F (vp )) features a chain of bunchellipses centered at the phase velocity vp . Panel marked δn at right displays the resonant density perturbation, split into its resistive (thick line) and reactive (thin line) components. Rightmost panel shows the potential φ of the Langmuir packet with characteristics: kL = 30, µ = 0.09

accumulation of electrons. Dotted contours indicate a negative value, i.e. a dearth of electrons. The bunching of the particles yields a chain of ellipses centered on the phase velocity vp and zeroes of the potential, which is shown in the rightmost panel. The panel marked δn displays both the resistive (thick line) and reactive (thin line) components of the density perturbation. It is clear that the linear perturbation is mostly resistive. In addition, from the maxima of δn being in phase with ∂φ/∂x > 0 we can conclude that the wave accelerates the electrons, whereby it is being damped. This is consistent with the power-law model, F (v) ∼ v −4 , we have chosen for drawing the plot. Note the slight tilt of the ellipses in Fig. 13.2. A consequence is that the result of a velocity-integration of the perturbed distribution critically depends on the integration window’s width and centering. For computing δn shown

13 Langmuir Wave-Electron Phase Correlation

321

in the mid-panel we used a 7% width on either side of vp . However, if the window is narrower and not centered on vp , the integration will emphasize the reactive component. Furthermore, the sign of the latter depends upon whether the window happens to be centered a little above or below vp . In Fig. 13.2 the parameter µ is small, µ = 0.09, hence the interaction is justiﬁably in the linear regime. However, when either the wave amplitude is larger or the packet is more extended, the linear solution to the Vlasov equation loses its validity and one must account for nonlinear corrections. These corrections to the orbits of resonant electrons δv and δx are diﬃcult to compute analytically. Instead, we resort to particle-in-cell simulations, where consistent interactions between ﬁeld and particles are automatically taken into account. The results of these simulations are shown in Fig. 13.3 which display the bunch-ellipses in a weakly nonlinear regime with µ = 1.2. Solid contours and shaded areas indicate a positive value, or an accumulation of electrons. Dotted contours indicate a negative value, or a dearth of electrons. Two points must be noted. First, the wave is here driven by a bump on the tail of the distribution function. Accordingly, resonant electrons are decelerated

Fig. 13.3. Bunch-ellipses in a weakly nonlinear regime, from a particle-in-cell simulation where the Langmuir packet is ampliﬁed by a bump on tail located at v = 10ve . Here µ = 1.2 and vp = 8.7ve . Accumulation locales (shaded in gray) have decelerated while dearth locales (dashed contours) accelerated. More in the text

322

C.A. Kletzing and L. Muschietti

by the wave and maxima of δn are in phase with ∂φ/∂x < 0. Second, since bunches of accumulated electrons are decelerated, the nonlinear correction to their orbit has δv < 0 and δx < 0. In contrast, bunches associated with a dearth of electrons are accelerated and thus have δv > 0 and δx > 0. Hence, the two types of bunch-ellipses are no longer aligned at v = vp as in Fig. 13.2. One can see in the plot that the solid contours have moved to the left while the dotted contours have moved to the right. In addition, their position in x drifts and slowly converges toward maxima of the potential, whereby their phase becomes more reactive.

13.3 Extended Wave Packet: A BGK Analysis The only nonlinear term in the Vlasov-Poisson system of equations is the acceleration term on the right hand side of (13.2). In Sect. 13.2, we chose to linearize this term by splitting the distribution in a large, homogeneous part F and a small, inhomogeneous part f , and then neglecting ∂f /∂v versus ∂F/∂v. We found that this was justiﬁed as long as µ 1. However, when the wave packet is so extended that a traversing electron can satisfy the resonance condition for a long time, this electron can be signiﬁcantly accelerated by the electric ﬁeld. Its orbit is then deeply altered. Therefore, the distribution function is strongly modiﬁed and the linearization procedure breaks down. The BGK method oﬀers another approach to solving the Vlasov-Poisson system of equations. Let us now imagine a very long wavepacket and examine the phase-space orbits of electrons in the so-called waveframe (frame moving with the wave phase velocity). These orbits are determined by the energy of the electrons ∂φ . (13.14) w = v 2 /2 − φ(x) where E(x) = − ∂x For this section, in order to simplify the notation, we introduce dimensionless units where length is normalized by λd , velocity is normalized by the electron ' thermal velocity ve = Te /m, and the electrostatic potential is normalized by Te /e. Electrons with w < 0 are trapped in local maxima of the wave potential. Electrons with w > 0 are untrapped and alternately accelerate and decelerate while passing over hill and dale of the potential. Now, any function of w where φ(x) is time-independent automatically satisﬁes Vlasov equation. The BGK approach exploits that property and assumes that the distributions are function of w only and that they are in a self-consistent steady state with the potential. This enables one to concentrate on solving the Poisson equation for a given model of wave ﬁeld, e.g. here a sinusoidal wave φ(x) = Ψ sin kx .

(13.15)

Let Fe (w) and Ft (w) be the distribution functions of, respectively, the passing and the trapped electrons. Poisson equation reads

13 Langmuir Wave-Electron Phase Correlation

d2 φ = −ni + dx2

∞

Ψ

Fe+ (w) + Fe− (w) √ dw + 2(w + φ)1/2

Ψ

−φ

√ 2Ft (w) dw . (w + φ)1/2

323

(13.16)

Terms on the right hand side represent, in order, the density of the ions, which are supposed to form a constant background, the density contribution from the passing electrons, and that from the trapped electrons. The passing electrons have been split into those moving to the right, Fe+ (w), and those moving to the left, Fe− (w). The trapped electrons, by contrast, must be symmetric with the same ﬂux of right and left moving particles, Ft+ (w) = Ft− (w) = Ft (w), in a stationary situation. 13.3.1 Passing Electrons A model of distribution for the ambient electrons that is practical for computational purposes and representative of the observed distributions is given by 1 2 Fe± (v) = , (13.17) π [1 + (v ± vp )2 ]2 where v is an absolute number measuring the velocity from the wave frame, and the direction is selected by the ± sign. Note that this distribution is normalized to unity and becomes a power-law in v −4 at large velocities. In the presence of the wave it translates into 2 Fe± (w) = (13.18) 2 2 with w > Ψ . √ π 1+ 2(w − ψ)1/2 ± vp After integrating this expression for the densities of right and left moving particles, one obtains the total density of passing electrons as

√ √ ϕ (1 + ϕ − vp2 ) 2 ϕ 2 1 np = 1 − − arctan (13.19) π vp4 + 2vp2 (1 − ϕ) + (1 + ϕ)2 π vp2 + 1 − ϕ where the notation ϕ ≡ 2(Ψ +φ) is introduced. The density is maximum where the potential is minimum, at ϕ = 0, and monotonically decreases toward a minimum where the potential is maximum, at ϕ = 4Ψ . We will assume that 4Ψ 1, which enables us to expand the complicated expression above into the simpler √ √ 16 2(5vp2 − 1) 4 2 1/2 (Ψ + φ) − (Ψ + φ)3/2 (13.20) np (φ) = 1 − π(vp2 + 1)2 3π(vp2 + 1)4 The most important term, in ϕ1/2 , has a coeﬃcient that is related to the value of the ambient distribution at the wave phase velocity ((13.17) with v = 0). This is in agreement with the intuitive'idea that electrons which move slowly along the separatrix (located at v = 2Ψ (1 + sin kx)) are strongly aﬀected by the potential and thus contribute the most to the density variations. By contrast, fast-moving electrons far from the separatrix hardly “notice” the potential and pass by quasi undisturbed.

324

C.A. Kletzing and L. Muschietti

13.3.2 Trapped Electrons We formally deﬁne the density of trapped electrons nt as an unknown function of φ through the expression Ψ √ 2Ft (w) dw . (13.21) nt (φ) ≡ (w + φ)1/2 −φ This integral equation can be inverted for Ft (w), which yields −w 1 d 1 nt (φ) dφ . Ft (w) = √ 2π −Ψ (−w − φ)1/2 dφ

(13.22)

The potential φ(x) is given by (13.15) and requires a net density perturbation ns (φ) = −k 2 φ with

ns = np + nt .

(13.23)

Substituting ns and np for nt in (13.22), one obtains after integration √ 4(5vp2 − 1) 2 2 − (Ψ − w)1/2 + (Ψ − w) (13.24) Ft (w) = 2 2 2 π(vp + 1) πvp π(vp2 + 1)4 with −Ψ ≤w 3.5 are preferred. The correlator was ﬂown on a rocket launched from Poker Flat, Alaska on March 4, 1988 which crossed several auroral arcs during the expansion phase of substorm. During the ﬂight, the parallel component of the electric ﬁeld reached amplitudes of 100 mV/m in the frequency band from 200 kHz to 5 MHz with a dominant frequency of 1.4 MHz indicating the presence of strong Langmuir oscillations. The output of the correlator showed several events with signiﬁcant correlation with electrons from the eight lowest energy channels which covered a range of energies from 380 eV to 3.2 keV. The correlations appeared ﬁrst in the higher energy channels and then moved to lower energy. Analysis of the data from this rocket ﬂight yielded ﬁve events with ﬂuctuations greater than 4σ. These ﬁve events, plotted as function of wave electric potential, are shown in Fig. 13.6 taken from Ergun et al. [9]. Each event is plotted with the associated error bar as determined from experimental uncertainties as well as the 1σ statistical error of the counts. All ﬁve events are located near 90◦ or 270◦ with respect to the wave electric

328

C.A. Kletzing and L. Muschietti

Fig. 13.6. Phase of bunched electrons for ﬁve events of correlations at levels greater than 4σ from Ergun et al. [9]. The phase of each event is plotted along with error bars derived from experimental uncertainties (Reprinted with permission from American Geophysical Union)

potential. This corresponds to 0◦ or 180◦ with respect to the wave electric ﬁeld. The frequency of all 5 events was 1.4 MHz which, when combined with the electron energies of 380 eV to 3.2 keV, yields wavelengths of 8 m to 20 m. Events 2 and 5 were such that more electrons were being accelerated by the wave than were being decelerated and thus corresponds to wave damping. They occurred during periods of weaker amplitudes of 10–20 mV/m. Events 1, 3, and 4 had more electrons decelerating than accelerating corresponding to wave growth. At these times the wave amplitudes were near 100 mV/m or greater. Following the analysis of Sect. 13.2, the value of µ can be calculated using the observed electric ﬁeld amplitude of 80 mV/m, the inferred wavelength λ = 12.8 m, and observed the plasma frequency of 1.4 MHz along with a value for kL. For these events, kL was estimated to be of the order of kL = 30 which yields a value of µ = 0.09. This is consistent with the linear analysis for a short wave packet which predicts the observation of the resistive component of the electron perturbation because in this regime, the resistive perturbation is generally larger than the reactive perturbation as illustrated in Fig. 13.2. 13.4.2 Measurements of the Reactive Component Further improvement in the measurement of the phase of electron bunching in Langmuir waves in the auroral zone was achieved using a new phase correlator developed at the University of Iowa (UI) in the late 1990’s. Figure 13.7 shows

13 Langmuir Wave-Electron Phase Correlation

329

VC O Mas ter C lock Divide by 16 C lock

Kept in phas e by P LL

Input Waveform

P has e B in at 1.6 Mhz

8/9

4/5

0/1

12/13

Fig. 13.7. The UI wave-particle correlator uses a phase-locked loop (PLL) locked to the measured waveform with a clock derived from master voltage-controlled oscillator (VCO) running at 16 times the frequency. This master clock subdivides the wave in 16 phase bins. Electron counts are sorted into the bins as they arrive to produce a map of phase bunching (Reprinted with permission from American Geophysical Union)

the principle of operation. The correlator used a voltage-controlled oscillator (VCO) running at 16 times the expected frequencies for Langmuir waves. The VCO clock signal was divided by 16 to provide a signal which was then aligned to the measured AC parallel electric ﬁeld through the use of a phase-locked loop (PLL). The PLL output a signal which indicated when the loop was properly locked and also the frequency of the locked signal which is then compared with the analog waveform data to ensure that the PLL was locked to signals of interest. Because the VCO master clock runs at 16 times the frequency of the wave to which the PLL is locked, it provides a highly accurate means of sub-dividing the measured wave into 16 phase bins. As electrons are counted by the detectors, they are sorted into the appropriate phase bin associated with their arrival time. Calibration of the detector, wave, and correlator electronics veriﬁed the accuracy of the phase bins and determined the timing delays through the system. The timing delays cause the absolute bin number which corresponds to 0◦ phase shift (as well as other phase angles) to shift as a function of input wave frequency. For example, at 1.6 MHz, 0◦ phase shift between the wave and the electrons occurs between correlator bins 8 and 9. The bottom of Fig. 13.7 shows the calibrated phase angles of the bins at a frequency of 1.6 MHz. The UI correlator was ﬂown on a rocket ﬂight which was launched from Poker Flat, Alaska in February 6, 2002. Strong Langmuir waves were observed as the rocket traversed an auroral form near the poleward boundary of the auroral precipitation. The Langmuir waves were associated with a burst of ﬁeld-aligned electrons at energies below 1 keV and well below the inverted-V

C.A. Kletzing and L. Muschietti

σ

330

Fig. 13.8. Two examples of signiﬁcant correlation of Langmuir waves with electrons at 468 eV. For each of the 16 phase bins, the count level is shown in terms of standard deviations away from the average number of counts that would be expected for a bin for the total number of counts received (Reprinted with permission from American Geophysical Union)

peak energy of 5 keV. During this period of Langmuir emission, two intervals of stronger emission with amplitudes of 60–200 mV/m were observed. Each of these intervals were of the order of 50 ms in duration. During both intervals, the ﬁeld-aligned wave power was 8.5–11 times that of the perpendicular power indicating ﬁeld-aligned waves. Because the payload telemetered continuous waveform data, it was possible to determine that the Langmuir waves were remarkably monochromatic with little or no variation in frequency. During each of the two larger amplitude bursts, the amplitudes varied slowly over hundreds of wave periods. During each of the two bursts of strong Langmuir waves, signiﬁcant waveparticle correlation was found as shown in Fig. 13.8. Each set of panels shows four consecutive sets of correlator measurements of electrons with energy of 468 eV and which were sampled every millisecond. Data are plotted as standard deviations away from the expected average count rate in each phase bin. To be regarded as signiﬁcant, we require a correlation level of at least 3.5σ. Although there is frequent variation of the order of 2σ shown in Fig. 13.8, these are not considered signiﬁcant because the probability of 2σ variation is once in 22 samples as discussed above. With this in mind, in the left set of panels, the ﬁrst and third panels show notable correlation. The ﬁrst panel shows a weakly signiﬁcant correlation with three adjacent phase bins with ∼2σ levels of correlation corresponding to a 3.3σ variation.

13 Langmuir Wave-Electron Phase Correlation

331

However, the third panel from the top has two consecutive phase bins with phase bunching above 3σ. The combined probability of two such bins occurring together exceeds that of a single 4σ event. The bins in which the phase bunching occurs are those at 90◦ with respect to the wave electric ﬁeld (positive away from the ionosphere) for the wave frequency of 1.6 MHz. The second example shown in the right set of panels in Fig. 13.8 shows a second event with similar high levels of signiﬁcance in the top three panels. Initially, a single channel with more than 4σ signiﬁcance occurs at 90◦ . In the second panel, this feature broadens and shifts to somewhat larger phase shift with four phase bins above 2σ representing a combined signiﬁcance of more than 4σ. The calculated probability of random occurrence of the counts represented by the four phase bins centered around 90◦ is one in 932,068 suggesting that this is a highly signiﬁcant correlation. The third panel shows further advance to larger phase angles, with most signiﬁcant set of signals now showing more than -4σ (5 channels at −2σ each) between 270◦ and 0◦ . Figure 13.9 shows three sequential distribution function plots from immediately before, during and immediately after the ﬁrst correlation event. The times given above each plot correspond to the center time of each 40 ms energy sweep. As can be seen, at the time of the correlation, a small, downward electron beam parallel to the magnetic ﬁeld is measured at the same energy as the correlated electrons. This is indicated by a small arrow at the velocity corresponding to 468 eV. The isolated beam is not present in the distribution function measured before the correlation or in the distribution measured after the correlation. Although the time resolution of these measurements is much lower than for the correlator, this suggests that the correlation arises from this beam. The brief increase in electron phase space density at the energy of correlated electrons illustrated in Fig. 13.9 was also seen for the second correlation event. In both cases the preceding and following energy sweeps did not show this feature, suggesting that the resonant electrons were only brieﬂy in the detectable energy range. From the energy of the correlated electrons (468 eV) and the frequency of the waves (1.6 MHz), we can derive the wavelength of the Langmuir waves as 8.2 m, similar to the results of the UC Berkeley measurements. The waveform data makes it possible to estimate the packet parameters to aid in the interpretation of correlation events. Figure 13.10 shows 15 ms of data which includes the interval of the second set of correlations. The interval during which the correlations were observed is indicated by a heavy line above the waveform data. As can be seen, the correlations occurred during the largest wave amplitudes. If we assume a background thermal energy of 0.2 eV and use the inferred wavelength of 8.2 m along with the observed Langmuir frequency of 1.6 MHz, then the group velocity for Langmuir waves, vg =

3v 2 k ∂ω 3v 2 th = th , ∂k ωpe λfpe

(13.25)

332

C.A. Kletzing and L. Muschietti

Fig. 13.9. Sequential distribution functions from before, during, and after the correlation event shown in the left panel of Fig. 13.8. The velocity corresponding to the correlated electrons is indicated with an arrow in the middle panel and shows that a ﬁeld-aligned beam is present at this time (Reprinted with permission from American Geophysical Union)

gives the value of vg = 8.02 km/s = 8 m/ms. If the wave is traveling down the ﬁeld line, then the amount of the wave packet above the correlation observations extends from the time of the correlations in Fig. 13.10 to the end of the packet some 3 ms later. This yields a length of the wavepacket above rocket payload of 24.1 m at the time of the correlations and corresponds to a value of kL = 18.5. A rough average for the amplitude of the electric ﬁeld during this interval is 130 mV/m. Combining this with the estimate of kL gives a value of µ = 0.12, consistent with the linear theory and similar to the value used in Fig. 13.2. It should be pointed out, however, that the usual limitations of single point spacecraft measurements apply to this interpretation. With the data at hand, we cannot rule out a scenario in which the electron beam has suddenly appeared and the wave packet has grown over time at the spacecraft location. Indeed, such a scenario would be consistent with the quasi-exponential increase in amplitude of the packet in the 10 ms preceding the correlation observations. In this case, there is no way to ascertain the amount of wave packet above the payload, but the shape of the packet suggests that the correlations were observed shortly after the linear growth phase ended and some type of non-linear saturation began to operate. The group velocity calculation still applies, but now becomes a lower bound on the length of the packet. By observing the packet for 3 ms we know that it extended at least 24 m above the rocket, but it may have extended much further, and then disappeared due to temporal eﬀects in the driving electron distribution. A third alternative is that the payload may have moved into and then out of a pre-existing region

13 Langmuir Wave-Electron Phase Correlation

333

Fig. 13.10. Langmuir waveform data which includes the interval of the second set of correlation events. The correlation interval is indicated by the solid bar above the waveform

of Langmuir waves such that the observed wave envelope is determined by the spatial structure of the Langmuir waves along the rocket track. However, given the rocket velocity of roughly 1 km/s, this latter scenario would require a rather small wave packet and does not seem likely.

13.5 Discussion The electron correlator observations reported by the UC Berkeley were predominantly resistive, that is, in-phase or 180◦ out-of-phase with the wave electric ﬁeld. Although they had expected to see the reactive component, the observations of the resistive component prompted the analysis developed in Sect. 13.2 and led to the conclusion that they had observed electrons which were bunched by relatively short wave packets with kL of the order of 50–100. This result was also consistent with the value for µ that was determined which suggested that thee observed waves were in the linear regime. As shown in the middle panel of Fig. 13.2, for µ = 0.09, the perturbation to the density and hence, to the electron distribution, is dominated by the resistive component for small values of µ. Although the reactive component is present, as shown in Fig. 13.1, its bipolar character requires relatively narrow response electron detectors so as not to average the positive and negative perturbation together, yielding no perturbation. The detectors which made these resistive correlator measurements had ∆v/v 16% and as shown in

334

C.A. Kletzing and L. Muschietti

Fig. 13.1, this is broader than the expected perturbation relative to the phase velocity. The electron phase bunching observed by the UI group was predominantly 90◦ out-of-phase with the electric ﬁeld, suggesting a trapped population of electrons. The wave ﬁeld in which the phase-bunched electrons were observed was long-lived in terms of wave periods, monochromatic, and showed slow modulation of the wave envelope. As discussed above, using the assumption that the wave packet moved past the payload due the group velocity of the waves yields a low value of µ which would suggest that the linear analysis also applies to this case. In examining Fig. 13.2, it is seen that although the overall perturbation to the density has a greater resistive component than reactive component, the reactive component does exist. On closer examination of Fig. 13.2, however, it can be seen that for a narrow range of velocities of the order of ∆v/v = 5%, on either side of the phase velocity, the reactive component is more dominant. This is seen in the shifting of the perturbed distribution toward being in phase with the potential for velocities above the phase velocity and a shift toward 180◦ out-of-phase for velocities below the phase velocity. Because the UI electron detectors had a comparably narrow velocity response ∆v/v 5%, they are capable of capturing one or the other side of the perturbation and could measure a reactive perturbation even for a short wave packet. In the case that the envelope of the observed waves is predominantly due to temporal evolution, then it is likely that the wave packet extends a significant distance above the rocket payload. This is suggested by the fact that the electron distribution will be unstable over a wide range of altitudes and would be expected to grow waves over a region of many wavelengths in extent. The character of the wave envelope also suggests a transition away from the linear stage of growth. Taken together, these two arguments suggest that an alternative explanation is that the correlations indicate trapping and thus the BGK analysis is appropriate. As shown in the right hand panel of Fig. 13.3, when the electron-Langmuir wave interaction becomes nonlinear, the density perturbation shifts toward a purely reactive phase as the electrons become trapped. Indeed the bunching of the electrons observed by the UI group is such that the positive perturbation was that of trapped electrons. To resolve this ambiguity will require future experiments which provide adjacent energy channels with narrow response so that the full character of the perturbed distribution function can be revealed. This would allow one to see if both reactive pieces, that is the positive perturbation below the phase velocity and the negative perturbation above the phase velocity shown in Fig. 13.1, are observed side-by-side as the linear model would suggest or if only a single reactive perturbation is observed, consistent with the BGK analysis.

13 Langmuir Wave-Electron Phase Correlation

335

13.6 Conclusions Multiple spacecraft observations have conﬁrmed the ubiquitous nature of Langmuir waves in the presence of auroral electrons. Early observations have shown clear evidence that the electrons show variations consistent with bunching at or near the Langmuir frequency. Linear analysis of the interaction of a ﬁnite Gaussian packet of Langmuir waves shows that there are two components to the perturbation to the electron distribution function, one in-phase (or 180◦ out-of-phase) with respect to the electric ﬁeld called the resistive component and one which is 90◦ (or 270◦ ) out-of-phase with respect to the electric ﬁeld. For small wave packets, the resistive perturbation dominates. For longer wave packets, a non-linear analysis is appropriate which suggests that the electrons have interacted long enough to become trapped and the reactive phase becomes dominant. Rocket observations of the phase bunching of the electrons using waveparticle correlators have measured both components. The UI observations [Kletzing et al., 16] diﬀer from those of the UC Berkeley observations [Ergun et al., 7, 9] in that a purely reactive phase bunching was observed as compared to a predominantly resistive perturbation. The resistive phase results of the UC Berkeley group were interpreted as arising from a short wave packet. The UI observations of the reactive phase can be explained by either a long, coherent train of Langmuir waves or that the narrower velocity response of the UI detectors made it possible to capture only one side of the reactive component of the perturbed distribution function for a short wave packet in the linear regime. Future wave-particle correlator experiments should be able to resolve these questions by providing more examples with better velocity space coverage.

References [1] Bale, S.D.: Observation of the topside ionospheric mf/hf radio emission from space, Geophys. Res. Lett. 26, 667, 1999. [2] Bauer, S.J. and R.G. Stone: Satellite observations of radio noise in the magnetosphere, Nature 218, 1145, 1968. [3] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and hf auroral hiss generated at low altitude, J. Geophys. Res. 94, 1359, 1989. [4] Boehm, M.H.: Waves and static electric ﬁelds in the auroral acceleration region, PhD thesis, University of California, Berkeley, 1987. [5] Boehm, M.H., G. Paschmann, J. Clemmons, H. H¨ ofner, R. Frenzel, M. Ertl, G. Haerendel, P. Hill, H. Lauche, L. Eliasson, and R. Lundin: The tesp electron spectrometer and correlator (F7) on Freja, Space Sci. Rev. 70, 509, 1995. [6] Bonnell, J.W., P.M. Kintner, J.E. Wahlund, and J.A. Holtet: Modulated langmuir waves: observations from freja and scifer, J. Geophys. Res. 102, 17233, 1997.

336

C.A. Kletzing and L. Muschietti

[7] Ergun, R.E., C.W. Carlson, and J.P. McFadden: Wave-particle correlator instrument design, In: R.F. Pfaﬀ, J.E. Borovsky, and D.T. Young (Eds.), Measurement Techniques in Space Plasmas: Particles, volume 102 of AGU Geophys. Monogr. Ser., p. 4325, AGU, Washington, D.C., 1998. [8] Ergun, R.E., C.W. Carlson, J.P. McFadden, J.H. Clemmons, and M.H. Boehm: Evidence of a transverse modulational instability in a space plasma, Geophys. Res. Lett. 18, 1177, 1991. [9] Ergun, R.E., C.W. Carlson, J.P. McFadden, D.M. TonThat, J.H. Clemmons, and M.H. Boehm: Observation of electron bunching during Landau growth and damping, J. Geophys. Res. 96, 11371, 1991. [10] Gough, M.P., P.J. Christiansen, and K. Wilhelm: Auroral beam-plasma interactions: particle correlator investigations, J. Geophys. Res. 90, 12287, 1990. [11] Gough, M.P. and A. Urban: Auroral beam/plasma interaction observed directly, Plan. Space Sci. 31, 875, 1983. [12] James, H.G., E.L. Hagg, and L.P. Strange: Narrowband radio noise in the topside ionosphere, AGARD Conf. Proc., AGARD-CP-138, 24–1–24–7, 1974. [13] Kelley, M.C. and G.D. Earle: Upper hybrid and Langmuir turbulence in the auroral e-region, J. Geophys. Res. 93, 1993, 1988. [14] Kellogg, P.J. and S.J. Monson: Radio emissions from the aurora, Geophys. Res. Lett. 6, 297, 1979. [15] Kintner, P.M., J. Bonnell, S. Powell, and J.E. Wahlund: First results from the freja hf snapshot receiver, Geophys. Res. Lett. 22, 287, 1995. [16] Kletzing, C.A., S.R. Bounds, J. LaBelle, and M. Samara: Observation of the reactive component of langmuir wave phase-bunched electrons, Geophys. Res. Lett., 32, L05106, doi:10.1029/2004GL021175, 2005. [17] McAdams, K.L. and J. LaBelle: Narrowband structure in hf waves above the electron plasma frequency in the auroral ionosphere, Geophys. Res. Lett. 26, 1825, 1999. [18] McAdams, K.L., J. LaBelle, M.L. Trimpi, P.M. Kintner, and R.A. Arnoldy: Rocket observations of banded stucture in waves near the langmuir frequency in the auroral ionosphere, J. Geophys. Res. 104, 28109, 1999. [19] McFadden, J.P., C.W. Carlson, and M.H. Boehm: High-frequency waves generated by auroral electrons, J. Geophys. Res. 91, 12079, 1986. [20] Muschietti, L., I. Roth, and R. Ergun: Interaction of Langmuir wave packets with streaming electrons: phase-correlation aspects, Phys. Plasmas 1, 1008, 1994. [21] Newman, D.L., M.V. Goldman, and R.E. Ergun: Langmuir turbulence in the auroral zone 2. nonlinear theory and simulations: J. Geophys. Res. 99, 6377, 1994. [22] Nicholson, D.R.: Introduction to Plasma Theory, Wiley, New York, 1983. [23] Reiner, M.J. and M.L. Kaiser: Complex type iii-like radio emissions observed from 1 to 14 mhz, Geophys. Res. Lett. 26, 397, 1999. [24] Reiner, M.J., M. Karlicky, K. Jiricka, H. Aurass, G. Mann, and M.L. Kaiser: On the solar origin of complex type iii-like radio bursts observed at and below 1 mhz, Astrophys. J. 530, 1049, 2000. [25] Samara, M., J. LaBelle, C. A. Kletzing, and S.R. Bounds: Rocket observations of structured upper hybrid waves at fuh = 2fce , Geophys. Res. Lett., submitted, 2004.

13 Langmuir Wave-Electron Phase Correlation

337

[26] Sanbonmatsu, K.Y., I. Doxas, M.V. Goldman, and D.L. Newman: NonMarkovian electron diﬀusion in the auroral ionosphere at high Langmuir-wave intensities, Geophys. Res. Lett. 24, 807, 1997. [27] Spiger, R.J., J.S. Murphree, H.R. Anderson, and R.F. Loewenstein: Modulation of auroral electron ﬂuxes in the frequency range 50 kHz to 10 MHz, J. Geophys. Res. 81, 1269, 1976. [28] Stasiewicz, K., B. Holback, V. Krasnoselskikh, M. Boehm, R. Bostr¨ om, and P.M. Kintner: Parametric instabilities of langmuir waves observed by freja, J. Geophys. Res. 101, 21515, 1996. [29] Walsh, D., F.T. Haddock, and H.F. Schulte: Cosmic radio intensities at 1.225 and 2.0 mc measured up to and altitude of 1700 km, Space Res. 4, 935, 1964. [30] Weatherwax, A.T., J. LaBelle, M.L. Trimpi, and R. Brittain: Ground-based observations of radio emissions near 2fce and 3fce in the auroral zone, Geophys. Res. Lett. 20, 1447, 1993.

Index

absorption D-region 163 negative 65 resonant 246 acceleration 205, 253 ‘inverted-V’ 107, 129 auroral electron 205 in electric ﬁeld 64 in lower hybrid waves 258 primary region 110 regions of 64 transverse 286 upward ions 111 active sounding 30 adiabatic motion 64 AKR 106, 129, 205, 298, 305 source region 129 bounded source 66 conditions for generation 56 elementary radiation events (ERE’s) 132 ﬁne structure 132 narrow band 132 relation to continuum 43 source extension 59 source life time 59 source region properties 63 statistical properties 58 Alfv´en waves 110 kinetic 116, 119 shear 116, 205 ampliﬁcation path 72 angle of propagation 71

antenna distributed dipoles 192 eﬀective length 199, 201 electric dipole 299 multiple 298 orientation 206 Appleton-Hartree equation 147 atmosphere 142 aurora auroral hiss 60, 68, 156, 177, 191, 201, 212 power ﬂux 204 Auroral Kilometric Radiation (AKR) 55, 205, 212 auroral oval 106, 205 auroral regions 107, 298 acceleration region 107 downward current 107 upward current 107 auroral roar 212 auroral wave emission 314 background plasma 114 black aurora 112 dayside 106 electric ﬁeld 205 radar 142 visible 142 beaming angle 41 Bernstein modes 9, 24, 41 BGK modes 121 Boltzmann equation 90 brightness temperature 55

340

Index

bulk ﬂows

114

cavities auroral 178 cavity width 286 ionospheric 66 observation of 68 plasmaspheric 44 refractive index 27 small scale 60, 214 spherical 274 Cherenkov radiation 29, 156, 202, 212 chord length 274 chorus 175 Cluster spacecraft 298 CMA diagram 6, 192 coherence length 91, 205 collapse lower hybrid 274 time 274, 289 collision frequency anomalous 258 comets 252 condensation of electrons at low v 65 convection 133 correlation auroral hiss-HF backscatter 164 correlation function 88 multi-point function 88 whistler-HF backscatter 164 correlator wave particle 214, 313 cosmic rays 106 secondaries 106 showers 106 current bifurcation 136 ﬁeld-aligned 107, 134 Hall 135 Pedersen 110 sheet 251, 252, 259, 314 tail current sheet 106 thin tail sheet 135 cutoﬀ 235 high frequency 28 low frequency 129 upper frequency 141 X-mode 71

density depletion 274, 278 depth of 292 diﬀusion 253 dispersion maximum growth rate 79 of plasma waves 5 relation ﬁnite geometry 78 distribution beam 129, 213, 252 auroral electron 110 electron conic 255 electron shell 255 hole 63 horseshoe 64, 128, 130, 212, 298 loss cone 63, 128, 213 Maxwellian 110 nonthermal 213 passing 125 perturbation of 313 ring 128 shell 298 trapped 64, 125 two electron 116 wave 300, 308 DNT resonance 19 “donkey ear—hyperpage 45 double layer 115, 125, 127, 255 electrostatic shock 116 ramp 128 shear ﬂow driven 120 strong 116 dual payloads 16 echo delay times 10 echo train 156 R-X mode echo 180 whistler echoes 28, 158, 180 Z mode echo 180 Z-mode, X-mode 14, 22 electric ﬁeld ﬁeld-aligned 112 reconnection 255 electrons anisotropic 41 auroral 129 conics 255 density scale height 212

Index electron beams 255 hard spectrum 98 inertial length 205 magnetized 133 oscillations, cylindrical 20 passing 323 precipitating 156 reactive bunching 313, 328 relativistic 129 resistive bunching 313, 325 soft spectrum 99 sounder-accelerated 17 trapped 324 trapped magnetospheric 156 wave packet perturbation 317 Ellis radio window 217 energy conversion 251 redistribution 253 transport 251 ﬂuctuations density 298 electromagnetic 298 small scale 91 Freja spacecraft 298 frequency electron cyclotron 38, 129, 141 lower hybrid 178 plasma 128, 141, 213, 252, 254 upper hybrid 28, 213 frequency gap 68 Gendrin angle 150, 168 Geos Spacecraft 298 geospace 298 harmonics electron cyclotron height apparent 4 heliosphere 252

212

impact parameter 275, 278 impedance 19 incoherence 205 index of refraction 15 refractive surfaces 17

341

surface of 150 inhomogeneity 235 ﬁeld-aligned 15 random 88 transverse 246 instability beam-beam 125 Buneman 126 current 127 gyro-resonant 156 modulational 289 two stream, modiﬁed 256 two-stream 125, 126 kinetic 128 whistler mode 156 interaction beam-plasma 202, 313 coherent 202 interacting wave bands 132 Langmuir-electron 315 nonlinear wave 315 resonant 141 wave-particle 141, 212, 253 Interball spacecraft 298 interferometry 299 invariant embedding method 240 ﬁrst adiabatic 64 ion gyro radius 252 ion inertial length 252 ion-acoustic speed 116 ionogram 4 ionosonde 4 ionosphere air-ionosphere boundary 153 lower edge 153 ionospheric cavity, Calvert 65 ionospheric reﬂection 4 irregularities density, ﬁeld aligned 5, 142 FAI 142 large scale 147, 156 small scale 147, 154 transverse scale 142 large-scale 91 small scale 19, 91 irregularities, ﬁeld aligned 41 Korteweg-de Vries equation

117, 121

342

Index

laboratory experiments 252 Landau damping 205, 313 lightning 141, 151, 161 linear instability 56 linear process 129 magnetic ﬂux tube 255 magnetized planets 252 magnetosphere 141 cusp 106 magnetopause 40, 106, 252 magnetotail 252 plasmasheet boundary layer 259 tail 106 current sheet 106 tail acceleration region 133 maser electron cyclotron 56, 129, 212, 298 ﬁnite geometry 74 ﬁnite geometry eﬀect 65 measurements burst mode 307 multi-component 298 mode Bernstein-Green-Kruskal (BGK) 121 connection 68 conditions for 73 through cavity boundaries 68 conversion 41, 81, 82, 143, 165, 212, 235 conversion radiation 212 coupling 22, 56, 235 crossover region 235 eigenmodes 212 electrostatic 212 energy leakage 79 extraordinary (X) 4 extraordinary (X,Z) 77, 130 free space 4 growth rate 79 identiﬁcation 70 identiﬁcation of 69 internally unstable 68 mode equation extraordinary (X,Z) 67 ordinary (O) 67 ordinary (O) 4, 41 quasi-electrostatic 17, 149, 193

refraction inside source 72 transmission coeﬃcients upper hybrid 31 W-mode 141 whistler 141, 191 Z-mode ducting 15

81

Nonlinear Schr¨ odinger equation nonthermal mechanism 55

121

Ohm’s law 114 generalized 114, 255 particles auroral energy ﬂux 107 nonthermal 87 pitch angle scattering 165 relativistic 94 ring current protons 165 solar energetic 106 trajectory stochastic 89 phase space ﬂow 64 phase space holes electron holes 205 phase-space holes 116, 120 BGK modes 121 bipolar electric ﬁelds 122 electron holes 121 ion holes 121, 122 observations of 122 simulations 123 tripolar electric ﬁelds 122, 126 width 124 plasma blobs 275 collisionless 251 density 206 inversion method 27 dielectric tensor 88 ionospheric 144 overdense 214 plasmagram 22, 24, 180 plasmapause 22, 37, 145, 314 plasmasphere 144 diﬀusive equilibrium 145 plasmaspheric hiss 166, 173, 176 plasmaspheric notch 45

Index plasmons 121 Poeverlein construction 8, 153 Poisson distribution 277 polar region low altitude 28 Polar spacecraft 298 polarization 235 left hand L 6 linear 195 right hand R 6 ponderomotive force 114 population inversion 56 potential barrier 126 ﬁeld-aligned drop 111 S-shaped 116 U-shaped 116 Poynting ﬂux 300 pressure anisotropy 114 gradient 114 process coherent 115 propagation angle 12 anisotropic 148 dispersive 148 ducted 12, 154 Earth-ionosphere wave guide 158 guiding 142 horizontal 12 inter-hemispheric 22 inter-satellite experiments 208 non-ducted 156 oblique 131 parallel 6, 131 perpendicular 6, 131 ray bending 163 reﬂection 142 refraction 142 scattering 142 stratiﬁed medium 12 transmission cone 153 vertical 12 waveguide duct coupling 171 proton spurs 20 pseudo-potential 116 radar

4

aurora 142 radiation 252 bremsstrahlung coherent 102 incoherent 102 coherent 56, 103, 130 continuum enhancement 40 escaping 40 kilometric 37, 42 non-thermal (NTC) 37 terrestrial 212, 314 trapped 40 dipole 19 direct 212 ﬁne structure 212 impulsive 151 incoherent 205 incoherent Cherenkov 17 incoherent cyclotron 17 indirect 212 isotropic 72 narrow-band 37 radiating diagram 71 solar radio bursts 101 sources location 38 narrow 82 regions 213 synchrotron 41, 100 diﬀusive 100 transition radiation 102 VLF 122, 128, 151 wave guide theory 130 whistler chorus 156 discrete equatorial 156 midlatitude hiss 156 periodic 156 quasi-periodic 156 radiation belts 142, 145 Radio Plasma Imager 157 radio wave cutoﬀ 6 emission from cavities 81 fading 142 interferometry 299 interplanetary bursts 252 pulse 4

343

344

Index

solar bursts 252, 314 trans-ionospheric 142 ray optics 206 ray path 16 ray tracing 169, 299 receivers plasma wave 142 reciprocity principle of 191 reconnection 106, 252 astrophysical implications 264 asymmetric 252 current closure 134 diﬀusion 133 fast 255 ﬂux transfer events, FTE 258 Hall current 133 ion dynamics 133 outﬂow region 253 radiation from 256, 264 reconnection site 133, 255 separatrix 253 small scale 136 tailward 106, 133 three-dimensional 136 whistler 255 whistlers 261 X-line 253 reﬂection from hemispheres 22 conjugate 22 local 22 refractive index 5, 148 cavity 24 relativistic eﬀects 56, 65 resistance anomalous 115, 258 resolution limitations 252 resonance 6, 235 cyclotron 214 DNT 19 group resonance 206 higher nfce 20 inverse 258 Landau 142 lower hybrid 150, 157 lower oblique 191 resonance cone 150, 197, 205

sounder-stimulated 31 upper oblique 7, 191, 207 resonance condition 56 resonator 298 Sagdeev potential 116 separatrix region 253 shock 251 simulations, numerical 252 skin depth 135 ion 135 solar wind 252 solitary waves (ESW) 255 soliton radiation 56 solitons 259 sounding frequency 4 sounding technique 4 space weather 142 spectrum bandwidth 164 dynamic 68 frequency-time spectrogram of inhomogeneity 102 spectral broadening 164 spectral shape 89 stratiﬁed medium 8 horizontally 152 structure formation 252 substorm 55, 106, 129, 133 activity 142 temporal sampling 282 time evolution deterministic 274 top-hat 278 topside sounder 10, 217 transfer equation 81 transmitters VLF 141 transport coeﬃcients 253 trapped electrons 65 Tsyganenko model 23 turbulence 205, 251 energy density 115 magnetic 91 two point study 31 upper hybrid

131

Index band 31 frequency 31, 38 mode 31 resonance 38, 246 W-mode 28 sources 159 waves Alfv´en 110 ampliﬁcation 298 BEN 259 Bernstein 41 broadband emission 131 chirps 216 coherent Langmuir train 313 electron cyclotron 256 harmonics 38, 216, 263 electrostatic 38 electrostatic solitary (ESW) 254 extended packet 322 ﬁnite size packet 316 Gaussian packet 313, 318 incoherent 115 ion acoustic 116, 122 ion cyclotron 116 Landau damping 149

345

Langmuir 214, 254, 256, 259, 313 lower hybrid 149, 164, 256, 274, 277 damping 285 lower hybrid-drift 253 multiplets 214 plane wave analysis 303 polarization 259 saucer emissions 122 shock wave 251 soliton 116, 124 upper hybrid 216, 254, 314 VLF whistlers 125 wave bands 214 wave form 214 wave normal 148 normal angle 153 wave packet 150, 313 whistler 123, 214, 254 magnetospheric whistlers 156 multi-component 156 Z-inﬁnity 7 Z-mode 5, 41 ducting 15 fast 18 slow 16

Lecture Notes in Physics For information about earlier volumes please contact your bookseller or Springer LNP Online archive: springerlink.com

Vol.640: M. Karttunen, I. Vattulainen, A. Lukkarinen (Eds.), Novel Methods in Soft Matter Simulations Vol.641: A. Lalazissis, P. Ring, D. Vretenar (Eds.), Extended Density Functionals in Nuclear Structure Physics Vol.642: W. Hergert, A. Ernst, M. Däne (Eds.), Computational Materials Science Vol.643: F. Strocchi, Symmetry Breaking Vol.644: B. Grammaticos, Y. Kosmann-Schwarzbach, T. Tamizhmani (Eds.) Discrete Integrable Systems Vol.645: U. Schollwöck, J. Richter, D. J. J. Farnell, R. F. Bishop (Eds.), Quantum Magnetism Vol.646: N. Bretón, J. L. Cervantes-Cota, M. Salgado (Eds.), The Early Universe and Observational Cosmology Vol.647: D. Blaschke, M. A. Ivanov, T. Mannel (Eds.), Heavy Quark Physics Vol.648: S. G. Karshenboim, E. Peik (Eds.), Astrophysics, Clocks and Fundamental Constants Vol.649: M. Paris, J. Rehacek (Eds.), Quantum State Estimation Vol.650: E. Ben-Naim, H. Frauenfelder, Z. Toroczkai (Eds.), Complex Networks Vol.651: J. S. Al-Khalili, E. Roeckl (Eds.), The Euroschool Lectures of Physics with Exotic Beams, Vol.I Vol.652: J. Arias, M. Lozano (Eds.), Exotic Nuclear Physics Vol.653: E. Papantonoupoulos (Ed.), The Physics of the Early Universe Vol.654: G. Cassinelli, A. Levrero, E. de Vito, P. J. Lahti (Eds.), Theory and Appplication to the Galileo Group Vol.655: M. Shillor, M. Sofonea, J. J. Telega, Models and Analysis of Quasistatic Contact Vol.656: K. Scherer, H. Fichtner, B. Heber, U. Mall (Eds.), Space Weather Vol.657: J. Gemmer, M. Michel, G. Mahler (Eds.), Quantum Thermodynamics Vol.658: K. Busch, A. Powell, C. Röthig, G. Schön, J. Weissmüller (Eds.), Functional Nanostructures Vol.659: E. Bick, F. D. Steffen (Eds.), Topology and Geometry in Physics Vol.660: A. N. Gorban, I. V. Karlin, Invariant Manifolds for Physical and Chemical Kinetics Vol.661: N. Akhmediev, A. Ankiewicz (Eds.) Dissipative Solitons Vol.662: U. Carow-Watamura, Y. Maeda, S. Watamura (Eds.), Quantum Field Theory and Noncommutative Geometry

Vol.663: A. Kalloniatis, D. Leinweber, A. Williams (Eds.), Lattice Hadron Physics Vol.664: R. Wielebinski, R. Beck (Eds.), Cosmic Magnetic Fields Vol.665: V. Martinez (Ed.), Data Analysis in Cosmology Vol.666: D. Britz, Digital Simulation in Electrochemistry Vol.667: W. D. Heiss (Ed.), Quantum Dots: a Doorway to Nanoscale Physics Vol.668: H. Ocampo, S. Paycha, A. Vargas (Eds.), Geometric and Topological Methods for Quantum Field Theory Vol.669: G. Amelino-Camelia, J. Kowalski-Glikman (Eds.), Planck Scale Effects in Astrophysics and Cosmology Vol.670: A. Dinklage, G. Marx, T. Klinger, L. Schweikhard (Eds.), Plasma Physics Vol.671: J.-R. Chazottes, B. Fernandez (Eds.), Dynamics of Coupled Map Lattices and of Related Spatially Extended Systems Vol.672: R. Kh. Zeytounian, Topics in Hyposonic Flow Theory Vol.673: C. Bona, C. Palenzula-Luque, Elements of Numerical Relativity Vol.674: A. G. Hunt, Percolation Theory for Flow in Porous Media Vol.675: M. Kröger, Models for Polymeric and Anisotropic Liquids Vol.676: I. Galanakis, P. H. Dederichs (Eds.), Halfmetallic Alloys Vol.678: M. Donath, W. Nolting (Eds.), Local-Moment Ferromagnets Vol.679: A. Das, B. K. Chakrabarti (Eds.), Quantum Annealing and Related Optimization Methods Vol.680: G. Cuniberti, G. Fagas, K. Richter (Eds.), Introducing Molecular Electronics Vol.681: A. Llor, Statistical Hydrodynamic Models for Developed Mixing Instability Flows Vol.682: J. Souchay (Ed.), Dynamics of Extended Celestial Bodies and Rings Vol.683: R. Dvorak, F. Freistetter, J. Kurths (Eds.), Chaos and Stability in Planetary Systems Vol.685: C. Klein, O. Richter, Ernst Equation and Riemann Surfaces Vol.686: A. D. Yaghjian, Relativistic Dynamics of a Charged Sphere Vol.687: J. W. LaBelle, R. A. Treumann (Eds.), Geospace Electromagnetic Waves and Radiation

The Lecture Notes in Physics The series Lecture Notes in Physics (LNP), founded in 1969, reports new developments in physics research and teaching – quickly and informally, but with a high quality and the explicit aim to summarize and communicate current knowledge in an accessible way. Books published in this series are conceived as bridging material between advanced graduate textbooks and the forefront of research to serve the following purposes: • to be a compact and modern up-to-date source of reference on a well-deﬁned topic; • to serve as an accessible introduction to the ﬁeld to postgraduate students and nonspecialist researchers from related areas; • to be a source of advanced teaching material for specialized seminars, courses and schools. Both monographs and multi-author volumes will be considered for publication. Edited volumes should, however, consist of a very limited number of contributions only. Proceedings will not be considered for LNP. Volumes published in LNP are disseminated both in print and in electronic formats, the electronic archive is available at springerlink.com. The series content is indexed, abstracted and referenced by many abstracting and information services, bibliographic networks, subscription agencies, library networks, and consortia. Proposals should be sent to a member of the Editorial Board, or directly to the managing editor at Springer: Dr. Christian Caron Springer Heidelberg Physics Editorial Department I Tiergartenstrasse 17 69121 Heidelberg/Germany [email protected]

James W. LaBelle Rudolf A. Treumann (Eds.)

Geospace Electromagnetic Waves and Radiation

ABC

Editors Professor James W. LaBelle Department of Physics Wilder Laboratory 6127 Dartmouth College Hanover, NH 03755 USA E-mail: [email protected]

Professor Rudolf A. Treumann Universität München Sektion Geophysik Theresienstraße 41 80333 München Germany E-mail: [email protected]

J.W. LaBelle and R.A. Treumann, Geospace Electromagnetic Waves and Radiation, Lect. Notes Phys. 687 (Springer, Berlin Heidelberg 2006), DOI 10.1007/b11580119

Library of Congress Control Number: 2005937155 ISSN 0075-8450 ISBN-10 3-540-30050-3 Springer Berlin Heidelberg New York ISBN-13 978-3-540-30050-2 Springer Berlin Heidelberg New York This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, speciﬁcally the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microﬁlm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permission for use must always be obtained from Springer. Violations are liable for prosecution under the German Copyright Law. Springer is a part of Springer Science+Business Media springer.com c Springer-Verlag Berlin Heidelberg 2006 Printed in The Netherlands The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a speciﬁc statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Typesetting: by the authors and TechBooks using a Springer LATEX macro package Printed on acid-free paper

SPIN: 11580119

54/TechBooks

543210

Preface

The “Ringberg Workshop on High Frequency Waves in Geospace” convened at Ringberg Castle, Bavaria, from July 11 to 14, 2004. Approximately 30 attendees from 11 countries gathered at the castle for a program of invited talks and posters focussed on outstanding problems in high-frequency waves, deﬁned broadly as waves exceeding a few kHz in frequency. Thirteen invited presentations comprise the contents of this volume. These articles provide introductions to current problems in geospace electromagnetic radiation, guides to the associated literature, and tutorial reviews of the relevant space physics. As such, this volume should be of value to students and researchers in electromagnetic wave propagation in the environment of the Earth at altitudes above the neutral atmosphere, extending from the ionosphere into outer space. The contributions are broadly grouped into three parts. Part I, entitled “High-Frequency Radiation” focusses on radiation processes in near-Earth plasmas. Benson et al. present a tutorial review of Z-mode emissions, which so far have received relatively little attention and are the subject of few such reviews despite their abundant presence in geospace. Hashimoto et al. continue with another tutorial review on the terrestrial continuum radiation, the relatively weak radio emissions that ﬁll the entire outer magnetosphere and provide information about the magnetospheric plasma boundaries and the state of the magnetospheric plasma density. Louarn reviews the ideas relevant to the generation of Auroral Kilometric Radiation (AKR), by far the most powerful and signiﬁcant of the high-frequency radiations in the magnetosphere. Fleishman introduces the topic of diﬀusive synchrotron radiation, a mechanism not widely appreciated by geophysicists, but which may play a role in several magnetospheric, heliospheric, and even astrophysical settings. Pottelette and Treumann end this chapter with a discussion of the latest ideas about the relationship between auroral acceleration processes and radiation processes such as AKR, a subject which has been transformed in the last decade due to observations with the FAST and CLUSTER satellites. Part II of this monograph, entitled “High-frequency waves,” focusses on wave physics. Sonwalkar presents a lengthy and comprehensive review of

VI

Preface

whistler-mode propagation in the presence of density irregularities. James’ paper deals with recent results from the OEDIPUS-C sounding rocket, combined with recent innovations in antenna theory, which lead to the provocative but signiﬁcant conclusion that ﬁeld strengths measured by many previous observations of auroral hiss using dipole antennas may need to be revised downward. Lee et al. present a novel theoretical method for analyzing modecoupling and mode-conversion of high-frequency waves, with applications to geophysical plasmas. Yoon et al. treat the subject of mode-conversion radiations, which are replete in the Earth’s environment, both in the ionosphere, magnetosphere, and solar wind. Vaivads et al. conclude this part with a review of high-frequency waves related to magnetic reconnection as the generator region of high-frequency waves and radiation in geospace, a very important and hot topic, especially in the light of recent CLUSTER satellite observations. Part III of the monograph is devoted to new analysis techniques and instrumentation transforming research on high-frequency waves. P´ecseli and Trulsen discuss novel ideas on the forefront of linking wave observations to theoretical models. Santol´ık and Parrot apply sophisticated wave propagation analysis tools to the study of AKR. Finally, Kletzing and Muschietti discuss wave particle correlators, describing the physics that can be investigated with them and including results from a recent state-of-the-art wave-particle correlator ﬂown in the Earth’s auroral ionosphere. This monograph would not have been possible without the assistance of the many referees. Special thanks are due to M. Andr´e, R.E. Ergun, J.R. Johnson, E.V. Mishin, R. Pottelette, O. Santol´ık, V.S. Sonwalkar, and A.T. Weatherwax. We thank Dr. Axel H¨ ormann and his team for creating the gracious, welcoming environment at Ringberg Castle, which allowed a creative workshop to take place and thereby inspired this volume. We also thank the International Space Science Institute Bern for support. Finally, the editors at Springer, especially Dr. Christian Caron, deserve thanks for supporting the timely publication of this work and helping to assure its high quality.

Hanover, New Hampshire, and Munich June 2005

James LaBelle Rudolf Treumann

Contents

Part I

High-Frequency Radiation

1 Active Wave Experiments in Space Plasmas: The Z Mode R.F. Benson, P.A. Webb, J.L. Green, D.L. Carpenter, V.S. Sonwalkar, H.G. James, B.W. Reinisch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Plasma Wave Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Sounder-Stimulated Z-Mode Waves in the Topside Ionosphere . . . . . 1.3.1 Single Spacecraft: Vertical and Oblique Propagation . . . . . . . 1.3.2 Single Spacecraft: Ducted Propagation . . . . . . . . . . . . . . . . . . . 1.3.3 Single Spacecraft: Wave Scattering . . . . . . . . . . . . . . . . . . . . . . 1.3.4 Dual Payloads: Slow Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.5 Dual Payloads: Fast Z . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.6 Possible Role of Z-Mode Waves in Sounder/Plasma Interactions . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Sounder-Stimulated Z-Mode Waves in the Magnetosphere . . . . . . . . 1.4.1 Remote O-Z-O-Mode Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2 Local Z-Mode Echoes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.3 Z-Mode Refractive-Index Cavities . . . . . . . . . . . . . . . . . . . . . . . 1.4.4 Whistler- and Z-Mode Echoes . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Active/Passive Investigation of Z-Mode Waves of Magnetospheric Origin . . . . . . . . . . . . . . . . . . . . . 1.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Review of Kilometric Continuum K. Hashimoto, J.L. Green, R.R. Anderson, H. Matsumoto . . . . . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Trapped and Escaping NTC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Continuum Enhancement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Generation Theories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 4 5 10 10 12 15 16 18 19 22 22 22 24 28 29 31 31 37 37 39 40 41

VIII

Contents

2.4 Kilometric Continuum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Geomagnetic Activity Dependence of Kilometric Continuum . . . . . . 2.6 Simultaneous Wave Observations by Geotail and IMAGE . . . . . . . . 2.7 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Generation of Auroral Kilometric Radiation in Bounded Source Regions P. Louarn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Spacecraft Observations in the Sources of the Auroral Kilometric Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Structure of Sources and Wave Properties . . . . . . . . . . . . . . . . 3.2.2 AKR Sources as Regions of Particle Accelerations . . . . . . . . 3.2.3 AKR Sources as Plasma Cavities . . . . . . . . . . . . . . . . . . . . . . . . 3.2.4 Free Energy for the Maser Process . . . . . . . . . . . . . . . . . . . . . . 3.2.5 FAST Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Cyclotron Maser in Finite Geometry Sources . . . . . . . . . . . . . . . . . . . 3.3.1 A Simple Model of AKR Sources . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 Generation in Plasma Cavities: A Simple Approach . . . . . . . . 3.3.3 Observations of Finite Geometry Eﬀects . . . . . . . . . . . . . . . . . 3.4 The Cyclotron Maser in Finite Geometry . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 Mathematical Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.2 Solutions of the Dispersion Relation . . . . . . . . . . . . . . . . . . . . . 3.4.3 The Problem of Energy Escape . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.1 O and X Mode Production in Narrow Sources . . . . . . . . . . . . 3.5.2 Fine Structure of Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.3 Conclusion and Pending Questions . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Generation of Emissions by Fast Particles in Stochastic Media G.D. Fleishman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Statistical Methods in the Theory of Electromagnetic Emission . . . 4.2.1 Spectral Treatment of Random Fields . . . . . . . . . . . . . . . . . . . . 4.2.2 Emission from a Particle Moving along a Stochastic Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.3 Kinetic Equation in the Presence of Random Fields . . . . . . . . 4.2.4 Solution of Kinetic Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Emission from Relativistic Particles in the Presence of Random Magnetic Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 General Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

42 46 49 50 51

55 55 57 57 59 59 61 63 64 65 65 66 68 74 74 78 79 82 82 82 83 83

87 87 88 88 89 90 92 94 94

Contents

IX

4.3.2 Special Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.3.3 Emission from an Ensemble of Particles . . . . . . . . . . . . . . . . . . 97 4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 5 Auroral Acceleration and Radiation R. Pottelette, R.A. Treumann . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 5.2 Morphology of the Auroral Acceleration Region . . . . . . . . . . . . . . . . . 107 5.2.1 Upward Current Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 5.2.2 Downward Current Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 5.3 Parallel Electric Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 5.3.1 Electrostatic Double Layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 5.3.2 Microscopic Structures – Phase Space Holes . . . . . . . . . . . . . . 120 5.3.3 Relation between Holes and Double Layers . . . . . . . . . . . . . . . 125 5.3.4 Holes in the Upward Current Region . . . . . . . . . . . . . . . . . . . . 128 5.4 Auroral Kilometric Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 5.5 The Tail Acceleration Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 5.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 Part II

High-Frequency Waves

6 The Inﬂuence of Plasma Density Irregularities on Whistler-Mode Wave Propagation V.S. Sonwalkar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 6.2 Magnetospheric Plasma Distribution: Field Aligned Irregularities . . 143 6.3 Propagation of Whistler-Mode Waves . . . . . . . . . . . . . . . . . . . . . . . . . . 147 6.3.1 Propagation of Plane Whistler-Mode Waves in Uniform Magnetoplasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 6.3.2 Propagation of Whistler-Mode Waves in the Magnetosphere 151 6.4 Observations and Interpretations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 6.4.1 W-Mode Observations When the Source is Below the Ionosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 6.4.2 W-Mode Observations When the Source is in the Magnetosphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 6.5 Discussion and Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . 181 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 7 Dipole Measurements of Waves in the Ionosphere H.G. James . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 7.2 Summary of OEDIPUS-C Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 7.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

X

Contents

7.2.2 Whistler-Mode Propagation Near the Resonance Cone . . . . . 197 Retrospective on Past Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 7.3.1 Dipole Eﬀective Length for EM Propagation . . . . . . . . . . . . . . 199 7.3.2 Intensity of Auroral Hiss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 7.3.3 Intersatellite Whistler-Mode Propagation Near a Resonance Cone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206 7.3.4 Intersatellite Slow-Z-Mode Propagation Near a Resonance Cone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207 7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209 7.3

8 Mode Conversion Radiation in the Terrestrial Ionosphere and Magnetosphere P.H. Yoon, J. LaBelle, A.T. Weatherwax, M. Samara . . . . . . . . . . . . . . . . 211 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 8.2 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213 8.2.1 Auroral-Zone Mode-Conversion Radiation . . . . . . . . . . . . . . . . 213 8.2.2 Magnetospheric Mode-Conversion Radiation . . . . . . . . . . . . . . 218 8.3 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222 8.3.1 Nonlocal Theory of Electrostatic Waves in Density Structures222 8.3.2 Trapped Langmuir Waves with Discrete Frequency Spectrum225 8.3.3 Discrete Upper-Hybrid Waves in Density Structures . . . . . . . 228 8.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 9 Theoretical Studies of Plasma Wave Coupling: A New Approach D.-H. Lee, K. Kim, E.-H. Kim, K.-S. Kim . . . . . . . . . . . . . . . . . . . . . . . . . . 235 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235 9.2 Numerical Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 9.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 9.4 Invariant Embedding Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240 9.5 Application of IEM to the Mode-Conversion of O and X Waves . . . 243 9.6 Discussion and Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 10 Plasma Waves Near Reconnection Sites A. Vaivads, Yu. Khotyaintsev, M. Andr´e, R.A. Treumann . . . . . . . . . . . . . 251 10.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251 10.1.1 Observations of Diﬀerent Wave Modes . . . . . . . . . . . . . . . . . . . 253 10.2 Numerical Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 10.3 Lower Hybrid Drift Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 10.3.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 10.3.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 10.3.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 10.4 Solitary Waves and Langmuir/Upper Hybrid Waves . . . . . . . . . . . . . 259

Contents

XI

10.4.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 10.4.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 10.4.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 10.5 Whistlers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.5.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.5.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.5.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 10.6 Electron Cyclotron Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 10.6.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 10.6.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 10.6.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 10.7 Free Space Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 10.7.1 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 10.7.2 Generation Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 10.7.3 Relation to Reconnection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265 10.8 Summary and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267 Part III High-Frequency Analysis Techniques and Wave Instrumentation 11 Tests of Time Evolutions in Deterministic Models, by Random Sampling of Space Plasma Phenomena H.L. P´ecseli, J. Trulsen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 11.2 Model Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 11.2.1 Spatial Sampling with One Probe Available . . . . . . . . . . . . . . 277 11.2.2 Spatial Sampling with Two Probes Available . . . . . . . . . . . . . 282 11.2.3 Temporal Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 11.3 Nonlinear Lower-Hybrid Wave Models . . . . . . . . . . . . . . . . . . . . . . . . . 283 11.4 Probability Densities for Observables . . . . . . . . . . . . . . . . . . . . . . . . . . 286 11.5 Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290 11.6 Discussions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292 11.6.1 Length Scales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 11.6.2 Time Scales . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294 11.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 12 Propagation Analysis of Electromagnetic Waves: Application to Auroral Kilometric Radiation O. Santol´ık, M. Parrot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 12.2 Analysis Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 12.2.1 Plane Wave Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 300 12.2.2 Wave Distribution Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304

XII

Contents

12.3 Analysis of Auroral Kilometric Radiation . . . . . . . . . . . . . . . . . . . . . . . 305 12.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 13 Phase Correlation of Electrons and Langmuir Waves C.A. Kletzing, L. Muschietti . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313 13.2 Finite-Size Wave Packet in a Vlasov Plasma . . . . . . . . . . . . . . . . . . . . 316 13.2.1 Linear Perturbation of the Electrons . . . . . . . . . . . . . . . . . . . . . 317 13.2.2 Case of a Gaussian Packet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 13.3 Extended Wave Packet: A BGK Analysis . . . . . . . . . . . . . . . . . . . . . . . 322 13.3.1 Passing Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 13.3.2 Trapped Electrons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324 13.4 Electron Phase Sorting Measurements . . . . . . . . . . . . . . . . . . . . . . . . . 324 13.4.1 Measurements of the Resistive Component . . . . . . . . . . . . . . . 325 13.4.2 Measurements of the Reactive Component . . . . . . . . . . . . . . . . 328 13.5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 13.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 335 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339

List of Contributors

Allan T. Weatherwax Siena College, Dept. Physics 515 Loudon Rd. Loudonville, NY 12211-1462, USA [email protected] Andris Vaivads Swedish Institute of Space Physics Box 537, Uppsala, SE 751 21, Sweden [email protected] Bodo W. Reinisch University of Massachusetts Lowell, MA 01854-0000, USA [email protected] Craig Kletzing University of Iowa/Phys. & Astronomy Iowa City, IA 52242-0000, USA [email protected] Donald L. Carpenter Stanford University Star Lab. Electr. Engng. Dept. Stanford, CA 94305-9515 [email protected] D.-H. Lee Kyung Hee University Dept. Astron. Space Sci. Yongin, Kyunggi 449-701, Korea [email protected]

E.-H. Kim Kyung Hee University Dept. Astron. Space Sci. Yongin, Kyunggi 449-701, Korea [email protected] Gregory D. Fleishman 26 Polytekhnicheskaya St. Petersburg 194021 Russian Federation [email protected] Hans L. Pecseli University of Oslo/Inst. Physics Box 1048, Blindern, N-0316 Oslo, Norway [email protected] Hiroshi Matsumoto Res. Inst. Sustainable Humanosphere Kyoto University Uji, Kyoto 611-0011, Japan [email protected] H. Gordon James Communications Research Center Canada Ottawa, Ontario KH2 882, Canada [email protected]

XIV

List of Contributors

James LaBelle Dartmouth College Dept. Phys. Astron. Wilder Laboratory Hanover, New Hampshire 03755 USA [email protected] James L. Green NASA/Goddard Space Flight Center MC 630, Bldg. 26, Greenbelt, MD 20771-1000, USA [email protected] Jan Trulsen University of Oslo Institute of Theoretical Astrophysics Box 1029 Blindern N-0315 Oslo, Norway [email protected] Kozo Hashimoto Res. Inst. Sustainable Humanosphere Kyoto University Uji, Kyoto 611-0011, Japan [email protected] K. Kim Ajou University Dept. Molecular Sci. Techn. Suwon, Kyunggi 443-749, Korea [email protected] K.-S. Kim Kyung Hee University Dept. Astron. Space Sci. Yongin, Kyunggi 449-701, Korea [email protected] Laurent Muschietti University of California Space Sciences Laboratory Berkeley, CA 94720-7450, USA [email protected]

Marilia Samara Dartmouth College Dept. Phys. Astron., Wilder Laboratory Hanover, NH 03755, USA [email protected] Mats Andr´ e Swedish Institute of Space Physics Box 537, Uppsala, SE 751 21 [email protected] Michel Parrot LPCE/CNRS 3a av. de Recherche Science 45071 Orleans, Cedex 02, France [email protected] Ond˘ rej Santol`ık Charles Univ. Prague Faculty of Math. & Phys. Prague 8, CZ-18000 Czech Republic [email protected] Peter H. Yoon University of Maryland Inst. Physical Sci. & Techn. College Park, MD 20742, USA [email protected] Philippe Louarn CNRS/CESR 9 av. Colonel Roche Toulouse, 31329 France [email protected] Phillip A. Webb NASA/GSFC Code 612.3 Greenbelt, MD 20771-0000, USA [email protected] Raymond Pottelette CNRS/CETP 4 av. de Neptune St. Maur des Foss´es Cedex, France [email protected]

List of Contributors

Robert F. Benson NASA/Goddard Space Flight Center MC 692, Bldg. 21, Room 252 Greenbelt, MD 20771-0001, USA [email protected] Roger R. Anderson University of Iowa Dept. Phys. & Astron. Iowa City, IA 52242-1479, USA [email protected] Rudolf A. Treumann Ludwig-Maximilians Universit¨ at M¨ unchen, Sektion Geophysik Theresienstr. 41, 80333 Munich Germany [email protected]

XV

Vikas S. Sonwalkar Univ. Alaska Fairbanks Dept. Electr. Engng. 306 Tanana Dr. Room 229 Duckering Fairbanks, AK 99775-0000, USA [email protected] Yuri V. Khotyaintsev Swedish Inst. Space Physics Box 537, Uppsala 75121, Sweden [email protected]

Part I

High-Frequency Radiation

1 Active Wave Experiments in Space Plasmas: The Z Mode R.F. Benson1 , P.A. Webb1 , J.L. Green1 , D.L. Carpenter2 , V.S. Sonwalkar3 , H.G. James4 , and B.W. Reinisch5 1

2 3 4 5

NASA/Goddard Space Flight Center, Greenbelt, MD USA [email protected] Stanford University, California USA University of Alaska Fairbanks, Alaska USA Communications Research Centre, Toronto, Canada University of Massachusetts Lowell, Massachusetts USA

Abstract. The term Z mode is space physics notation for the low-frequency branch of the extraordinary (X) mode. It is an internal, or trapped, mode of the plasma conﬁned in frequency between the cutoﬀ frequency fz and the upper-hybrid frequency fuh which is related to the electron plasma frequency fpe and the electron 2 2 2 = fpe + fce ; fz is a function of fpe cyclotron frequency fce by the expression fuh and fce . These characteristic frequencies are directly related to the electron number density Ne and the magnetic ﬁeld strength |B|, i.e., fpe (kHz)2 ≈ 80.6Ne (cm−3 ) and fce (kHz)2 ≈ 0.028|B|(nT). The Z mode is further classiﬁed as slow or fast depending on whether the phase velocity is lower or higher than the speed of light in vacuum. The Z mode provides a link between the short wavelength λ (large wave number k = 2π/λ ) electrostatic (es) domain and the long λ (small k) electromagnetic (em) domain. An understanding of the generation, propagation and reception of Z-mode waves in space plasma leads to fundamental information on wave/particle interactions, Ne , and ﬁeld-aligned Ne irregularities (FAI) in both active and passive wave experiments. Here we review Z-mode observations and their interpretations from both radio sounders on rockets and satellites and from plasma-wave receivers on satellites. The emphasis will be on the scattering and ducting of sounder-generated Z-mode waves by FAI and on the passive reception of Z-mode waves generated by natural processes such as Cherenkov and cyclotron emission. The diagnostic applications of the observations to understanding ionospheric and magnetospheric plasma processes and structures beneﬁt from the complementary nature of passive and active plasma-wave experiments.

Key words: Auroral kilometric radiation, Z-mode, free space radiation, wave transformation, radiation escape, cavity modes, active experiments

R.F. Benson et al.: Active Wave Experiments in Space Plasmas: The Z Mode, Lect. Notes Phys. 687, 3–35 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

4

R.F. Benson et al.

1.1 Introduction According to cold plasma theory, at high frequencies there are two characteristic electromagnetic (em) waves, or modes, that can propagate in a magnetoplasma. They are often referred to as the free-space ordinary (O) and extraordinary (X) modes because waves propagating in these modes can smoothly connect to free space. The X mode has two branches. In addition to the freespace mode, it has a mode called the slow branch. This name is used because it is restricted to propagation velocities less than the vacuum speed of light c. Since this mode only exists within a plasma, there was considerable interest in explaining observations indicating that it was responsible for a unique signature on early ground-based radars designed to probe the ionosphere. In their most common application these radars, called ionosondes, operate by transmitting a radio pulse of short time duration at a particular frequency and receiving, at the same frequency, for a time interval suﬃcient to receive an echo from the ionosphere overhead. This process is repeated over a range of frequencies likely to produce reﬂections. The resulting record is called an ionogram. Normally, there are two ionospheric reﬂections, one due to the O mode and one due to the X mode. “On rare occasions”, as ﬁrst reported by Eckersley [23], there is a third reﬂection with the same polarization as the O mode. This third reﬂection trace, corresponding to the slow X-mode branch, was dubbed the Z mode in ionospheric research; a designation commonly used in space physics. In order to explain the presence of the Z mode at ground level, i.e., far below the ionospheric plasma, and the polarization (same as the O mode), a Z-O mode coupling process involving obliquely-propagating Omode waves was introduced by Ellis [24] as discussed in Sect. 13.5 of Ratcliﬀe [57]. An ionogram showing this triple splitting of the ionospheric reﬂection is schematically illustrated in Fig. 1.1. Here the apparent height h (or apparent, or virtual, range) corresponds to ct/2, where t is the round trip echo delay time, and the frequency f is the sounding frequency. For a description of the sounding technique, and the inversion from h (f ) to Ne (h), where Ne is the electron number density and h is the true altitude, see Thomas [64] and Reinisch [58] and references therein. In order to understand how the Z mode is related to the free-space O and X modes it is necessary to discuss plasma-wave dispersion. This topic will be addressed in Sect. 1.2. Since the Z mode is an internal (or trapped) mode of the plasma, the emphasis in this paper will be on the reception of the Z mode by space-borne receivers during active and passive experiments. Sections 1.3 and 1.4 will deal with sounder-stimulated Z-mode waves in the ionosphere and the magnetosphere, respectively. Particular attention will be given to the information that the sounder-stimulated Z-mode waves provide concerning magnetic-ﬁeld aligned Ne irregularities (FAI). FAI are irregularities in Ne transverse to the direction of the background magnetic ﬁeld B that are maintained for long distances along B. They eﬃciently scatter and

1 Active Wave Experiments in Space Plasmas: The Z Mode

Apparant Height (h')

Z

O

5

X

fH Sounding Frequency (f)

Fig. 1.1. Ground-based ionogram schematic illustrating Z-, O-, and X-mode reﬂection traces. Here the ionospheric notation for the electron cyclotron frequency, fH , is used [adapted from 57]

duct sounder-stimulated Z-mode waves. Section 1.5 discusses a combined active/passive investigation of Z-mode waves generated by natural processes. A summary is presented in Sect. 1.6. There have been many spacecraft that have generated Z-mode waves in the ionosphere and magnetosphere using radio sounders. Similarly, there have been many satellites that have detected Z-mode waves of magnetospheric origin using plasma wave receivers [LaBelle and Treumann, 43, included a review of auroral Z-mode observations and theory]. Our goal is not to review the Z-mode observations from all of these missions. Rather, it is to select speciﬁc examples that illustrate the range of Z-mode phenomena observed in active space wave-injection experiments and to demonstrate their diagnostic capability. In the case of the ionosphere, we will mainly use data from two missions, namely, (1) the ISIS (International Satellites for Ionospheric Studies) satellites [Jackson and Warren, 33] and (2) the OEDIPUS sounding rocket double payloads (Observations by Electric-ﬁeld Determinations in the Ionospheric Plasma-A Unique Strategy) [see, e.g., 30, 36]. In the case of the magnetosphere, data from the Radio Plasma Imager (RPI) [Reinisch et al., 59] on the IMAGE (Imager for Magnetopause-to-Aurora Global Exploration) satellite [Burch, 15] will be used.

1.2 Plasma Wave Dispersion Waves in a cold plasma are described by a dispersion relation, i.e., the scalar relation expressing the angular frequency ω = 2πf in terms of the propagation vector k, which is related to the refractive index n by n = kc/ω where k = |k| = (2π/λ) and λ is the wavelength.This description has been given in a number of books and review papers [see, e.g., 1, 19, 27, 39, 57, 63]. Figure 1.2 presents dispersion curves for waves propagating in a homogeneous cold

6

R.F. Benson et al. ο

ο ο θ = 0 90 0

R-X

ο

θ=0

(a)

c

(b)

c

θ = 90 θ=0

ο

fx fuh fce

CMA 4

Z (X) Z (R-X) L-O

ο

θ=0

Whistler (R)

θ = 90 ο

CMA 3

θ =0

CMA 6a CMA 7

fpe

ο

ο

θ =0

)

fce

ο

CMA 3 Z (X)

L-X )

fz

ο

0

Z (L-X

L-O

Z(

f

ο

R-X

fx fuh fpe

90

fz

Whistler (R)

k

k

fpe > fce

fpe < fce

Fig. 1.2. Schematic dispersion diagrams. (a) Example when fpe /fce > 1. (b) Example when fpe /fce < 1 [adapted from 27, 62] (Reprinted with permission of the American Geophysical Union)

plasma, where ion motions are neglected, with k making an angle θ relative to B. The ﬁgure shows the dispersion curves for θ = 0 and θ = π/2 cases for a range of frequency and wave number. The region between these two limiting cases is shown by various shades of gray – indicating various modes – where the propagation at oblique wave normal angles is permitted. The waves are labelled based on their polarization for parallel or perpendicular propagation, i.e., R or L for right- or left-hand polarization (with respect to the direction of B) when θ = 0, and X or O for extraordinary and ordinary mode polarization when θ = π/2. In some regions, only one letter is used indicating that propagation is not possible for both θ = 0 and θ = π/2. Thus Z(X) indicates that the Z mode does not include the condition θ = 0 in the region indicated based on the cold-plasma approximation. The Z-mode regions in Fig. 1.2 are also labelled with the CMA designation using the notation of Stix [63]. Thus Z(X) occurs in CMA region 3 where k, and thus n = |n|, can become large leading to a condition (n = ∞) known as resonance; the condition k = 0 (or n = 0) is known as a cutoﬀ. The plasma resonances and cutoﬀs in Fig. 1.2 are given by the following expressions: fce (kHz) =

|e| |B| ≈ 0.028 |B(nT)| 2πme )

(1.1)

1 Active Wave Experiments in Space Plasmas: The Z Mode

7

1

2 1 1 e2 fpe (kHz) = Ne2 ≈ 80.6 Ne (cm−3 ) 2 (4π 2 0 me )

fuh

=

fx

=

fz

1

2 2 2 (fpe + fce )

12 2 fpe 1+4 2 fce

fce 1+ 2 12 2 fpe fce ≡ fx − fce −1 + 1 + 4 2 = 2 fce

(1.2) (1.3) (1.4)

(1.5)

where e is the electron charge, me is the electron mass and 0 is the permittivity of free space. For θ < π/2, the resonance condition that replaces (3) above, in CMA 3 of Fig. 1.2, is known as the Z-inﬁnity and is given by 12 1 1 2 4 2 2 fuh + (fuh − 4fce fpe cos2 θ) 2 fZI = √ 2

(1.6)

The Z inﬁnity is also referred to as the upper oblique resonance [see, e.g., Beghin et al., 4]. The above cutoﬀs and resonances are described using diﬀerent notations in Sects. 6.4 and 6.5 of Ratcliﬀe [57] and Sects. 1–5 of Stix [63]. Figure 1.2 is often presented in the form of ω vs. k. In this presentation the magnitudes of the phase and group velocities, ω (1.7) |vp | = k and

∂ω |vg | = , ∂k

(1.8)

respectively, correspond to the slope of the line from the origin to a particular point on a dispersion curve, and to the slope of a line tangent to the dispersion curve at that point, respectively. In Fig. 1.2 there is a slanting line labelled c to indicate that it corresponds to free-space propagation. The curves to the left of this line (labelled R-X, L-O and L-X) have vp ≥ c and those to the right have vp < c. Accordingly, the Z-mode waves labelled L-X are called fast Z (CMA 4 in Fig. 1.2a and CMA 7 in Fig. 1.2b) and those labelled Z(X) are called slow Z (CMA 3 in Figs. 1.2a and 2b). Both fast and slow Z-mode waves can be found in the region labelled R-X (CMA 6a in Fig. 1.2b). The ionosphere and magnetosphere, contrary to the conditions appropriate to Fig. 1.2, are neither homogeneous nor cold and the ions are not motionless. Yet the dispersion properties derived by using these assumptions, and illustrated by Fig. 1.2, have proved very successful in describing many phenomena. The standard approach is to consider the ionosphere as a horizontallystratiﬁed medium with Ne varying only in the vertical direction. Then the wave is considered to behave as if it were in a homogeneous medium at each

8

R.F. Benson et al.

ionospheric level. Both diagrams in Fig. 1.2 correspond to a speciﬁc value of fpe /fce . The curves change shape as fpe /fce changes. For example, in Fig. 1.2a, the √ band of no propagation between fce and fZ only appears when fpe /fce > 2. Also, the Z(X) region, i.e., CMA-region 3 in Figs. 1.2a and 1.2b, is maximum for the condition fpe = fce . The progress of a radio wave through the non-homogeneous ionosphere can be modelled by considering the change in the shape of the refractive-index surface as fpe /fce changes. This process is illustrated in Fig. 1.3 to illustrate how Z-mode signals at a frequency f from a high-latitude source of natural origin could propagate over great distances in the horizontal direction. The conditions correspond to a source location where f ≈ fpe < fce . The left side of the diagram shows the evolution of the refractive-index surfaces from low to high altitudes corresponding to plasma conditions changing from CMA regions 7 to 6a to 3 in Fig. 1.2b. Gurnett et al. [28] used a construction technique introduced by Poeverlein [54], based on Snell’s law, to argue that a wave at frequency f originating in the region where f < fce will be refracted at the f = fce level and will be able to propagate long distances in the horizontal direction. In Fig. 1.3, the arrows originating at the intersections of the vertical dashed line with these refractive index surfaces indicate the direction of vg in (benson-eq8). Note the change from a closed refractive index surface to an open surface as CMA region 3 is encountered.

RESONANCE

f = f uh

Bo CMA 3

f = f ce

INDEX OF REFRACTION SURFACE CMA 6a

Vg

f = f pe

SOURCE

CMA 7

f = fz Z-MODE CUTOFF

Fig. 1.3. Illustration of the large change in the shape of the refractive-index surface as CMA region 3 is encountered and the ability for long-range horizontal propagation in the Z mode in the polar regions where B is nearly vertical [adapted from 28] (Reprinted with permission of the American Geophysical Union)

1 Active Wave Experiments in Space Plasmas: The Z Mode

9

NORMALIZED FREQUENCY f/fce

6

5 4 3

2

1 0 10 -4

10 -3

10 -2

10 -1

1

10

NOR M ALIZE D WAV E NUMBER kR

Fig. 1.4. Normalized calculated dispersion diagram for a wide range of wavelengths with electron thermal motions included where f is normalized by fce and k is normalized by 1/R where R is the electron cyclotron radius for the case fpe /fce = 1.6 [adapted from Oya, 53] (Reprinted with permission of the American Geophysical Union)

Figure 1.4 shows the result of numerical solutions of the dispersion relations, for a particular case of fpe /fce > 1, for all θ, and for θ ranging from 0 to π/2, when hot-plasma eﬀects are included. In these solutions, electron thermal motions are included and a Maxwellian electron velocity distribution function is assumed but collisions are neglected and the ions are still considered to provide an immobile neutralizing background [Oya, 53]. The right portion of the diagram corresponds to the electrostatic (es) domain and the left portion corresponds to the electromagnetic (em) domain. New wave modes, known as the Bernstein modes [Bernstein, 13], now appear between the nfce harmonics in the es-domain. The Bernstein modes correspond to undamped modes with θ = π/2. Damping rapidly increases for these modes as θ departs from π/2. These es-modes are coupled to the em-domain, with negligible damping, through the Z-mode when θ ≈ π/2, corresponding to frequencies close to fuh , each, in turn, as fuh increases for increasing fpe /fce values. Thus the Z-mode near θ = π/2, i.e., near fuh , is of prime importance in the coupling of energy resulting from wave/particle interactions in the es-domain into the em-domain where the information can be transmitted out of the plasma.

10

R.F. Benson et al.

1.3 Sounder-Stimulated Z-Mode Waves in the Topside Ionosphere 1.3.1 Single Spacecraft: Vertical and Oblique Propagation In a space plasma such as the ionosphere the Z mode can be directly detected by ionospheric topside sounders. Figure 1.5a presents an example of a midlatitude ionogram where ionospheric reﬂections form a clear Z mode trace in addition to O- and X-mode traces. Also seen in Fig. 1.5a are sounderstimulated plasma resonances at fce = fH , fpe = fN , fuh = fT and 2fce = 2fH and an oblique Z trace labelled Z’. The resonances can be used to accurately determine the ambient |B| and Ne values from (1.1)–(1.3) as, e.g., given in the reviews of Muldrew [50] and Benson [7]. The presently accepted interpretation for these principal plasma resonances stimulated by ionospheric topside sounders is based on the investigation by Calvert [16] of the Z trace. He showed that this trace, which lies between fpe and fuh , is the result of ionospheric reﬂections of obliquelypropagating Z-mode waves. These oblique reﬂections result from the shape of the refractive index surfaces, in particular from the change in the shape to form a sphere plus a line parallel to B when the downward-propagating Z-mode wave encounters the level where f = fpe (see the left side of Fig. 1.3). The Z trace is caused by waves reﬂected at this level. This condition was called a “spitze” by Poeverlein [54] and is discussed in some detail in Budden [14]. The Z trace could be explained by using ray tracing with the cold plasma theory. Calvert [16] did not restrict his calculations to the cold plasma approximation, however, and found ray paths that could return to the spacecraft that included propagation beyond the resonance-cone angle limit of cold plasma theory. These echoes were electrostatic in nature and had echo times much greater than the observed Z echoes so the solutions were discarded. McAfee [46], in his investigation of the plasma resonance observed at fpe , found that these hot-plasma solutions had echo delay times comparable to the delays observed on topside ionograms when frequencies very close to fpe were considered. This oblique-echo model was later extended to the plasma resonances observed at fuh [47] and at nfce [50]. Thus, the investigation of oblique Zmode propagation by Calvert [16] provided the fundamental ﬁrst step toward our understanding of sounder-stimulated plasma resonances. Even though the Z-mode is deﬁned as the slow branch of the X mode, it is important to note that the Z and X modes have comparable group velocities near their respective cutoﬀs. This behavior is clearly illustrated in Fig. 1.5b. Though Z-mode waves cannot travel as far from the spacecraft as the freespace O- and X-mode waves, they are useful for determining the vertical Ne distribution out to a few hundred km from the spacecraft as seen in Fig. 1.5c. The good agreement between the Ne values obtained in Fig. 1.5c by inverting the Z-, O- and X-mode ionospheric-reﬂection traces of Fig. 1.5a provides

1 Active Wave Experiments in Space Plasmas: The Z Mode

11

Fig. 1.5. (a) Alouette-2 mid-latitude ionogram, in negative format with signal reception in white on a black background, showing Z, O and X traces. Here the ionospheric notation of H, N , and T is used to represent the subscripts ce, pe, and uh, respectively, in the present work. Also, “S” is used to designate the cutoﬀ frequencies at the satellite. The fzI label and arrow below the ionogram, and near the 0.9 MHz frequency marker, identiﬁes the Z inﬁnity condition given by (6). (b) Corresponding group velocities (Ne = 6800 cm−3 corresponds to fpe = 0.74 MHz from (2), B = 0.254 = 0.254 × 10−4 T = 0.254 × 105 nT corresponds to fce = 0.7112 MHz from (1) and fuh = 1.026 MHz from (3)). (c) Calculated Ne values from each of the traces [adapted from Jackson, 32]

12

R.F. Benson et al.

conﬁdence that the two main assumptions used in the inversion process, namely, vertical propagation and a horizontally-stratiﬁed ionosphere, were justiﬁed since this process is independent for each trace [Jackson, 32]. For vertical propagation the propagation angle relative to B is 90◦ -dip angle, or φ = 51.4◦ using Jackson’s notation of Fig. 1.5b. Using this φ value in (6), with the other values from Fig. 1.5b where fpe /fce = 1.04, yields fZI = 0.968 MHz in agreement with the observed narrow vertical modulated feature observed from zero to approximately 3,000 km apparent range; it is labelled fzI just below the ionogram. This feature can be observed because the wide receiver bandwidth (3 dB bandwidth of 37 kHz) allows long-duration signals returning from the previous pulse (diﬀering by only a few kHz) to be observed at the start of the receiving interval, i.e., it corresponding to a wrap-around of the apparent-range scale for the asymptotic Z-mode echo. Figure 1.6 shows examples of Z-mode echoes that clearly illustrate the limiting behavior of the Z-mode cold-plasma dispersion curves in Figs. 1.2 and 1.4 for conditions of nearly parallel and nearly perpendicular propagation relative to B. In Fig. 1.6a, corresponding to high latitude and thus nearly parallel propagation, the Z- and O-mode traces touch one another near 700 km apparent range and 0.9 MHz suggesting coupling like in the dispersion diagrams in Figs. 1.2 and 1.4 for θ = 0. Also, the Z trace has a large apparent range at this frequency, coinciding with the combined fpe and fce plasma resonances, as would be expected from (1.6) with θ = 0, i.e., fZI = fpe = fce . From the observed plasma resonances and wave cutoﬀs in Fig. 1.6a, and equations (1.3)–(1.5), fpe /fce = 0.92/0.935 = 0.98 for this ionogram. In Fig. 1.6b, corresponding to low latitude and thus perpendicular propagation, the Z-mode trace becomes asymptotic to fuh , again, as expected from (1.6) (now with θ = π/2). In this case, fpe /fce = 1.89/0.565 = 3.35. When nearly parallel or nearly perpendicular propagation is involved in the presence of FAI, dramatic ionogram signatures can be produced due to ducting and scattering, respectively, of the sounder-generated Z-mode signals. 1.3.2 Single Spacecraft: Ducted Propagation When ionospheric topside sounders encounter equatorial plasma bubbles, dramatic ﬂoating X-mode echoes are observed that resemble epsilons [Dyson and Benson, 22]. They are called ﬂoating because they are not tied to the zerotime baseline at the top of the ﬁgure. These traces are the result of soundergenerated signals that echo in both the local and conjugate hemispheres (relative to the location of the satellite) due to ducted propagation in FAI that are maintained from one hemisphere to the other. The bottom portion of Fig. 1.7 illustrates the top segment of such an X-mode epsilon in the frequency range above about 2.9 MHz and at apparent ranges beyond 2200 km. The X-mode echo just above this echo signature, i.e., corresponding to virtual ranges less than 2200 km in the bottom portion of Fig. 1.7, is due to ducted propagation in the local hemisphere. In the top portion of Fig. 1.7 the distances to the

1 Active Wave Experiments in Space Plasmas: The Z Mode

13

Fig. 1.6. (a) ISIS-2 high-latitude ionogram showing Z, O and X traces under conditions of nearly parallel propagation (Resolute Bay digital ionogram obtained from http://nssdc.gsfc.nasa.gov/space/isis/isis-status.html corresponding to 1500:28 UT on day 126 of 1973; 62.8◦ latitude, −101.8◦ longitude, 1394 km altitude, 83◦ dip). (b) Low-latitude Alouette-2 ionogram showing Z, O and X traces under conditions of nearly perpendicular propagation [adapted from Benson, 6]. As in Fig. 1.5, the ionospheric notation of N , and T is used to represent the subscripts pe, and uh, respectively, and numerals are used to identify the nfH = nfce resonances. The 3fce resonance contains two spurs on the low-frequency side (see Sect. 1.3.6) with delay times near 4 and 7 ms. The insert shows ﬁve selected receiver amplitude vs. time traces corresponding to these spur observations. Each trace represents a vertical scan line on the ionogram display, the amplitude modulation on the insert traces corresponding to the intensity modulation on the ionogram scan lines. The initial systematic positive and negative spikes on each of the insert traces are calibration and sync pulses [see Fig. 31 of Franklin and Mclean, 25]; the time-delay zero point was taken as the left side of the dashed line segment on the insert trace labelled (a). The nfce resonances indicated that fce = 0.565 MHz corresponding to τp = 3.25 ms (Reprinted with permission of American Geophysical Union)

14

R.F. Benson et al.

Fig. 1.7. ISIS-1 ionograms recorded 1000 km above the dip equator showing Zand X-mode echoes from within an equatorial plasma bubble [9]. The ionospheric notation of H, and N is used to represent the subscripts ce and pe, respectively; numerals are used to identify the nfH = nfce resonances (Reprinted with permission of American Geophysical Union)

reﬂection levels in each hemisphere were the same, as the two traces merge and have the same virtual ranges from about 2200 to 2400 km. Z-mode echoes from waves that are ducted along FAI can form truncated versions of these ﬂoating X-mode epsilons. In both portions of Fig. 1.7 the ducted Z-mode waves are conﬁned to a narrow frequency range below the label Z at the top of the ﬁgure. As in the case of the ducted X-mode signals, the Z-mode traces tied to the zero virtual-range scale correspond to signals ducted within the local hemisphere and those beyond about 2000 km in virtual range correspond to signals that experience ducted propagation into both

Virtual Range (km)

1 Active Wave Experiments in Space Plasmas: The Z Mode

0

H

2

15

1112 13 3 4 5 6 7 8 9 10 ZNX 14

1000 2000 3000 4000 0.1 0.2

0.5 0.55

0.9

1.25 1.5 1.6

2.5 2.0

4.5

6.5 7.0

8.5 10.5

Frequency (MHz) Fig. 1.8. A portion of an Alouette-2 ionogram showing Z-mode ﬂoating echoes, attributed to wave ducting in FAI, from fZ , to a maximum frequency prior to fpe labelled Z and N at the ﬁgure, respectively. The label notation is the same as in Fig. 1.7 [adapted from Benson, 9] (Reprinted with permission of the American Geophysical Union)

hemispheres. Note that the ducted Z-mode traces terminate before they reach the local fpe , designated by N at the top of the ﬁgure. Similar frequency restrictions of the reception of sounder-stimulated Z-mode waves attributed to wave ducting were commonly observed in an investigation of FAI near 500 km based on Alouette-2 perigee observations [9]; an example is shown in Fig. 1.8. In a comparison of wave ducting in diﬀerent wave modes, assuming small propagation angles inside a duct produced by a small increase in refractive index, Calvert [17] showed that under conditions of high Ne , similar to the conditions of Fig. 1.8, Z-mode ducting should be stronger than X- and O-mode ducting. The strongest Z-mode ducting was found to occur in the frequency range from fZ to midway between fZ and fpe where a transition from trough ducting to crest ducting occurs due to a curvature reversal in the refractiveindex surface. Calvert [17] argues that ducting cannot be maintained across the curvature reversal in agreement with the upper-frequency truncation of the ﬂoating Z-mode signals attributed to FAI wave ducting in Fig. 1.7 and Fig. 1.8. 1.3.3 Single Spacecraft: Wave Scattering Sounder-generated Z-mode waves that are scattered by FAI lead to strong signal returns in the frequency region between the greater of fce and fpe and less than fuh , i.e., in CMA region 3 in Figs. 1.2 and 1.4. Z-mode ray-tracing calculations indicate that the ray becomes horizontal in the ionosphere as the refractive index tends to inﬁnity [Lockwood, 45] leading to a condition Muldrew [49] termed “wave trapping”. The resulting signature is labelled as a noise band in Fig. 1.9. Denisenko et al. [21] used such signatures in the

16

R.F. Benson et al.

fN

fH

fT

APPARENT RANGE (Km)

0

1000

2000

3000 NOISE BAND

4000

0.2

0.5

0.9

1.25

1.5 1.6

2.0

3.5

FREQUENCY (MHz) o

o

29 DECEMBER , 1967 2257/14 GMT (99 W, 59 N) SATELLITW HEIGHT 1550Km

Fig. 1.9. An Alouette-2 ionogram showing sounder-generated Z-mode signal returns attributed to wave scattering in FAI (labelled “noise band”) [Muldrew, 49]. The label notation is the same as in Fig. 1.5a (Copyright 1969 by IEEE; reprinted with permission)

COSMOS-1809 topside-sounder data to investigate the global distribution of small-scale (∼10–100 m) FAI in the 940 to 980 km altitude range. 1.3.4 Dual Payloads: Slow Z The above discussion pertained to observations where the receiver and transmitter were on the same spacecraft and they shared a common antenna. Thus the received Z-mode signals correspond to echo returns either from short distances, due to scattering from FAI in the vicinity of the satellite (Fig. 1.9), or from long distances (∼100’s km) due to vertical propagation (Z trace in Fig. 1.5a), oblique propagation (Z trace in Fig. 1.5a), or to nearly parallel propagation that is ducted along FAI (Figs. 7 and 8). The Z mode has also been investigated during space experiments involving wave propagation between receivers and transmitters on diﬀerent payloads. James [34, 35] performed such experiments during a high-latitude rendezvous between ISIS 1 and ISIS 2. The operating frequency was in the range fpe < fce < f < fuh , i.e., corresponding to “slow Z-mode” propagation in CMA-region 3 of Fig. 1.2b. These waves were observed to propagate over distances of several hundred km. It was found [James, 35] that the observed transmission-reception signal delay times could be explained by ray optics but that the observed distortion of the received pulse relative to the transmitted pulse indicated the importance of signal scattering by FAI. The OEDIPUS-A dual-payload rocket was launched from the Andøya Rocket range in Norway on January 30, 1989. It was dedicated to such

1 Active Wave Experiments in Space Plasmas: The Z Mode

17

two-point measurements and provided an additional opportunity to investigate slow Z-mode propagation [James, 36]. In this case, fce < fpe < f < fuh , i.e., corresponding to slow Z-mode propagation in CMA-region 3 of Fig. 1.2a. The transmitting and receiving payloads were separated by nearly 1 km and the separation direction diﬀered from the B direction by only a few degrees. The observed delay times for the Z-mode were too large to be explained by free-space propagation and the full em solution of the hot plasma dispersion equation, based on the work of Lewis and Keller [44] and Muldrew and Estabrooks [51], was used to investigate the problem. Using this hot-plasma approach, James [36] constructed refractive-index surfaces appropriate to the problem (see Fig. 1.10). Note how diﬀerent these hot-plasma Z-mode refractive-index surfaces are from the cold plasma surfaces illustrated on the left side of Fig. 1.3 in the CMA-region 3. James [36] found that direct ray paths connecting the two payloads had |vg | values too small to explain the observed delays in the large n region assuming a smooth horizontally-stratiﬁed medium. Thus waves corresponding to such large-n dispersion solutions would arrive well after the OEDIPUS A ionogram display time limit (by approximately a factor of 10) and would not be detected. When the region of smaller n was investigated, labelled “electromagnetic and quasi-electrostatic domain” in Fig. 1.10, and the payloads were assumed to be within an Ne depletion duct (with cross-B dimension ∼ 100 m), ducted ray paths could be found that were consistent with the observations. James [36] suggested that such ducting may be common in the auroral ionosphere and that it should be considered when trying to interpret natural Z-mode emissions. The OEDIPUS-C rocket dual-payload was launched from the Poker Flat rocket range in Alaska on November 7, 1995. Again, the sounder-transmitter was on one payload and the sounder-receiver was on the other, and the separation direction between the payloads, now separated by more than a 1 km, differed from the B direction by only a few degrees. In this case, James [37] investigated the slow Z-mode propagation corresponding to fpe < fce < f < fuh , i.e., to the CMA-region 3 of Fig. 1.2b. He found, using hot-plasma dispersion theory, that the calculated propagation times for rays directly connecting the observed payloads were typically more than a factor of three greater than the observed time delays between signal transmission and reception. Thus waves corresponding to such solutions would arrive well beyond the observing time base and would not be detected. In all other cases, there were no solutions corresponding to the desired direction. The received signals could be explained, however, in terms of incoherent Cherenkov and cyclotron radiation from sounder-accelerated electrons (SAE). Particle detectors on both payloads detected SAE following sounder transmissions from the transmitting payload. James [37] could reproduce the observed signal delay times and could predict (within an order of magnitude) the observed signal intensities.

18

R.F. Benson et al.

Fig. 1.10. Hot-plasma Z-mode refractive-index surfaces for f = 2.534 MHz, fce = 1.2 Hz, Te = 2000 K and fpe , labelled fp in the ﬁgure, in the range 2.233 = fpe = 2.533 MHz [James, 36] (Reprinted with permission of the American Geophysical Union)

1.3.5 Dual Payloads: Fast Z Horita and James [29, 30] investigated fast Z-mode propagation using OEDIPUS-C data corresponding to frequencies below fpe and above the greater of fZ or fce , i.e., to CMA-region 4 in Fig. 1.2a where there are no competing cold-plasma wave modes to complicate the interpretation. The Z

1 Active Wave Experiments in Space Plasmas: The Z Mode

19

mode was found to be stronger than almost all the other cold-plasma modes and to be strongest at frequencies just below fpe for fpe /fce near and greater than 1 but strongest just above fZ for higher fpe /fce values (between 2 and 3). Using cold-plasma dispersion theory, the Balmain [2] antenna-impedance theory and the Kuehl [40] dipole radiation theory they found that the observed and calculated signal intensities were generally in good agreement. They attribute the strength of the Z-mode signals, relative to the other free-space modes, to antenna-impedance values that permit eﬃcient coupling between the antenna and the transmitter and receiver. 1.3.6 Possible Role of Z-Mode Waves in Sounder/Plasma Interactions Among the plasma instability and nonlinear phenomena stimulated by ionospheric topside sounders investigated by Benson [8] was a diﬀuse feature observed in the frequency range above the greater of fce and fpe and below fuh , i.e., in the CMA 3 slow Z-mode regions of Fig. 1.2. It was designated as the DNT resonance because of its generally diﬀuse appearance on topside ionograms and its location between fN and fT (ionospheric notation for fpe and fuh ). A weak short-duration (< 1 ms) example of this resonance is shown in Fig. 1.6a. It is often observed for up to about 5 ms. It is not observed over the entire listening range, however, and thus is distinguished from the noise band in this frequency range. This noise band, attributed to wave scattering (see Sect. 1.3.3), is evident in Fig. 1.6a and, more prominently, in Fig. 1.9. No theoretical interpretation has been oﬀered that explains the frequency and time-duration characteristics of this resonance. It has been attributed by Pulinets [55], Pulinets et al. [56] to the scattering of sounder-generated Z-mode waves and used as a diagnostic tool for the investigation of the distribution of small-scale FAI in the topside ionosphere. James [38] attempted to explain the DNT resonance as observed by the ISIS-II sounder in terms of radiation from SAE in analogy with the successful explanation of Z-mode signals observed by the OEDIPUS-C sounder receiver as described in Sect. 1.3.4. While this explanation was not found to explain the ISIS-II observations, he concluded that SAE may still play a role because SAE pulses that persisted for milliseconds were observed when the OEDIPUS-C sounder transmitter was tuned to the DNT frequencies. Features have been observed on topside ionograms that imply that ion motions must be considered for a proper interpretation. They appear either as prominent protrusions (called spurs) on the electron resonances (most often from the low-frequency side) or as narrow (in time delay, i.e., apparent range) emissions between the resonances. In either case, they appear with delay times that correspond to multiples of the proton gyroperiod 1/fcp . One of these phenomena, the proton spurs on the nfce resonances, appears to be strongly inﬂuenced by Z-mode transmissions [Benson, 6]. The spurs are

20

R.F. Benson et al.

greatly enhanced when fZ , from (1.5), is near, but slightly less than, nfce for n = 2, 3, 4, . . . and the largest spurs are observed for large n. Figure 1.6 illustrates the spurs observed on the 3fce resonance when fZ ≈ 3fce . It was suggested that the Z mode may be more eﬃcient at coupling energy into the plasma under these conditions. Note that this frequency region just above fZ corresponds to the fast Z region where Horita and James [29, 30] found the strongest Z-mode signals which they attributed to optimum antenna-impedance values (see Sect. 1.3.5). Their larger fpe /fce values, which produce the strongest signals just above fZ , would correspond to larger fZ values which, in turn, would correspond to higher nfce resonances satisfying the condition fZ ≈ nfce ; the largest proton spurs were observed under just such conditions [Benson, 6]. Unique Z-mode topside-ionogram signatures have been observed in low latitudes that suggest topside sounders are capable of stimulating, or enhancing, FAI when they encounter the plasma conditions fpe /fce ≈ n where n is an integer larger than 3 [Benson, 10]. An illustration for the case of n ≈ 5 is presented in Fig. 1.11. Note that well-deﬁned Z, O and X traces are clearly seen for the cases fpe /fce < 5 (top panel) and fpe /fce > 5 (bottom panel) but that the Z trace is masked by a long-duration diﬀuse signal that extends from fZ to part way to fpe . It was argued that the frequent occurrence of such signatures made it unlikely that the spacecraft was just encountering FAI when the ambient conditions fpe /fce ≈ n were satisﬁed. Thus the sounder-generated Z-mode waves were considered to be ducted in FAI stimulated, or enhanced, by the sounder on a short time scale (1 s). The possibility that this sensitive diagnostic role of the Z-mode waves could be due to eﬃcient scattering when fpe /fce ≈ n was investigated by Zabotin et al. [69]. They did not ﬁnd any sensitivity in the scattering of Zmode waves by FAI near the magnetic equator to the conditions fpe /fce ≈ n and concluded that the above examples were either due to sounder stimulation, or enhancement, as proposed or to ducting conditions sensitive to these conditions. No study of the sensitivity of Z-mode ducting by FAI to the conditions fpe /fce ≈ n has been made. As pointed out in Sect. 1.3.2, however, Calvert [17] found that under high Ne conditions (as indicated in Fig. 1.11) Zmode ducting in the frequency range between fZ and midway to fpe should be stronger than O- or X-mode ducting, a prediction consistent with the observations in the middle panel of Fig. 1.11. Thus the sensitivity to the fpe /fce ≈ n condition is likely in the generation, or enhancement of existing, FAI by the sounder. Benson [10] gave other examples of sounder-stimulated plasma phenomena when fpe /fce ≈ n and suggested that more energy is deposited into the plasma under these conditions and, particularly, when fpe /fce ≥ 4. Osherovich et al. [52] investigated large-amplitude cylindrical electron oscillations appropriate to FAI, with initial conditions chosen so as to favor Z-mode stimulation, and found that the resulting frequency spectrum was very sensitive to the fpe /fce value, with larger amplitudes, and more nonlinear frequency components, observed when fpe /fce ≈ n, and that the eﬀect

1 Active Wave Experiments in Space Plasmas: The Z Mode

21

Fig. 1.11. Consecutive Alouette-2 low-latitude ionograms revealing long-duration Z-mode echoes only when fpe /fce ≈ 5 at 05:15 UT on 17 June 1969 [Benson, 10]. The label notation is the same as in Fig. 1.7 (Reprinted with permission from Elsevier)

was observed to increase with increasing n. Kuo et al. [41] investigated the creation of FAI by a Z-mode pump wave under the above resonant conditions. They added the constraint that the Z-mode pump wave frequency fo ≈ fpe at a short distance from the satellite where a four-wave coupling process takes place. One of the products in this coupling process corresponds to short-scale (meter size) FAI. They propose that the observed Z-mode diﬀuse signals, like those in the middle panel of Fig. 1.11, are caused by scattering from these FAI but stress the need for additional research to identify mechanisms that could generate (or enhance) large-scale FAI (>100 m) capable of supporting wave ducting.

22

R.F. Benson et al.

1.4 Sounder-Stimulated Z-Mode Waves in the Magnetosphere 1.4.1 Remote O-Z-O-Mode Coupling Radio sounding in the magnetosphere is challenging because the distances are so large and Ne is so low. These constraints motivated the RPI design for the IMAGE mission. As a result, the IMAGE satellite contains the largest structures ever placed on a spinning satellite, namely, the RPI spin-plane crossed dipole antennas (originally 500 m tip-to-tip length for each). Soon after the IMAGE launch on 25 March 2000 into an elliptical polar orbit, with an apogee of 8 Earth radii (RE ) geocentric distance and a perigee altitude of about 1000 km, the RPI detected discrete long-range echoes outside the plasmasphere in the north polar region. Reinisch et al. [60] and Carpenter et al. [19] attributed them to signals propagating in the X mode in FAI to the polar ionosphere where they were reﬂected; they were clearly distinguished from the diﬀuse shorter-range echoes from the nearby highly-irregular plasmapause boundary. Later, echo signatures indicating inter-hemisphere propagation, similar to the ionospheric example shown in Fig. 1.7, were identiﬁed by Fung et al. [26]; an example where the Z mode played a prominent role in the interpretation is shown in Fig. 1.12 from Reinisch et al. [61]. This record is called a plasmagram and is the magnetospheric analog of the ionospheric topside-sounder ionogram examples shown in Figs. 1.5–1.9 and 1.11. The virtual range vs. frequency curves through the data points corresponding to X-mode reﬂections from the local and conjugate hemispheres (labelled SX and N X, respectively) were used to derive the hemisphere-tohemisphere Ne distribution along the magnetic ﬁeld line through the satellite [31]. This ﬁeld-aligned Ne distribution was then used to calculate the reﬂections expected for transmitted O-mode signals that coupled to Z-mode signals, which reﬂected at the distances where the transmitted frequencies were equal to the fZ cutoﬀ frequencies given by (1.5), and then coupled back to the O-mode signals that were received at the satellite. These O-Z-O traces are labelled N Z and SZ for the echoes of this type from the northern and southern hemispheres, respectively, in Fig. 1.12. This mode-coupling interpretation of Reinisch et al. [61], involving the Z-mode to explain the weaker companion echoes to the inter-hemisphere magnetospheric RPI X-mode echoes, differs from the interpretation of Muldrew [48] of inter-hemisphere ionospheric echoes observed by Alouette 1 in that Muldrew [48] did not invoke O-Z-O mode coupling. 1.4.2 Local Z-Mode Echoes Echoes of Z-mode signals are often directly observed by RPI in the magnetosphere. Figure 1.13 shows an example when IMAGE was near the plasmapause and a strong Ne gradient could be determined from multiple plasma

1 Active Wave Experiments in Space Plasmas: The Z Mode

23

RPI Plasmagram 17:50 October 24, 2000 7.0

SX+NX

Virtual Range (RE)

6.0

NZ

5.0

NP

4.0

NX

SP SZ

3.0 2.0

SX

ce

2

4

1.0 30

100

300

1000

Frequency (kHz) Fig. 1.12. An RPI plasmagram showing echoes from both hemispheres in the frequency region above about 300 kHz, with interpretive traces through the data points, and short-duration plasma resonances at fce , 2fce and 4fce (lower left). The labels S, N, X, and Z denote southern and northern hemispheres and X and O-Z-O traces respectively. The N X trace has been extrapolated to higher frequencies. The insert shows the orbit, location of IMAGE (x) and the L = 4 dipole ﬁeld lines; NP and SP indicate the magnetic poles [adapted from Reinisch et al., 61] (Reprinted with permission of the American Geophysical Union)

resonances and wave cutoﬀs. The frequency range of this high-resolution plasmagram fortuitously included the Z- and X-mode cutoﬀs (fZ and fX , respectively) and several plasma resonances. (The linear frequency step size between transmissions of single 3.2 ms pulses was equal to the RPI bandwidth of 300 Hz.) The plasma conditions corresponded to Fig. 1.2a; the Z mode from fZ to fpe is in CMA region 4 and is the only cold-plasma wave mode. The diﬀuse nature is attributed to scatter returns from FAI. These scatter returns become more prominent in the CMA region 3 between fpe and fuh (see Fig. 1.2a). This enhanced scatter forms a noise band analogous to the noise band in the ionospheric example of Fig. 1.9 (which, however, corresponds to the CMA 3 region of Fig. 1.2b). Using the values scaled from Fig. 1.13 in (1.3)– (1.5) reveals that consistent solutions cannot be obtained for these equations with constant fce and fpe values over the 41-s time interval required to record this plasmagram. Three independent fpe determinations can be made, however, if the Tsy 96–1 model magnetic-ﬁeld values [Tsyganenko, 65, 66, 67],

24

R.F. Benson et al.

Fig. 1.13. An RPI plasmagram recorded during an outbound plasmapause crossing revealing Z- and X-mode cutoﬀs and plasma resonances at fpe , fuh , and 4fce labelled at the top by z, x, pe, uh, and 4, respectively. Also labelled, as Q3, at the top is one of the resonances associated with the Bernstein modes discussed in Sect. 1.2 in connection with Fig. 1.4 [adapted from Benson et al. 11] (Reprinted with permission of the American Geophysical Union)

with a percentage oﬀset correction based on the observed 4fce plasma resonance in Fig. 1.13, are used corresponding to the spacecraft locations at the times of the recording of fZ , fuh , and fX . The plasma resonance observed at fpe in Fig. 1.13 provides a fourth measurement of fpe and it is independent of the fce value. While the deduced Ne gradient is large (an order-of-magnitude decrease in a change in L value of approximately 1) [Benson et al., 11], it is only about 1/10 the gradient of a well-developed plasmapause [Carpenter and Anderson, 18]. 1.4.3 Z-Mode Refractive-Index Cavities Among the most spectacular echo signatures observed on RPI plasmagrams are those corresponding to the direct transmission and reception of Z-mode waves that are ducted along FAI within refractive-index cavities [Carpenter et al., 20]. Examples are shown in Fig. 1.14. They were obtained when RPI was operating with a linear frequency step size of 900 Hz. From the observed fZ values in Fig. 1.14 (corresponding to the onset of the Z-mode traces), and the model values for fce (oﬀ scale to the right in both plasmagrams), it is deduced from (1.5) that fpe is also oﬀ scale to the right in both plasmagrams. In both cases, fpe < fce , i.e., propagation corresponding to CMA region 7 in Fig. 1.2b is involved.

1 Active Wave Experiments in Space Plasmas: The Z Mode

25

4.0 (a)

3.0 2.0

50 40

fZ

0

fZ

4.0

30 20

3.0

AmpX (dBnV/m)

VIRTUAL RANGE (RE )

60

1.0

10 W

2.0

0

1.0 (b)

0 180

200 220 FREQUENCY (kHz)

240

Fig. 1.14. RPI plasmagrams showing multicomponent Z-mode echoes recorded near perigee in the region of Ne gradients between the southern aurora zone and the plasmasphere. (a) L ≈ 3.2, altitude ∼3800 km, fZ = 194 kHz, fce (model) = 382 kHz implies fpe ≈ 334 kHz or fpe /fce ≈ 0.9; 0824 UT on 26 July 2001. (b) L ≈ 2.9, altitude ∼ 2700 km, fZ = 216 kHz, fce (model) = 469 kHz implies fpe ≈ 384 kHz or fpe /fce ≈ 0.8; 0245 UT on 12 July 2001 (whistler-mode echoes, due to reﬂections from the bottom side of the ionosphere [62], are marked “W”) [adapted from Carpenter et al. 20] (Reprinted with permission of the American Geophysical Union)

The virtual ranges of the observed echoes starting near 1.5 RE , which appear as upward slanting epsilons in Figs. 1.14a and 1.14b, are too short to be explained in the same manner as used for the echoes shown in Fig. 1.12, i.e., they are too short to be attributed to echoes from the conjugate hemisphere. They can be explained, however, in terms of ducted echoes returned from within a refractive-index cavity in the hemisphere containing the IMAGE satellite. The interpretation presented by Carpenter et al. [20] is illustrated in Fig. 1.15 for the case when the IMAGE satellite is assumed to be located below a relative minimum in the proﬁle of fZ along B, as deduced from (1.5) with fce and fpe values based on models and RPI observations. The most prominent features in Fig. 1.14a are reproduced in Fig. 1.15a with labels corresponding to the ray paths deﬁned in Fig. 1.15b which displays

26

R.F. Benson et al.

Fig. 1.15. (a) Reproduction of the most prominent echoes in Fig. 1.14a. (b) Schematic fZ proﬁle limiting ray paths A and B for representative frequencies fi , fj , and fk , relative to the location of the IMAGE satellite. These ray paths correspond to the labels used for the echo traces in (a) [Carpenter et al., 20] (Reprinted with permission of American Geophysical Union)

an idealized fZ proﬁle along B. When the sounder frequency reaches the frequency fi , corresponding to the condition fi = fZ at the satellite level as illustrated in Fig. 1.15b, a wave can propagate upward along path B into the region where fi > fZ and be reﬂected at a higher altitude where the condition fi = fZ is again satisﬁed. This returning wave is responsible for the echo with a virtual range of ≈ 1.5 RE , i.e., the nose of the ﬁrst upward slanting epsilon signature in Fig. 1.15a. This wave is reﬂected again at the fi = fZ condition at the satellite and the process is repeated. Two such repetitions are evident in the data of Fig. 1.14a and the reproduction in Fig. 1.15a. Higher sounder frequencies, such as fj in Fig. 1.15b, correspond to the condition fj > fZ at the satellite level, and waves can now propagate both upward along path B and downward along path A, within the region where fj > fZ , and be reﬂected at both higher and lower altitudes where the condition fj = fZ

1 Active Wave Experiments in Space Plasmas: The Z Mode

27

Fig. 1.16. Ne along the L = 3.2 magnetic-ﬁeld line above the IMAGE satellite derived from inverting the “B” Z trace of Fig. 1.15a compared with an RPI-derived empirical model of Huang et al. [31] for a diﬀerent day (8 June 2001) and a scaling of that model by a factor of 0.7. The portion of the lowest curve corresponding to magnetic latitude values less than 13◦ is an extrapolation [Carpenter et al. 20] (Reprinted with permission of the American Geophysical Union)

is satisﬁed. Multiple combinations of these echoes produced the elements of the epsilon signatures seen in Figs. 1.14a and 1.15a. Carpenter et al. [20] also presented examples of a special form of echo signature on RPI plasmagrams that corresponds to ducted Z-mode propagation along FAI within Z-mode refractive-index cavities when IMAGE is assumed to be located above the relative minimum in the proﬁle of fZ along B. Carpenter et al. [20] introduced an inversion method to determine the Ne distribution along B from the upward propagating signals within Z-mode refractive-index cavities of the type discussed above. The results of applying this method to the trace corresponding to B in Fig. 1.15a are presented in Fig. 1.16. Also presented in Fig. 1.16 are predicted values from an empirical Ne model based on the inversion of RPI X-mode echoes from signals that propagated on multiple ﬁeld-aligned paths on a diﬀerent day, namely, 8 June 2001 [Huang et al., 31]. The lower Ne values derived from the Z-mode data were attributed to the movement of IMAGE through a region of plasmapause Ne gradients at the time of the measurements. Since these Z-mode echoes from waves trapped in Z-mode refractive-index cavities are often the only prominent echoes observed on a single plasmagram, this inversion method provides a valuable diagnostic tool for determining the Ne distribution along B.

28

R.F. Benson et al.

1.4.4 Whistler- and Z-Mode Echoes In an investigation of IMAGE/RPI data in the inner plasmasphere and at moderate to low altitudes over the polar regions, Sonwalkar et al. [62] found diﬀuse Z-mode echoes often accompanied whistler (W)-mode echoes. An example from their study is shown in Fig. 1.17. The W-mode echo in this ﬁgure, with narrowly deﬁned time delay as a function of frequency, is an example of a discrete echo. The Z-mode echo, with a time-delay spread that increases with frequency, is an example of a diﬀuse echo. As discussed by Carpenter et al. [20], this Z-mode pattern is characteristic of the low altitude polar region and the plasma condition fpe /fce < 1. The abrupt high-frequency cutoﬀ of this Z-mode echo, and the long-duration sounder-stimulated plasma resonance at ∼787 kHz in Fig. 1.17, provides a measure of fuh [Benson et al., 14]; the gap, or decrease in echo spreading at ∼685 kHz, provides a measure of fce [Carpenter et al., 20]. From these values fpe is calculated to be ∼387 kHz from (1.3). Whistler-mode echoes with a much broader range of time delays with frequency than those shown in Fig. 1.17 are also observed on IMAGE. They are called diﬀuse W-mode echoes [Sonwalkar et al., 62]. In regions poleward of the plasmasphere, diﬀuse Z-mode echoes of the kind illustrated in Fig. 1.17 were found to accompany both discrete and diﬀuse W-mode echoes 90% of the time, and were also present during 90% of the soundings when no W-mode echoes were detected. Based on comparisons of ray tracing simulations with the observed dispersion of W- and Z-mode echoes, Sonwalkar et al. [62] proposed that: (1) the observed discrete W-mode echoes are due to RPI signal reﬂections from the

Fig. 1.17. RPI plasmagram displaying both W-mode echoes (frequencies below ∼300 kHz) and diﬀuse Z-mode echoes (frequencies above ∼300 kHz) labelled WM and ZM, respectively. The minimum observable time delay is 13 ms due to the 3.2ms minimum transmitted pulse length and additional time needed for the receiver to recover from the high voltage generated during the transmitter pulses. The amplitude scale is coded from 10 to 50 dB nV/m [adapted from Sonwalkar et al., 62] (Reprinted with permission of the American Geophysical Union)

1 Active Wave Experiments in Space Plasmas: The Z Mode

29

lower boundary of the ionosphere, (2) the diﬀuse W-mode echoes are due to scattering of RPI signals by FAI located within 2000 km earthward of IMAGE and in directions close to that of the ﬁeld line passing through IMAGE, and (3) the diﬀuse Z-mode echoes are due to scattering of RPI signals from FAI within 3000 km of IMAGE, particularly to signals propagating in the generally cross-B direction. This interpretation suggests that Z-mode echoes occur most frequently (∼ 90%), both in the presence and absence of whistler-mode echoes, because the Z-mode waves capable of returning to the sounder can propagate long distances in all directions, i.e., not only close to the ﬁeld lines as in the case of the whistler mode waves that are capable of returning to the sounder. Thus there is a much larger probability of encountering plasma irregularities which may lead to Z-mode echoes. These RPI results are consistent with previous investigations in that they indicate that the high-latitude magnetosphere is highly structured with FAI that exist over cross-B scales ranging from 10 m to 100 km and that these FAI profoundly eﬀect W- and Z-mode propagation.

1.5 Active/Passive Investigation of Z-Mode Waves of Magnetospheric Origin The ﬁrst observation of enhanced Z-mode radio emissions of natural origin, corresponding to CMA-region 3 in Fig. 1.2, were made during the radioastronomy rocket experiment of Walsh et al. [68]. They ruled out a thermal source due to the large signal intensities and suggested Cherenkov radiation as a likely source mechanism because of the large refractive index (and hence low wave phase velocities) in this frequency domain. Bauer and Stone [3], using satellite observations, were the ﬁrst to show that the observed frequency limits of this CMA region 3 Z-mode radiation could be used to determine the magnetospheric Ne . Several later experiments have investigated these emissions attributed to CMA region 3 Z-mode radiation by comparing the observed frequencies with fpe values determined by active techniques. Beghin et al. [4] used the AUREOL/ARCADE-3 mutual impedance probe in the 400–2000 km altitude region, Kurth et al. [42] used a sounder during a brief (5 min) period of the single pass through the terrestrial magnetosphere by Cassini, and Benson et al. [12] used active soundings by the RPI on four passes of IMAGE in the vicinity of the plasmapause region. In each of these investigations, it was concluded that the upper and lower frequency boundaries of an observed intense upper-hybrid band corresponded to fuh and fpe , respectively, in the region where fpe > fce . In the study by Benson et al. [12] these frequency identiﬁcations were found to hold to an accuracy of a few per cent in fpe by interpolating between active soundings to the intervening passive dynamic spectra. Figure 1.18 shows the results of superimposing the plasmagram-determined fce , fpe and fuh values from active

30

R.F. Benson et al. 200 100

dBV/ Hz

50

10

-90

5 200

a 25 March 2003

100

fuh

50 20

fce

10 fpe

4 2 0 -2 -4

-120 -130 -140

02

-150

00

0 -2 -4 -6 -8 X (RE)

5 0000

-100 -110

Z (RE)

Frequency (kHz)

20

b 0030

0100

0130

0200

UT Fig. 1.18. (a) Passive RPI dynamic spectrum. (b) Same as (a) except with superimposed fce , fpe and fuh values determined from active RPI plasmagrams [adapted from Benson et al., 12] (Reprinted with permission of the American Geophysical Union)

sounding on the passive RPI dynamic spectrum corresponding to the same time interval. Comparing Figs. 1.18a and 1.18b illustrates the beneﬁt of having active soundings to conﬁdently determine fpe , particularly when fpe < fce . The sounder-derived fpe values (white triangles) follow the upper edge of an intense, presumably whistler mode, emission. They deviate from this upper edge, however, at a location that would be diﬃcult to determine without the active soundings. Also, in this fpe < fce frequency domain, the upper-hybrid band enhancement, in this case between fce and fuh , is often not very well deﬁned as fuh approaches fce . Beghin et al. [4] never observed an enhancement when fpe < fce ; they attribute this ﬁnding to a lack of instability growth of Z-mode waves in the upper-hybrid band under these conditions. The upperhybrid band is relatively broad (extending from ∼50 to 60 kHz) near 00:30 UT where fpe ∼ fce , and it narrows in bandwidth as time progresses. The scaled fpe and fuh frequencies in the region beyond about 00:30 UT, i.e., in the region

1 Active Wave Experiments in Space Plasmas: The Z Mode

31

where fpe > fce , allow the boundaries of the upper-hybrid band to be identiﬁed and distinguished from the slanting ﬁnger-like higher-frequency emissions which are attributed to Bernstein-mode emissions discussed in connection with Fig. 1.4. Figure 1.18 suggests that there are two diﬀerent sources of the observed W-mode emissions, one more intense than the other. The most intense one extends out to slightly beyond 01:00 UT and is limited by the minimum in fpe near 10 kHz as determined from the active sounding. The weaker one extends out to about 01:45 UT and is limited by fce from this point backward in time to 00:30 where it is limited by fpe . At earlier times, the weak emissions in the frequency domain from fpe to fuh could be either L-O or CMA region 6 a Z-mode emissions (see Fig. 1.2b). Conﬁrming identiﬁcations of passive dynamic-spectral features by nearlysimultaneous active determinations of fpe , such as illustrated in Fig. 1.18, provides conﬁdence in the interpretation of the passive dynamic spectra when supporting active measurements are not available. It also provides conﬁdence in the use of the passive dynamic spectra to help interpret plasmagrams when the spectrum of sounder-stimulated resonances is complex [Benson et al., 11].

1.6 Summary Even though the Z mode is an internal, or trapped, mode of the plasma it has valuable diagnostic applications in space plasmas in both active and passive wave experiments. In active experiments discrete Z-mode echo traces can be inverted to provide Ne proﬁles, diﬀuse traces provide information about FAI, and two-point propagation studies provide information concerning wave propagation, wave ducting and wave/particle interactions. In passive experiments, intense Z-mode signals of magnetospheric origin provide valuable ambient Ne information.

Acknowledgements We are grateful to the reviewer for many helpful comments on the manuscript. The work at the University of Alaska Fairbanks was supported by NASA under contract NNG04GI67G. Support for B. W. R. was provided by NASA under subcontract 83822 from SwRI.

References [1] Allis, W.P., S.J. Buchsbaum, and A. Bers: Waves in anisotropic plasmas, MIT Press, Cambridge, 1963.

32

R.F. Benson et al.

[2] Balmain, K.G.: Dipole admittance for magnetoplasma diagnostics, IEEE Trans. Antennas Propag. 17, 389–392, 1969. [3] Bauer, S.J. and R.G. Stone: Satellite observations of radio noise in the magnetosphere, Nature 218, 1145–1147, 1968. [4] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and HF auroral hiss generated at low altitude, J. Geophys. Res., 94, 1359–1379, 1989. [5] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and HF auroral hiss generated at low altitude, J. Geophys. Res. 94, 1359–1379, 1989. [6] Benson, R.F.: Ion eﬀects on ionospheric electron resonance phenomena, Radio Sci. 10, 173–185, 1975. [7] Benson, R.F.: Stimulated plasma waves in the ionosphere, Radio Sci. 12, 861– 878, 1977. [8] Benson, R.F.: Stimulated plasma instability and nonlinear phenomena in the ionosphere, Radio Sci. 17, 1637–1659, 1982. [9] Benson, R.F.: Field-aligned electron density irregularities near 500 km – Equator to polar cap topside sounder Z mode observations, Radio Sci. 20, 477–485, 1985. [10] Benson, R.F.: Evidence for the stimulation of ﬁeld-aligned electron density irregularities on a short time scale by ionospheric topside sounders, J. Atm. and Solar-Terr. Phys. 59, 2281–2293, 1997. [11] Benson, R.F., V.A. Osherovich, J. Fainberg, and B.W. Reinisch: Classiﬁcation of IMAGE/RPI-stimulated plasma resonances for the accurate determination of magnetospheric electron-density and magnetic ﬁeld values, J. Geophys. Res. 108, 1207, doi:10.1029/2002JA009589, 2003. [12] Benson, R.F., P.A. Webb, J.L. Green, L. Garcia, and B.W. Reinisch: Magnetospheric electron densities inferred from upper-hybrid band emissions, Geophys. Res. Lett. 31, L20803,doi:1029/2004GL020847, 2004. [13] Bernstein, I.B.: Waves in a plasma in a magnetic ﬁeld, Phys. Rev. 109, 10–21, 1958. [14] Budden, K.G.: The propagation of radio waves, the theory of radio waves of low power in the ionosphere and magnetosphere, 669 pp., Cambridge University Press, New York, 1985. [15] Burch, J.L.: The ﬁrst two years of IMAGE, Space Sci. Rev. 109, 1–24, 2003. [16] Calvert, W., Oblique z-mode echoes in the topside ionosphere, J. Geophys. Res. 71, 5579–5583, 1966. [17] Calvert, W.: Wave ducting in diﬀerent wave modes, J. Geophys. Res., 100 (A9), 17,491–17,497, 1995. [18] Carpenter, D.L. and R.R. Anderson: An ISEE/whistler model of equatorial electron density in the magnetosphere, J. Geophys. Res. 97, 1097–1108, 1992. [19] Carpenter, D.L., M.A. Spasojevic, T.F. Bell, U.S. Inan, B.W. Reinisch, I.A. Galkin, R.F. Benson, J.L. Green, S.F. Fung, and S.A. Boardsen: Small-scale ﬁeld-aligned plasmaspheric density structures inferred from RPI on IMAGE, J. Geophys. Res., 107(A9), 1258, doi:10.1029/2001JA009199, 2002. [20] Carpenter, D.L., T.F. Bell, U.S. Inan, R.F. Benson, V.S. Sonwalkar, B.W. Reinisch, and D.L. Gallagher: Z-mode sounding within propagation “cavities” and other inner magnetospheric regions by the RPI instrument on the IMAGE satellite, J. Geophys. Res. 108, 1421, doi:10.1029/2003JA010025, 2003.

1 Active Wave Experiments in Space Plasmas: The Z Mode

33

[21] Denisenko, P.F., N.A. Zabotin, D.S. Bratsun, and S.A. Pulinets: Detection and mapping of small-scale irregularities by topside sounder data, Ann. Geophysicae 11, 595–600, 1993. [22] Dyson, P.L. and R.F. Benson: Topside sounder observations of equatorial bubbles, Geophys. Res. Lett. 5, 795–798, 1978. [23] Eckersley, T.L.: Discussion of the ionosphere, Proc. Roy. Soc. A141, 708–715, 1933. [24] Ellis, G.R.: The Z propagation hole in the ionosphere, J. Atmosph. Terr. Phys. 8, 43–54, 1956. [25] Franklin, C.A. and M.A. Maclean: The design of swept-frequency topside sounders, Proc. IEEE 57, 897–929, 1969. [26] Fung, S.F., R.F. Benson, D.L. Carpenter, J.L. Green, V. Jayanti, I.A. Galkan, and B.W. Reinisch: Guided Echoes in the Magnetosphere: Observations by Radio Plasma Imager on IMAGE, Geophys. Res. Lett. 3, 1589, doi:10.1029/2002GL016531, 2003. [27] Goertz, C.K. and R.J. Strangeway: Plasma waves, in: Introduction to Space Physics, edited by M.G. Kivelson, and C.T. Russell, pp. 356–399, Cambridge University Press, New York, 1995. [28] Gurnett, D.A., S.D. Shawhan, and R.R. Shaw: Auroral hiss, Z mode radiation, and auroral kilometric radiation in the polar magnetosphere: DE 1 observations, J. Geophys. Res. 88, 329–340, 1983. [29] Horita, R.E. and H.G. James: Enhanced Z-mode radiation from a dipole, Adv. Space Res. 29, 1375–1378, 2002. [30] Horita, R.E. and H.G. James: Two-point studies of fast Z-mode waves with dipoles in the ionosphere, Radio Sci., 39, RS4001, doi:10.1029/2003RS002994, 2004. [31] Huang, X., B.W. Reinisch, P. Song, P. Nsumei, J.L. Green, and D.L. Gallagher: Developing an empirical density model of the plasmasphere using IMAGE/RPI observations, Adv. Space Res. 33, 829–832, 2004. [32] Jackson, J.E.: The reduction of topside ionograms to electron-density proﬁles, Proc. IEEE 57, 960–976, 1969. [33] Jackson, J.E. and E.S. Warren: Objectives, history, and principal achievments of the topside sounder and ISIS programs, Proc. IEEE 57, 861–865, 1969. [34] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites, 1, General results, Radio Sci. 13, 531–542, 1978. [35] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites 3. Z mode pulses, J. Geophys. Res. 84, 499–506, 1979. [36] James, H.G.: Guided Z mode propagation observed in the OEDIPUS A tethered rocket experiment, J. Geophys. Res. 96, 17,865–17,878, 1991. [37] James, H.G.: Slow Z-mode radiation from sounder-accelerated electrons, J. Atmos. Solar-Terr. Phys. 66, 1755–1765, 2004. [38] James, H.G., Radiation from sounder-accelerated electrons, Adv. Space Res., in press, 2005. [39] Kelso, J.M.: Radio ray propagation in the ionosphere, 408 pp., McGraw Hill, New York, 1964. [40] Kuehl, H.H.: Electromagnetic radiation from an electric dipole in a cold anisotropic plasma, Phys. Fluids 5, 1095–1103, 1962.

34

R.F. Benson et al.

[41] Kuo, S.P., M.C. Lee, and P. Kossey: Excitation of short-scale ﬁeld-aligned electron density irregularities by ionospheric topside sounders, J. Geophys. Res. 104, 19,889–19,894, 1999. [42] Kurth, W.S., G.B. Hospodarsky, D.A. Gurnett, M.L. Kaiser, J.-E. Wahlund, A. Roux, P. Canu, P. Zarka, and Y. Tokarev: An overview of observations by the Cassini radio and plasma wave investigation at Earth, J. Geophys. Res. 106, 30239–30252, 2001. [43] LaBelle, J. and R.A. Treumann: Auroral radio emissions, 1. hisses, roars, and bursts, Space Sci. Rev. 101, 295–440, 2002. [44] Lewis, R.M. and J.B. Keller: Conductivity tensor and dispersion equation for a plasma, Phys. Fluids, 5, 1248–1264, 1962. [45] Lockwood, G.E.K.: A ray-tracing investigation of ionospheric Z-mode propagation, Can. J. Phys. 40, 1840–1843, 1962. [46] McAfee, J.R.: Ray trajectories in an anisotropic plasma near plasma resonance, J. Geophys. Res. 73, 5577–5583, 1968. [47] McAfee, J.R.: Topside ray trajectories near the upper hybrid resonance, J. Geophys. Res. 74, 6403–6408, 1969. [48] Muldrew, D.B.: Radio propagation along magnetic ﬁeld-aligned sheets of ionization observed by the Alouette topside sounder, J. Geophys. Res. 68, 5355–5370, 1963. [49] Muldrew, D.B.: Nonvertical propagation and delayed-echo generation observed by the topside sounders, Proc. IEEE 57, 1097–1107, 1969. [50] Muldrew, D.B.: Electron resonances observed with topside sounders, Radio Sci. 7, 779–789, 1972. [51] Muldrew, D.B. and M.F. Estabrooks: Computation of dispersion curves for a hot magnetoplasma with application to the upper-hybrid and cyclotron frequencies, Radio Sci. 7, 579–586, 1972. [52] Osherovich, V.A., J. Fainberg, R.F. Benson, and R.G. Stone: Theoretical analysis of resonance conditions in magnetized plasmas when the plasma/gyro frequency ratio is close to an integer, J. Atm. and Solar-Terr. Phys. 59, 2361–2366, 1997. [53] Oya, H.: Conversion of electrostatic plasma waves into electromagnetic waves: numerical calculation of the dispersion relation for all wavelengths, Radio Sci. 6, 1131–1141, 1971. [54] Poeverlein, H.: Strahlwege von Radiowellen in der Ionosph¨ are, Z. Angew. Phys. 1, 517, 1949. [55] Pulinets, S.A.: Prospects of topside sounding, in WITS handbook N2, edited by C.H. Liu, pp. 99–127, SCOSTEP Publishing, Urbana, Illinois, 1989. [56] Pulinets, S.A., P.F. Denisenko, N.A. Zabotin, and T.A. Klimanova: New method for small-scale irregularities diagnostics from topside sounder data, in: SUNDIAL Workshop, McLean, Virginia, 1989. [57] Ratcliﬀe, J.A.: The Magneto-Ionic Theory and its Applications to the Ionosphere, 206 pp., Cambridge University Press, New York, 1959. [58] Reinisch, B.W.: Modern Ionosondes, in: Modern Ionospheric Science, edited by H. Kohl, R. Ruster, and K. Schlegel, pp. 440–458, European Geophysical Society, Katlenburg-Lindau, Germany, 1996. [59] Reinisch, B.W., D.M. Haines, K. Bibl, G. Cheney, I.A. Gulkin, X. Huang, S.H. Myers, G.S. Sales, R.F. Benson, S.F. Fung, J.L. Green, W.W.L. Taylor, J.-L. Bougeret, R. Manning, N. Meyer-Vernet, M. Moncuquet, D.L. Carpenter, D.L.

1 Active Wave Experiments in Space Plasmas: The Z Mode

[60]

[61]

[62]

[63] [64] [65] [66]

[67]

[68]

[69]

35

Gallagher, and P. Reiﬀ: The radio plasma imager investigation on the IMAGE spacecraft, Space Sci. Rev. 91, 319–359, 2000. Reinisch, B.W., X. Huang, D.M. Haines, I.A. Galkin, J.L. Green, R.F. Benson, S.F. Fung, W.W.L. Taylor, P.H. Reiﬀ, D.L. Gallagher, J.-L. Bougeret, R. Manning, D.L. Carpenter, and S.A. Boardsen: First results from the radio plasma imager on IMAGE, Geophys. Res. Lett. 28, 1167–1170, 2001a. Reinisch, B.W., X. Huang, P. Song, G.S. Sales, S.F. Fung, J.L. Green, D.L. Gallagher, and V.M. Vasyliunas: Plasma density distribution along the magnetospheric ﬁeld: RPI observations from IMAGE, Geophys. Res. Lett. 28, 4521– 4524, 2001b. Sonwalkar, V.S., D.L. Carpenter, T.F. Bell, M.A. Spasojevic, U.S. Inan, J. Li, X. Chen, A. Venkatasubramanian, J. Harikumar, R.F. Benson, W.W.L. Taylor, and B.W. Reinisch: Diagnostics of magnetospheric electron density and irregularities at altitudes 5. There were almost no Geotail kilometric continuum radiation observations when Kp > 5 in 1996. Figure 2.9 provides dramatic evidence for the enhancement of KC at solar maximum (right panel) over solar minimum (left panel). The cause of this solar cycle diﬀerence is unknown. Both studies show that kilometric continuum radiation is observed even if Kp = 0. In order to check the Kp dependence, an example when kilometric continuum radiation was observed and Kp = 0 is shown in Fig. 2.10. The kilometric continuum is observed from 06–18 UT on Geotail on December 20, 1996, as shown in the top dynamic spectra. The magnetic latitudes of the satellite are shown for one day in the center of the upper right panel next to the spectrogram and the circles indicate the observations of kilometric continuum. Kp indices are shown above the panel for the day in the second line and for the previous day in the ﬁrst line. Kp had been less or equal to 1 for more than 24 hours. This demonstrates that kilometric continuum occurs even in such a very quiet time. Dst had been almost zero as seen from the lower right panel. These observations pose a problem for the energy source of the emissions.

48

K. Hashimoto et al.

Fig. 2.10. Geotail observations for December 20, 1996, when Kp was always less than 1. Kilometric continuum radiation is observed from 06 UT to 18 UT

Fig. 2.11. Geotail observations April 8, 2001, when Kp ranged from 2 to 7. Kilometric continuum radiation was not observed until after 22 UT even though Geotail was within 10 degrees of the equator since 09 UT

Figure 2.11 is an example of occurrence of large Kp without kilometric continuum. The continuum is observed only after 22 UT although the satellite was within 10 degrees of the equator since 09 UT, Kp was large for a long time, and AKR was often observed. It should be noted that the kilometric continuum radiation from about 600 kHz to about 800 kHz was quite strong from 22 UT to 05 UT the following day.

2 Review of Kilometric Continuum

49

2.6 Simultaneous Wave Observations by Geotail and IMAGE Data from the extreme ultraviolet (EUV) imager [47] of the IMAGE satellite demonstrated the existence of notches, where the electron densities are low near the equatorial plasmapause, and the notches are the source of kilometric continuum radiation. The radio plasma imager (RPI) of the satellite indicated that the electromagnetic waves were not only trapped in the low density region but also escape radially out or are generated outside as kilometric continuum. The IMAGE satellite observed the notch near 4 UT in the dayside as shown in Fig. 2.12 [10]. The orbit is the black curve and the plasmapause is shown as the light curve. Near 190◦ in the magnetic longitude, the density is decreased and the plasmapause position was inside the normal one. Kilometric continuum is observed inside the notch region.

Fig. 2.12. Notch observation by IMAGE on April 8, 2001 [10] (Reprinted with permission of the American Geophysical Union)

On the same day, Geotail did not observe kilometric continuum under very high Kp as shown in Fig. 2.11 until 18 hours later. The notch was observed around local noon, but Geotail was in the nightside. This fact indicates that kilometric continuum is radiated during high Kp, but Geotail was not able to observe it. The satellite position could be the cause of the low occurrence probability at high Kp. Simultaneous observations of kilometric continuum with IMAGE RPI and Geotail PWI are displayed in a frequency range of 300 − 800 kHz in the top and the bottom of Fig. 2.13, respectively [Hashimoto et al., 20]. Intense kilometric continuum was received during the disturbed time, especially for Kp > 7 from 20 UT to 03 UT. It should be noted that both kilometric continuum spectra show quite good similarity including the ﬁne structures from 21 UT to 06 UT. IMAGE moved from the southern hemisphere to 30◦ N. On the other hand, Geotail moved in the equatorial region from 4.4◦ N to 12.3◦ N at 01 UT and

50

K. Hashimoto et al.

Fig. 2.13. IMAGE and Geotail observations and their orbits on May 29–30, 2003. Note the similarity of the spectra [20] (Reprinted with permission of the American Geophysical Union)

then back down to 2.4◦ N as shown in the right hand panel of Fig. 2.13. Both satellites observed almost the same spectra in a wide latitude range of more than 30◦ . Their longitudes are close within 10◦ . IMAGE RPI observed the emission in a wide latitudinal range diﬀerent from general trends reported by Hashimoto et al. [19] and Green et al. [10]. The vertical line at 0420 UT is a type III burst. The intensity observed by IMAGE is weaker around 400 kHz after 0430 UT when the satellite is at latitudes higher than 25◦ . It would be diﬃcult to explain these quite similar spectra by multiple narrow beam sources. Rather, this can be explained if the sources radiate uniformly in wide emission cones in latitude and both satellites receive the emissions from the same sources, which is contrary to the beaming theory.

2.7 Summary and Conclusions Non-thermal continuum radiation is one of the fundamental electromagnetic emissions in planetary magnetospheres [cf. the review by Kaiser, 27]. It has been observed in every planetary magnetosphere visited by spacecraft armed with wave instruments and has even been found to be generated in the magnetosphere of the Galilean moon Ganymede [Kurth et al., 33]. Although

2 Review of Kilometric Continuum

51

this emission has been observed and studied for more than 35 years, there are still several unveriﬁed theories on how this emission is generated. There is also much more which we do not know about this emission and its relationship to the dynamics of the plasmasphere. Many of the characteristics of the lower frequency portion of the nonthermal continuum (trapped component) have been diﬃcult to determine due to the multiple reﬂections of the emission from the magnetopause and plasmapause. Recently there is renewed interest in studying the high frequency extension of this emission (the escaping component), especially the extension into the kilometric frequency range. Kilometric continuum has been reported to be observed by Polar and Cluster [Menietti et al., 39, 40] and INTERBALL-1 [Kuril’chik et al., 34, 35] in addition to Geotail, IMAGE, and CRRES [10, 11, 19, 20, and the present paper]. It has been conﬁrmed that the kilometric continuum is generated at steep density gradients at density irregularities in the equatorial region. These irregularities do not only exist at the plasmapause, but also inside the plasmapause and in notches. Although the observations are consistent with the mode conversion mechanism at the plasma frequency, they are not consistent with the beaming model of Jones [22, 23]. The simultaneous observations given in Fig. 2.13 provide striking evidence against the latter although this is discussed in more detail by Hashimoto et al. [20]. The relations to solar and geomagnetic activities are also interesting topics. Several new features of the high frequency escaping kilometric continuum, such as the narrow latitudinal beam structure and relationship to plasmaspheric notch or notch structures, provide a new opportunity to observe the triggering of this emission and its relationship to plasmaspheric dynamics. The insight gained in performing multi-spacecraft correlative measurements should provide key measurements on separating spatial from temporal eﬀects that are essential to verifying existing theories. Observing the radiation while the instability has been initiated and grows, and examining the dynamics of the large-scale plasmasphere should lead to signiﬁcant advances in delineating the best theory for the generation of this emission.

References [1] Anderson, R.R., D.A. Gurnett, and D.L. Odem: CRRES plasma wave experiment, J. Spacecraft Rockets 29, 570, 1992. [2] Barbosa, D.D.: Low-level VLF and LR radio emissions observed at earth and Jupiter, Rev. Geophys. Space Phys. 20, 316, 1982. [3] Carpenter, D.L. and R.R. Anderson: An ISEE/whistler model of equatorial electron density in the magnetosphere, J. Geophys. Res. 97, 1097, 1992. [4] Carpenter, D.L., R.R. Anderson, W. Calvert, and M.B. Moldwin: CRRES observations of density cavities inside the plasmasphere, J. Geophys. Res. 105, 23323, 2000.

52

K. Hashimoto et al.

[5] D´ecr´eau, P.M.E. et al.: Observation of continuum radiations from the Cluster ﬂeet: First results from direction ﬁnding, Ann. Geophys. 22, 2607, 2004. [6] Filbert, P.C. and P.J. Kellogg: Observations of low-frequency radio emissions in the earth’s magnetosphere, J. Geophys. Res. 94, 8867, 1989. [7] Frankel, M.S.: LF radio noise form the Earth’s magnetosphere, Radio Sci. 8, 991, 1973. [8] Fung, S.F. and K. Papadopoulos: The emission of narrow-band Jovian kilometric radiation, J. Geophys. Res. 92, 8579, 1987. [9] Green, J.L. and S.A. Boardsen: Conﬁnement of nonthermal continuum radiation to low latitudes, J. Geophys. Res. 104, 10307, 1999. [10] Green, J.L., B.R. Sandel, S.F. Fung, D.L. Gallagher, and B.W. Reinisch: On the origin of kilometric continuum, J. Geophys. Res. 107, doi: 10.1029/ 2001JA000193, 2002. [11] Green, J.L. et al.: Association of kilometric continuum radiation with plasmaspheric structures, J. Geophys. Res. 109, A03203, doi: 10.1029/2003JA010093, 2004. [12] Green, J.L. and S.F. Fung: Advances in Inner Magnetospheric Passive and Active Wave Research, In: AGU Monograph on Physics and Modeling of the Inner Magnetosphere, AGU, Washington D.C., in press, 2005. [13] Gough, M.P.: Nonthermal continuum emissions associated with electron injections: Remote plasmapause sounding, Planet. Space Sci. 30, 657, 1982. [14] Gurnett, D.A. and R.R. Shaw: Electromagnetic radiation trapped in the magnetosphere above the plasma frequency, J. Geophys. Res. 78, 8136, 1973. [15] Gurnett, D.A.: The Earth as a radio source: The nonthermal continuum, J. Geophys. Res. 80, 2751, 1975. [16] Gurnett, D.A. and L.A. Frank: Continuum radiation associated with low-energy electrons in the outer radiation zone, J. Geophys. Res. 81, 3875, 1976. [17] Gurnett, D.A., W. Calvert, R.L. Huﬀ, D. Jones, and M. Sugiura: The polarization of escaping terrestrial continuum radiation, J. Geophys. Res. 93, 12817, 1988. [18] Hashimoto, K., K. Yamaashi, and I. Kimura: Three-dimensional ray tracing of electrostatic cyclotron harmonic waves and Z mode electromagnetic waves in the magnetosphere, Radio Science 22, 579, 1987. [19] Hashimoto, K., W. Calvert, and H. Matsumoto: Kilometric continuum detected by Geotail, J. Geophys. Res. 104, 28645, 1999. [20] Hashimoto, K., R.R. Anderson, J.L. Green, and H. Matsumoto: Source and Propagation Characteristics of kilometric continuum observed with multiple satellite, J. Geophys. Res., 110, A09229, doi: 10.1029/2004JA010729, 2005. [21] Jones, D.: Source of terrestrial nonthermal radiation, Nature 260, 686, 1976. [22] Jones, D.: Latitudinal beaming of planetary radio emissions, Nature 288, 225, 1980. [23] Jones, D.: Beaming of terrestrial myriametric radiation, Adv. Space Res. 1, 373, 1981. [24] Jones, D.: Terrestrial myriametric radiation from the Earth’s plasmapause, Planet. Space Sci. 30, 399, 1982. [25] Jones, D., W. Calvert, D. A. Gurnett, and R. L. Huﬀ: Observed beaming of terrestrial myriametric radiation, Nature 328, 391, 1987.

2 Review of Kilometric Continuum

53

[26] Jones, D.: Planetary radio emissions from low magnetic latitudes – Observations and theories, In: Planetary Radio Emissions II (H.O. Rucker, S.J. Bauer, and B.M. Pedersen. Eds., Austrian Acad. Sci., Vienna) p. 245, 1988. [27] Kaiser, M.L.: Observations of non-thermal radiation from planets, In: Plasma Waves and Instabilities at Comets and in Magnetospheres, (B. Tsurutani and H. Oya, Eds., AGU, Washington) p. 221, 1989. [28] Kasaba, Y., H. Matsumoto, K. Hashimoto, R.R. Anderson, J.-L. Bougeret, M.L. Kaiser, X.Y. Wu, and I. Nagano: Remote sensing of the plasmapause during substorms: GEOTAIL observation of nonthermal continuum enhancement, J. Geophys. Res. 107, 20389, 1998. [29] Kurth, W.S., J.D. Craven, L.A. Frank, and D.A. Gurnett: Intense electrostatic waves near the upper hybrid resonance frequency, J. Geophys. Res. 84, 4145, 1979. [30] Kurth, W.S., D.A. Gurnett, and R.R. Anderson: Escaping nonthermal continuum radiation, J. Geophys. Res. 86, 5519, 1981. [31] Kurth, W.S.: Detailed observations of the source of terrestrial narrowband electromagnetic radiation, Geophys. Res. Lett. 9, 1341, 1982. [32] Kurth, W.S.: Continuum radiation in planetary magnetospheres, In: Planetary Radio Emissions III (H. O. Rucker, S. J. Bauer, and M. L. Kaiser, Eds., Austrian Acad. Sci., Vienna) p. 329, 1992. [33] Kurth, W.S., D.A. Gurnett, A. Roux, and S.J. Bolton: Ganymede: A new radio source, Geophys. Res. Lett. 24, 2167, 1997. [34] Kuril’chik, V.N., M.Y. Boudjada, and H.O. Rucker: Interball-1 observations of the plasmaspheric emissions related to terrestrial “continuum” radio emissions, In: Planetary Radio Emissions V (H.O. Rucker, M.L. Kaiser and Y. Leblanc, Eds., Austrian Acad. Sci., Vienna) p. 325, 2001. [35] Kuril’chik, V.N., I.F. Kopaeva, and S.V. Mironov: INTERBALL-1 observations of the kilometric “continuum” of the Earth’s magnetosphere, Cosmic Research 42, 1, 2004. [36] Lee, L.C.: Theories of non-thermal radiations from planets, In: Plasma Waves and Instabilities at Comets and in Magnetospheres (B. Tsurutani and H. Oya, Eds., AGU, Washington) p. 239, 1989. [37] Matsumoto, H., I. Nagano, R.R. Anderson, H. Kojima, K. Hashimoto, M. Tsutsui, T. Okada, I. Kimura, Y. Omura, and M. Okada: Plasma wave observations with GEOTAIL spacecraft, J. Geomagn. Geoelectr. 46, 59, 1994. [38] Melrose, D.B.: A theory for the nonthermal radio continuum in the terrestrial and Jovian magnetospheres, J. Geophys. Res. 86, 30, 1981. [39] Menietti, J.D., R.R. Anderson, J.S. Pickett, D.A. Gurnett, and H. Matsumoto: Near-source and remote observations of kilometric continuum radiation from multi-spacecraft observations, J. Geophys. Res. 108, 1393, doi: 10.1029/2003JA009826, 2003. [40] Menietti, J.D., O. Santolik, J.S. Pickett, and D. A. Gurnett: High resolution observations of continuum radiation, Planet. Space Sci. 53, 283, 2005. [41] Morgan, D.D. and D.A. Gurnett: The source location and beaming of terrestrial continuum radiation, J. Geophys. Res. 96, 9595, 1991. [42] Okuda, H., M. Ashour-Abdalla, M.S. Chance, and W.S. Kurth: Generation of nonthermal continuum radiation in the magnetosphere, J. Geophys. Res. 87, 10457, 1982.

54

K. Hashimoto et al.

[43] Oya, H., M. Iizima, and A. Morioka: Plasma turbulence disc circulating the equatorial region of the plasmasphere identiﬁed by the plasma wave detector (PWS) onboard the Akebono (EXOS-D) satellite, Geophys. Res. Lett. 18, 329, 1991.) [44] Reinisch, B.W. et al.: The Radio Plasma Imager investigation on the IMAGE spacecraft, Space Sci. Rev. 91, 319, 2000. [45] R¨ onnmark, K.: Emission of myriametric radiation by coalescence of upper hybrid waves with low frequency waves, Ann. Geophys. 1, 187, 1983. [46] R¨ onnmark, K.: Conversion of upper hybrid waves into magnetospheric radiation, In: Planetary Radio Emissions III (H.O. Rucker, S.J. Bauer, and M.L. Kaiser, Eds., Austrian Acad. Sci., Vienna) p. 405, 1992. [47] Sandel, B.R., R.A. King, W.A. King, W.T. Forrester, D.L. Gallagher, A.L. Broadfoot, and C.C. Curtis: Initial results from the IMAGE Extreme Ultraviolet Imager, Geophys. Res. Letts. 28, 1439, 2001. [48] Vesecky, J.F. and M.S. Frankel: Observations of a low-frequency cutoﬀ in magnetospheric radio noise received on IMP 6, J. Geophys. Res. 80, 2771, 1975. [49] Yamaashi, K., K. Hashimoto, and I. Kimura: 3-D electrostatic and electromagnetic ray tracing in the magnetosphere, Mem. Natl. Inst. Polar Res., Spec. Issue 47, 192, 1987.

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions P. Louarn Centre d’Etudes Spatiales du Rayonnement, Toulouse, France [email protected]

Abstract. Owing to the complete and precise wave/particle measurements performed by the Viking and FAST spacecraft in the sources of the terrestrial Auroral Kilometric Radiation, the conditions of generation of this coherent radio emission are now well documented. It has been demonstrated that the production of radio waves occurs in thin laminar regions that exactly correspond to the regions of auroral particle acceleration. We present and discuss observations, mainly performed by Viking, that have led to the conception of a model of wave generation by ﬁnite geometry sources. The corresponding theoretical analysis is also presented. This model has certainly a very broad domain of applications, including radio emissions from other planets, from the Sun and more distant astrophysical objects.

Key words: Auroral Kilometric Radiation, free space radiation, magnetoionic modes, eﬀect of inhomogeneity, cyclotron maser mechanism, loss cone distribution, horseshoe, ring distribution

3.1 Introduction As seen from deep space, the Earth is a powerful natural radio source emitting 107 to 108 W with a maximum spectral density at ∼250 kHz. Its dominant emission – the Auroral Kilometric Radiation (AKR hereafter) – is generated in the night sector of the auroral zone, at magnetic latitudes larger than 65◦ and altitudes ranging from 5000 km to 15000 km. Its power increases with magnetospheric activity, especially when substorms develop. This has been realized ﬁrst by Benediktov et al. [6] and conﬁrmed by others [9, 20, 21, 23]. This radiation is characterized by a brightness temperature (>1015 K) much larger than any plasma temperature in planetary magnetospheres. It cannot be explained by a thermal generation mechanism and coherent emission processes such as plasma instabilities must be invoked for its generation. The AKR is not a singular phenomenon. In its non-thermal origin, high polarization, temporal variability and spectral ﬁne structure AKR is similar P. Louarn: Generation of Auroral Kilometric Radiation in Bounded Source Regions, Lect. Notes Phys. 687, 55–86 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

56

P. Louarn

to the radio emissions that emanate from other magnetospheres, at Jupiter, Saturn, Uranus and Neptune [see, e.g., Zarka, 60]. It can be compared to solar radio bursts (microwave spike bursts) [1, 15] and stellar radio emissions [32, 39]. The AKR generation mechanism has thus to be considered as a general process able to eﬃciently convert some forms of free energy present in magnetized plasmas into radiating electromagnetic waves. Due to the proximity of its sources, the study of the AKR oﬀers a splendid opportunity to test all the details of this radiation process. It was actually possible to determine some important characteristics of AKR from early space measurements [6, 21, 22]. The evidence for a dominant X polarization came from Voyager measurements [Kaiser et al., 26], as the probes began their odyssey. This was conﬁrmed later from DE-1 measurements published by Shawhan and Gurnett [55] and Mellott et al. [38], although the existence of a fainter O mode was also detected. The good correlation between inverted V precipitation and AKR [Green et al., 20] has given support to the idea that the AKR is generated by electrons accelerated above the auroral regions. Important progress has been made by the crossing of the sources regions themselves, ﬁrst reported by Benson and Calvert [7] using ISIS 1 spacecraft. These direct in-situ measurements have revealed that the AKR is generated: • in strongly magnetized regions (fp /fc 1 where fp is the plasma frequency and fc the electron gyrofrequency), • at frequencies close to the local cut-oﬀ of the X mode: fx = 12 [fc + (fc2 + 4fp2 )1/2 ], • with waves propagating at large angles with respect to the geomagnetic ﬁeld B0 [7, 8]. Numerous theories have been proposed for the interpretation of this coherent radiation: nonlinear mode coupling, soliton radiation and linear instabilities. In view of the observational constraints summarized above, a linear instability – the cyclotron maser – ﬁrst proposed by Wu and Lee [59] and Lee and Wu [27] and further developed by several authors [28, 29, 30, 40, 41, 42], has been acknowledged by a majority as being the most promising generation mechanism. Though this relativistic instability was known before [see Trubnikov, 57], the main progress realized by Wu and Lee [59] was to show that the relativistic correction in the electron motion is crucial even at the modest keV energies of auroral electrons. In the tenuous auroral plasmas, the X mode cut-oﬀ frequency is indeed just a few percent above fc and the relativistic correction must be considered in the wave/particle resonant condition: ω − k v − ωc /Γ = 0 where Γ = (1 − 1 v 2 /c2 )− 2 is the Lorentz factor and k the wave vector component parallel to the static magnetic ﬁeld. The relativistic eﬀects permit an exchange of energy between X mode waves and electrons. It can be shown that the plasma may act as a coherent radiating source if the electron distribution presents an inversion of population in the form of positive gradients ∂f /∂v⊥ > 0. Since the most common distribution functions observed in the auroral zone – the loss-cone

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

57

distributions – precisely present this type of free energy, an eﬃcient X mode ampliﬁcation may take place at frequencies close to fc . This good agreement between theoretical predictions and observations has explained the success of this direct ampliﬁcation mechanism in the community. The cyclotron maser model was thus soon recognized as the most plausible generation mechanism of non-thermal radio emission radiated by highly magnetized planetary, solar and even stellar plasmas. Our understanding of the AKR generation was greatly improved by the measurements made by the Swedish satellite Viking in the heart of the AKR sources. It was possible to understand the connection between the particle acceleration and the production of the radiations and to get a precise knowledge of the plasma conditions inside the sources and in their near vicinity. As it is reviewed here, these observations have motivated new theoretical analyses that were severely tested. These studies have contributed to establish a sophisticated model of the AKR generation and, by extension, of the planetary radio emissions. Ten years after Viking, the superb FAST measurements have conﬁrmed and updated this picture of wave ampliﬁcation in thin laminar sources. In Sect. 3.2, we present some of the Viking observations performed in the AKR sources. The observational eﬀects linked to the ﬁnite geometry of the sources are discussed in Sect. 3.3. The theoretical analysis in made in Sect. 3.4 before the discussion and the conclusion proposed in Sect. 3.5.

3.2 Spacecraft Observations in the Sources of the Auroral Kilometric Radiation 3.2.1 Structure of Sources and Wave Properties The Viking spacecraft was the ﬁrst to explore, with a complete set of experiments, the central part of the AKR sources, at altitudes of the order of one Earth’s radius (RE ) [see the review by Roux et al., 53]. Frequency-time spectrograms corresponding to AKR source crossings by Viking have been presented by Bahnsen et al. [2, 3], Pottelette et al. [44], Louarn et al. [33], and Ungstrup et al. [58]. Figure 3.1 shows such a source crossing. The source is deﬁned in the data by an intensiﬁcation of the signal at frequencies very close to fc . For the example shown in Fig. 3.1 (orbit 165), this occurs from ∼2033:00 to ∼2033:30 UT. During this 30 s time period, the spectral density of the AKR is maximum just at the frequency channel that contains fc . Two additional sources are also crossed during short time periods, at ∼2032:30 and 2034:00 UT. For the purpose of a statistical study, Hilgers et al. [24] computed the integrated electric energy EAKR in the vicinity of the electron gyrofrequency (from fc − 5 kHz to fc + 10 kHz) for the whole set of Viking data. They deﬁned the sources as the regions along the orbit where this quantity is maximum.

58

P. Louarn

Fig. 3.1. An example of source crossed by Viking. The AKR source is indicated and is characterized by an increase of the AKR wave power at and, in some cases, even below the gyrofrequency

Almost 40 source crossings have been identiﬁed using this method. The statistical wave properties inside or near the sources were then determined with the following results: • The lowest frequency peak fpeak of the AKR is on average a few percent above fc , but it may also be up to 3% below it. • To determine the lowest frequency of the AKR, flc is deﬁned as the ﬁrst frequency below fpeak where the spectral energy has decreased by 10%. This frequency is signiﬁcantly below fpeak , by 2% in the average. A significant power is thus produced down to and often below fc . • The refractive index is close to unity everywhere but close to the cutoﬀ where values smaller than unity are measured [Bahnsen et al., 3]. • Concerning the polarization, Hilgers et al. [25] showed that the orientation of the electric ﬁelds is not the same within the source at f ∼ fc and outside the sources at f fc . Inside the sources, the modulation pattern due to the spin of the spacecraft is consistent with a wave electric ﬁeld being conﬁned within 10◦ in the plane perpendicular to B0 , as expected for X mode waves propagating close to the direction perpendicular to B0 . Outside the sources, the modulation pattern is less pronounced. The waves being transverse, this suggests that they propagate at a more oblique angle with respect to B0 . Concerning the source extension and its lifetime, the fact that wellidentiﬁed AKR patterns are seen in the dynamic spectrograms over several tens of minutes shows that the source exists for at least that long, essentially at the same location and, even on the same ﬁeld lines. In most situations, the spectrum covers a broad frequency range. Since the emission is generated at or very close to fc , it is possible to translate the frequency bandwidth into an altitude range for the source. The 100–200 kHz bandwidth of the AKR then corresponds to an extension of the order of 2500 km below the spacecraft (at a typical altitude of 6000 km). As discussed by De Feraudy et al. [13], the

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

59

observation of broad spectra also implies that propagation takes place in a ﬁlled cone or, more likely, that the source is extended in east-west direction along the auroral oval. The north-south extent is always small, often less than 100 km, and is directly measured since the satellite orbit is nearly in the meridian plane. Altogether, this suggests that the AKR sources are relatively long lived regions, at a mean altitude of 6000 km, with a dimension along the Earth magnetic ﬁeld of ∼2000 km and more. They would thus be ribbons or laminar structures, limited to a width of 20–200 km north-south and much more extending east-west. 3.2.2 AKR Sources as Regions of Particle Accelerations An example of an AKR source was discussed in detail by Louarn et al. [33]. It occurred during the orbit 849 and has lasted a suﬃciently long time to permit the complete measurements of the distribution functions. The corresponding wave and particle measurements are displayed in Fig. 3.2. Using the criteria presented in Sect. 3.2.1, the source crossing occurred in between the two vertical red lines. This time period corresponds to important modiﬁcations in the electron and ions energy spectrograms. The energy of the electrons is maximum just outside the source where it reaches 10 keV. This energy decreases to 5 keV inside the source as upward propagating ion beams are simultaneously detected with energies of ∼5 keV. As discussed in Louarn et al. [33], this can be explained by assuming that the AKR source is an acceleration region characterized by V-shaped iso-potential surfaces. These observations indeed suggest that Viking was successively below, inside, and again below a zone of nonzero parallel electric ﬁeld, this acceleration region coinciding with the AKR source (see the lower panels in Figs. 3.2 and 3.3). As discussed below, the fact that the sources are regions of parallel particle acceleration has several important consequences regarding the wave generation mechanism. 3.2.3 AKR Sources as Plasma Cavities That regions of nonzero parallel electric ﬁeld are regions where the plasma density is perturbed is not a surprise. This has, nevertheless, a special importance here. The plasma density is a crucial parameter for the cyclotron maser instability: it contributes to determine the diﬀerent regimes of the process and the properties of the most unstable waves. Diﬀerent methods can be used to obtain the density. For example, the low frequency emission seen below 50 kHz in Figs. 3.1 and 3.2 (the auroral “hiss”) are whistler waves with a upper-cutoﬀ frequency at the local plasma frequency [see the comparison with relaxation sounder measurements by Perraut et al., 43]. This cut-oﬀ is larger than 20 kHz outside the source. This corresponds to densities above 6 cm−3 . It clearly decreases in the source (see the line in Fig. 3.2). A careful analysis shows that it at that point approaches or even

60

P. Louarn

kHz 400

Source

200

100

Maximum AKR level Plasma cavity

Energy decrease

Upward ion beams

Source

U-shaped potential Spacecraft trajectory

E||

Fig. 3.2. Wave and particle measurements performed by Viking inside a source of AKR. The particle measurements are fully consistent with the crossing of a region of particle acceleration. The white line in the upper panel is the gyro frequency

goes below 10 kHz, the lower- frequency threshold of the receiver. The density of the plasma is then below 1.1 cm−3 in the central part of the source. This observation can be generalized to other source crossings. Hilgers et al. [24] have even shown that the auroral small-scale cavities with densities typically of the order of or less than 1.5 cm−3 detected by Viking coincide almost systematically with AKR sources. To get more indications on the plasma parameters, Louarn et al. [33] have estimated the electron density within and in the vicinity of the source from the electron detector. While the density of high-energy (1–40 keV) electrons remains almost constant across the source, the density of low-energy electrons (below 1 keV) decreases by a factor larger than 10. The density variation is thus associated with a large decrease of the density of cold plasma. A

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

61

Fig. 3.3. A schematic representation of the AKR source/auroral acceleration region

simple interpretation would be that the low energy particles coming from the ionosphere have been evacuated from the source by the accelerating electric ﬁeld. 3.2.4 Free Energy for the Maser Process Let us determine the free energy that may drive the maser. At ﬁrst glance, the problem could be reduced to the search for positive ∂f /∂v⊥ slopes in the distribution functions. Nevertheless, the determination of the non-thermal feature that drives the maser is ambiguous: • several ∂f /∂v⊥ features are able to feed the instability, and • the diﬀusion associated with the waves can level out, at least partly, the “operating” ∂f /∂v⊥ slopes. Theoretical arguments are thus useful here for the identiﬁcation of the most eﬃcient form of free energy. In Fig. 3.4 examples of non-thermal auroral electron distribution functions are given, together with the properties of the waves that can be ampliﬁed via

62

P. Louarn

v

Loss Cone

T

Resonant Ellipse

v|| upward radiation f > fce , in dominant cold plasma v

Hole, Horseshoe

T

Resonant Ellipse

v|| upward radiation f > fce , in dominant cold plasma v

T

Resonant Ellipse

Trapped

v||

perpendicular radiation f < fce , in dominant energetic plasma Fig. 3.4. Forms of free energy that may drive the maser process and expected properties of the unstable waves. In gray: regions of positive ∂f /∂v⊥

the cyclotron maser instability by such free energy sources. A crucial parameter for determining the dominant free energy source is the proportion of hot electrons in the plasma. In a plasma dominated by cold electrons, X mode waves propagate above the electron gyrofrequency and the resonance condition, ω − k v − ωc /Γ = 0, can only be fulﬁlled if k is not null. Therefore, the resonant curve – an ellipse in the v , v⊥ phase space – is not centered at v = 0, v⊥ = 0 and the growth rate which is proportional to the integral of v⊥ ∂f /∂v⊥ along the resonant curve is large only if free energy exists for v = 0. This is the case of the “loss cone” and the “hole” distribution functions, as indicated in Fig. 3.4. Conversely, in an energetic plasma with no or a very tenuous cold component, X mode propagation is possible below the electron gyro-frequency. The Doppler shift term (k v ) is no longer required to

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

63

fulﬁll the resonant condition and free energy centered at v = 0 can be used. The detailed theoretical analysis even shows that, for a given energy of the particles, inversions of populations centered at v = 0 are the most eﬃcient to drive the maser. The three-dimensional plots displayed in Fig. 3.5 show the electron distribution functions just outside and inside an AKR source. The distribution function measured just outside the source is characterized by a dense thermal core and an enhanced population of energetic particles up to 10 keV. Two kinds of ∂f /∂v⊥ non-thermal features are clearly apparent: a loss cone, for electrons moving upward and a “hole” (Fig. 3.5b) for electrons moving downward. Note that the “hole” distribution functions are similar to the “horse-shoe” distributions described by Delory et al. [14], using FAST data. The source of the AKR, as identiﬁed from the criteria discussed in Sect. 3.2.1, does not coincide with these non-thermal features. This suggests that neither a loss cone nor a hole alone suﬃces to generate the AKR. Inside the potential structure (distributions D, E, F), the thermal core is highly depleted. The loss cone and the hole are still present but less pronounced than outside the source. The most prominent feature is a broad plateau in v⊥ (with sometimes a faint indication of positive ∂f /∂v⊥ slopes), which indicates that electrons tend to accumulate in this region of the phase space. This characteristic feature of the distribution functions inside the source leads Louarn et al. [33] to propose that the trapped electrons are the free energy source of the AKR. As shown by Eliasson et al. [16] and by Louarn et al. [33, 34], this trapped electron component can be produced by a time-varying or by a spatially-varying parallel electric ﬁeld. 3.2.5 FAST Observations To date, the FAST satellite, launched in 1996, was the most recent spacecraft to explore the high altitude auroral zones. It crossed the source regions of the AKR with experiments characterized by a remarkable time resolution, the electron distributions being measured in less than a few 10 ms for example [see, 12, 19]. FAST has largely conﬁrmed the results of Viking with even more precision on several crucial points [Ergun et al., 17, 18]: • the correspondence between the acceleration regions and the sources of AKR, • the fact that only accelerated electrons are present in the source, the low energy component being completely evacuated, • the presence of the trapped distribution, • the relatively small dimension of the sources, • the ﬁne temporal/spectral structuring of the radiations, in bursts of several tenths of seconds and frequency bandwidth thinner than 1 kHz. This point was not accessible to Viking observations. These observations have reinforced the status of the cyclotron maser process as the most plausible wave generation mechanism. It was nevertheless

64

P. Louarn

Fig. 3.5. Distribution functions measured outside and inside the AKR source. Different forms of free energy are indicated. The trapped population in only found in the source

advocated that horseshoe distribution functions, thus diﬀerent from the trapped distribution, would be used by the maser [see Delory et al., 14]. Horseshoe distributions result from the same processes as trapped distributions: a combination of electric ﬁeld acceleration and the motion in phase space due to the conservation of the ﬁrst adiabatic invariant. This cannot be considered as a fundamental diﬀerence between the Viking and the FAST observations. 3.2.6 Summary To conclude, an important result from Viking was to reveal that the AKR sources correspond to acceleration regions. They are laminar structures with

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

65

a small north-south extension (a few 10 km, typically). The plasma in these regions is tenuous (n ∼ 1 cm−3 ) and essentially constituted by energetic particles ( E = 3 − 10 keV). The sources are separated from the denser and colder external plasma (n ∼ 5−10 cm−3 , E ∼ a few 10 eV) by sharp density gradients with typical scale lengths smaller than 1 km. An important feature of the electron distribution functions observed inside the sources is an electron accumulation at low parallel velocities and keV energies. These electrons are trapped between their magnetic mirror point and an electrostatic reﬂection point, leading to distributions presenting positive ∂f /∂v⊥ slopes at low v and large v⊥ . The theory shows that this constitutes a very eﬃcient source of free energy for the maser. In the tenuous plasma that ﬁlls the sources and even at moderate energy ( E < 1 keV), it can be shown that relativistic eﬀects must be taken into account in the full X mode propagation and not only in the perturbation eﬀect that leads to wave ampliﬁcation [47, 48, 49, 56]. The relativistic dispersion is then the “zero order” theory for any realistic model of the generation of the AKR. The cyclotron maser instability is just the relativistic negative-absorption eﬀect linked to the presence of positive ∂f /∂v⊥ slopes in the electron distributions. However, the limited extension of the sources modiﬁes the properties of the generation process, with observable consequences regarding the radiated waves. An ensemble of papers [35, 36, 50, 51, 52] have considered this speciﬁc problem of AKR generation in ﬁnite geometry sources. The observation of ﬁnite geometry eﬀects, the model of laminar sources, and its mathematical analysis are discussed in the next sections.

3.3 Cyclotron Maser in Finite Geometry Sources The ﬁrst detailed mentioning of ﬁnite geometry eﬀects on AKR generation was due to Calvert [10, 11]. He proposed that the ﬁnest structures in the AKR observed by ISIS 1 could be explained by a feedback lasing eﬀect due to wave reﬂection in the sources. However, this model is considerably diﬀerent from the one discussed here. Before Viking, the AKR source regions were indeed assimilated to the auroral “cavity”, a generic term that designates a wide region of low plasma density corresponding, as a whole, to the region of auroral precipitations. The lasing eﬀect studied by Calvert was supposed to take place due to reﬂections on the edges of this large scale region. The sources explored by Viking are actually much smaller and are embedded in this large scale auroral cavity. 3.3.1 A Simple Model of AKR Sources We have chosen the simplest model of the sources (see Fig. 3.6). They are supposed to be laminar structures, limited in one direction perpendicular to the static B ﬁeld (x direction, the source width being 2l) and unlimited in the y-z directions (z being the direction of B0 ). Inside the source, the electron

66

P. Louarn

Fig. 3.6. A simple slab model for the sources of AKR

population is constituted by energetic particles with an idealized ring-like distribution: f (v , v⊥ ) = (2πv0 )−1 δ(v )(v⊥ − v0 ). The external plasma is cold and the source boundaries are sharp density gradients. The idealized distribution neglects the thermal eﬀects and could appear oversimpliﬁed. However, as shown by Pritchett [47], Strangeway [56], and Le Qu´eau and Louarn [31], it leads to dispersion equations that take into account the relativistic eﬀects without mathematical complications. The use of more realistic distributions does not signiﬁcantly improve the model. For a ﬁrst approach of ﬁnite geometry eﬀects, this simple description can be considered as precise enough. Only four parameters must be taken into account: • • • •

the the the the

density inside the source: ni (plasma frequency: ωpi ), energy of the internal electron population: E = 12 mv02 , density of the external plasma: n0 (plasma frequency: ωp0 ) and, width (l) of the source.

The ionic component is supposed to be a motionless neutralizing background. We will use the following normalized parameters: i = (ωpi /ωc )2 , o = (ωpo /ωc )2 , δ = (v0 /c)2 and L = ωc l/c. For a typical source, ni ∼ 1 cm−3 , no ∼ 10 cm−3 , energy of the electron ∼5 keV, fc ∼ 200 kHz and l ∼ 15 − 30 km; these parameters are: i = 0.2×10−2 , o = 2×10−2 , δ = 10−2 and L = 60. The non-homogeneity of the geomagnetic ﬁeld is another important parameter. For a dipolar ﬁeld: one has B(z) ∼ B0 (l − z/H); with H = R/3 where R is the distance to the center of the planet. 3.3.2 Generation in Plasma Cavities: A Simple Approach Some of the physical eﬀects linked to ﬁnite geometry can be discussed in a simple way by comparing the relativistic dispersion curve and the “cold” one. As it is calculated in Sect. 3.4.1, the relativistic equation that describes the

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

67

wave propagation at frequencies close to fc has two solutions [Le Qu´eau and Louarn, 31]:

2 1 − 2χ − N2 k⊥ c 2 2 (3.1) = 1 − χz and N⊥ = N⊥ = ωc 1−χ where χz = (ωp /ωc )2 , and χ = 12 [(∆ω−δN2 )]/(∆ω+δ)2 . The ﬁrst equation is the usual dispersion equation of the ordinary mode (O mode), here developed at the zero order in ∆ω = (ω−ωc )/ωc . The second equation corresponds to the extraordinary (X or Z) mode. In Fig. 3.7, the corresponding dispersion curves are displayed, both for an energetic plasma (left panel) and a cold plasma (right panel). At leading order in ∆ω and i,o (which are small quantities of the order of 10−2 throughout this study), the cut-oﬀ and the resonance frequencies assume simple expressions: (1) in cold plasma, the X mode cut-oﬀ and Z mode resonance are ∆ωX = o /2 and ∆ωZ = o /4, (2) in relativistic plasma, they become: ∆ωX = o /2 − δ and ∆ωZ = o /4 − δ.

Re ω

Re ω

Re ω

Im ω

Unstable

X-mode O-mode N Z-mode

X-mode

O-mode

N Z-mode

Relativistic plasma inside source

Cold plasma outside source

Fig. 3.7. Dispersion curves in relativistic and cold plasmas, for quasi-perpendicular propagation. Only a narrow frequency band centered at the gyrofrequency is considered here. The width of the unstable domain is δ

At ﬁrst order in δ, the O-mode dispersion is not modiﬁed by relativistic eﬀects and is stable. Conversely, due to the resonant denominator (∆ω + δ)2 , an instability develops on the respective frequency domains [∆ωX , δ] for the X mode and [−δ, ∆ωZ ], for the Z mode [Le Qu´eau and Louarn, 31]. In the complex frequency plane, the unstable frequencies lie on the circle ||∆ω|| = δ. The maximum growth rate (Im [∆ω] = δ) is obtained for Re [∆ω] = 0. Let us note the existence of an energy threshold: for δ < 3i /8 the maximum growth rate is on the Z mode, when for δ > 3i /8 it is on the X mode. Taking i = 2.5 × 10−2 (corresponding to fp = 10 kHz and fc = 200 kHz), this critical electron energy is δ = 0.094 × 10−3 corresponding to 500 eV. This is below the typical electron energy measured in the majority of sources. The most

68

P. Louarn

frequent regime is rather δ > i which is consistent with the observation of a dominant X mode inside the source. The examination of the dispersion curves suggests that the internally unstable waves are not easily connected to the external X mode, when the connections to the external Z and O modes are systematically possible. The direct internal/external X connection would imply: δ > o /2. For an external plasma frequency of ∼30 kHz, this corresponds to δ = 1.12 × 10−2 or, to energies of ∼5.5 keV. In most cases, the energy measured by Viking is smaller, with δ < o /2. Let us also note that the maximum instability takes place at Re(∆ω) = 0 thus always below the external X mode cut-oﬀ. One concludes that the most unstable internal waves are rarely directly connected to the external X mode. This leads to a few questions: • If the most unstable waves are directly connected to the external Z and O modes only, why is the level of Z and O modes as low as it is observed? • Can the electromagnetic energy created at Re(∆ω) = 0 be transmitted to the external X mode, and by what mechanism? • How important is the condition δ > o /2 in the generation of AKR? These questions are addressed by analyzing the dynamic spectra measured by Viking during the crossing of AKR sources. 3.3.3 Observations of Finite Geometry Eﬀects Selection of Source Crossings The dynamic spectra corresponding to four source crossings are presented in Fig. 3.8. They occurred during the orbits 165, 176, 237 and 1260. The electron gyrofrequency is indicated by a white line. Two frequency ranges have been selected for each orbit. The 10–60 kHz band contains the “auroral hiss”. As mentioned before, these whistler waves have a frequency cut-oﬀ at the local plasma frequency providing the plasma density. The radiation in the upper band is the AKR (typically 200–400 kHz). During small periods of time (a few 10 s at most, in between the vertical bars), the AKR intensiﬁes, and its low-frequency cut-oﬀ shifts down to fc . These are the AKR sources. They always correspond to a decrease of the power of the hiss, its cut-oﬀ being less apparent and often shifted below the band of analysis (10 kHz). This directly illustrates that the sources are plasma cavities. For each source, the conditions of mode connection can be studied at the two interfaces corresponding to the crossings of the source frontiers. The ratio fp /fc at each interface, just outside the sources, is deduced from the position of the hiss cut-oﬀ. The lower value of this ratio is observed for orbit 1260 (0.06) and the higher for orbit 176 (0.208). This corresponds to frequency-gaps between fc and the cut-oﬀ of the external X mode (fxo ) that vary from 1.2 kHz to more than 7 kHz. This is also the gap between the domain of wave ampliﬁcation and the frequency above which the connection with the external X mode

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

69

Fig. 3.8. Four examples of source crossings. The diﬀerent components of the AKR are indicated. The black line is the gyro frequency

becomes possible. The density inside the source is not precisely known. Since the hiss cut-oﬀ is below 10 kHz, it is expected to be smaller than ∼1 cm−3 . The duration of the crossing gives an upper limit for the source width. The thinnest sources would be 20–40 km wide, the widest 150–200 km. Note that oblique crossings being possible, these numbers must thus be considered as upper limits. Mode Identiﬁcations, Internal/External Connections Two components in the AKR can be distinguished in Fig. 3.8. The most intense has a bowl shaped cut-oﬀ at a few frequency channels above fc (see

70

P. Louarn

Fig. 3.9. Detailed views of the sources. The modulation due to the spin is wellapparent and is used in the mode identiﬁcation

orbits 165 and 176). The analysis of the polarization shows that it is the X mode component. Between this component and fc , a fainter O mode radiation is often detected. It can be relatively powerful as for orbit 176. In Fig. 3.9, detailed views of the same source crossings are presented. For each orbit, three panels are shown: (1) a dynamic spectra near fc , (2) a compressed “hiss” dynamic spectra and (3) the angle between the antenna and the geomagnetic ﬁeld. This angle will be used for the study of the polarization. The temporal resolution is 2.4 s and the spectral one is 2 kHz (4 kHz above

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

71

256 kHz). The source frontiers are indicated by tick marks. We also indicate the gyrofrequency and the X mode cut-oﬀ. Between fc and the X cut-oﬀ in the external plasma (fx0 ) the level of the AKR is more than 20 dB greater inside the sources than outside. This feature of the dynamic spectra shows that the electromagnetic energy is conﬁned inside the sources, at least close to fc . This is clearly apparent for obits 165 and 176. This diﬀerence between the internal and the external wave level decreases as the frequency increases. At frequencies a few percent above fc , the energy propagates outward more easily or even freely. Outside the sources, the most intense external waves are then observed above fxo . The modes of propagation can be determined from the spin modulation. Inside the sources and whatever the frequency, the maxima of the AKR Eﬁeld are observed for a perpendicular orientation of the antenna which is consistent with the extraordinary mode polarization. Outside the sources and for f > fxo , the maxima of the E-ﬁeld are also observed when the antenna is perpendicular (see the example shown in the panels of orbit 165). The internal and the external ﬁelds have thus a similar polarization which explains the easy energy escape in this frequency domain. Still outside the sources but for fc < f < fxo , the maxima are now observed for a parallel orientation of the antenna. This is the expected polarization of O mode waves. In this frequency domain, the internal/external polarization then diﬀers which could explain the attenuation observed at the crossing of the source frontier. It is interesting to note that the relative importance of the X and O components varies with the ratio fc /fp at the interfaces. For high values of this ratio, the O component is relatively intense. The O mode level is even larger than the X one for the densest case (left part of orbit 176). For low values of this ratio, the O component is comparatively less powerful and can even be undetectable (as in the case of orbit 1260). The relative proportion of O mode thus increases for sources embedded in dense plasma. The angle of propagation of the diﬀerent AKR components can also be estimated from the dynamic spectra. Given the geometry of the Viking orbit, radiation with a largely opened bowl-shaped dynamic spectra propagates at larger angles with respect to B0 than radiation with close bowl shape. One sees that the O mode essentially propagates perpendicular to the geomagnetic ﬁeld when the X mode propagates upward at oblique angles. This angle of propagation has been evaluated at the interfaces of the diﬀerent sources. It clearly decreases, i.e. the propagation is more parallel, when the ratio fp /fc increases. For large fp /fc (case of the orbit 176), very oblique angles of propagation (20◦ ) are observed [see Louarn and Le Qu´eau, 35]. Structure of the Electromagnetic Field and Radiating Diagram The ratio E /E⊥ (ratio between the electric ﬁeld measured when the antenna is respectively parallel and perpendicular to B0 ) can be used to follow the wave refraction in the sources. This ratio increases as the frequency increases. On

72

P. Louarn

average for the sources presented here, one gets E /E⊥ ∼0.1 for (f − fc )/fc = 0.01 and E /E⊥ ∼ 0.8 for (f −fc )/fc = 0.1. This increase can be related to the refraction of the waves inside the source. Since the waves are generated very close to fc , the analysis of waves with increasing frequencies is equivalent to the analysis of waves generated farther below. It is then possible to describe the evolution of the wave vector due to the upward propagation from the evolution of the spin modulation with the frequency. This nevertheless requires some theoretical considerations. The three components of the wave electric ﬁeld are related by the following relations: 1 − χ − N2 Ey χ N N⊥ (1 − χ − N 2 ) Ey , Ez = i 2) χ(1 − N⊥

Ex = −i

(3.2) (3.3)

where z is the direction of B0 , and the wave vector is in the x, z-plane. It 1 can be shown that E /E⊥ [Ez /(Ex2 + E − y 2 ) 2 ] depends more strongly on the variations on N than of the frequency. The evolution of this ratio provides thus actually an indication of the wave refraction in the source. As discussed in Louarn and Le Qu´eau [35], the increase of the ratio E /E⊥ from 0.1 to 0.8 would indicate that the wave vector rotates from a nearly perpendicular direction to an oblique one (∼45–60◦ ), over an altitude range of 100–200 km. The measurements of E /E⊥ oﬀer another indication. In the case of an isotropic radiation ( Ex ∼ Ey ), it can be veriﬁed that the theoretical E /E⊥ value hardly exceeds 0.4 even for very oblique propagation. This is signiﬁcantly smaller than the observed values. This can be solved by noting that the polarization can be very elliptic in the x-y plane. For example, for N ∼ 0.1 and ky ∼ 0, one gets Ex /Ey ∼ 0.05. If the radiating diagram is anisotropic in the x-y plane, the measurements of E may vary signiﬁcantly with the orientation of the antenna in the x-y plane. Much larger E /E⊥ could be measured by antenna almost perpendicular to the dominant x/y polarization or, equivalently, if the plane of the antenna is close to the main direction of the radiating diagram (see Fig. 3.10). Given the geometry of the orbit and the cartwheel spin mode of Viking, this means that the radiation coming from the sources studied here mainly propagates in the meridian plane (north-south direction). It could appear surprising that the four orbits examined here present this particular orientation. The bias certainly comes from the choice of long-duration source crossings. This indeed leads to the selection of tangential-to-the-source crossings, thus presenting a similar meridian orientation. This cannot be considered as a generality. Despite this particularity of these source crossings, the study shows that the waves are rather emitted tangentially to the sources which suggests that they optimize their ampliﬁcation path in the sources.

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

73

Fig. 3.10. Radiating diagram of AKR sources. The waves are preferentially emitted tangentially to the source

Summary The main conclusions of this observational study are the following: • The energy is systematically produced on the X mode. It is conﬁned inside the source from fc (the frequency of wave ampliﬁcation) to fx0 (the frequency of possible connection with the external X mode). In this frequency range, the connection is possible with the external O and Z modes. The eﬃciency of transmission to the O mode is small (attenuation by 25 dB). Nevertheless, it increases as fp /fc increases in the external plasma. Sources embedded in a particularly dense plasma can thus even emit a dominant O mode. The Z mode transmission coeﬃcient can be considered as null. • The energy generated in the source propagates upward before escaping. This corresponds to a progressive refraction of the waves that is actually measured. As discussed later, the range of upward propagation is linked to the frequency gap that separates fc and fxo . • The angle of propagation in the external plasma varies from 80◦ –70◦ to less than 20◦ . This angle decreases as the density gap at the source frontiers increases. A quite parallel propagation is observed for large fp /fc . • The radiation diagram of a source is not isotropic: the preferred direction of emission is the tangential-to-the-source direction. As shown by these observations, the ﬁnite geometry of the sources deeply inﬂuences the properties of the radiations. We will examine below how these eﬀects are explained by the theory of the cyclotron maser in ﬁnite geometry.

74

P. Louarn

3.4 The Cyclotron Maser in Finite Geometry 3.4.1 Mathematical Formulation A priori, the classical dielectric tensor – the one obtained in homogeneous plasma – is not adapted to the present analysis. Given the geometry chosen here, a Fourier transform is indeed not possible in the x direction and general solutions of the form H(x) · exp[i(ky y + k z − ωt)] must be considered. The corresponding general current perturbation is: J(x, ky , k , ω) = dx σ(x, x ky , k , ω)E(x , ky , k , ω) (3.4) The x convolution is an important mathematical complication. Nevertheless, as discussed now, the possibility of neglecting Larmor radius eﬀects greatly simpliﬁes the problem. The conductivity tensor indeed becomes purely local: σ(x, x , ky , k , ω) = σ(ky , k , ω)δ(x − x ) and the convolution is reduced to a simple multiplication. The conductivity tensor can also be directly computed starting from the Vlasov equation ∂f ∂f ∂f +v· − e (E + v × B) · =0 ∂t ∂x ∂p

(3.5)

where p is the relativistic momentum. Using cylindrical coordinates, p = (p cos θ, p sin θ, p ), the zero and ﬁrst order linear equations are: ∂f ∂f ∂f0 v· − e (v × B0 ) · =0 ∂t ∂x ∂p ∂ ∂ −iω + ik v + vx − vy ∂x ∂y ωc ∂ ∂f0 + δf = e [δE + v × δB] · Γ ∂θ ∂p

(3.6)

(3.7)

The zero order equation reads Γ v⊥ ∂f0 ∂f0 + =0. cos θ ωc ∂x ∂θ

(3.8)

If the Larmor radius (ρ) is neglected, one notes that the idealized ring-like distribution (see Sect. 3.2.1) is a solution of (3.8). Concerning the ﬁrst order (3.7), supposing that the transverse scale of the perturbation (L = 1/k for a wave) is large compared to ρ, one can neglect the terms vx ∂/∂x and vy ∂/∂y. They are indeed of the order ωc ρ/L and thus very small compared to ω and ωc . The ﬁrst-order equation then reduces to ωc ∂ ∂f0 . (3.9) −iω + ik v + δf = e [δE + v × δB] · Γ ∂θ ∂p

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

75

This is simply the classical equation obtained in an homogeneous plasma supposing k⊥ = 0. The conductivity tensor can be deduced from (3.9). Equivalently, one can make the assumption k⊥ = 0 in the general expression of this tensor that writes [see, e.g., Bekeﬁ, 5, p. 229] ωp2 σ = i2π0 ω

∞

∞ p⊥ dp⊥

0

with:

Mm

−∞

∞ dp Mm Γ m=−∞ ω − k v − mωc /Γ

(3.10)

2 mJ 2 mJm Jm mJm m v U −iv U W v ⊥ ⊥ ⊥ m b 2 b mJm Jm 2 v = iv⊥ U U (J ) −iv W J J ⊥ ⊥ m m m m b 2 mJm 2 −iv v U U J J v W J m m m m 2

where U Wm

∂f ∂f ∂f = me Γ ω + k p⊥ − p ∂p⊥ ∂p ∂p⊥

∂f ∂f ∂f mωc = me Γ ω − − p p⊥ ∂p Γ p⊥ ∂p ∂p⊥

where the Jm are the Bessel functions of order m, with argument b = Γ k⊥ v⊥ /ωc , and me is the rest mass of the electrons. For the problem under consideration here, i.e. the generation of supra-luminous mode in a moderately energetic plasma, the argument b is small. Retaining only the leading term in the development of the Bessel functions, one gets: ∞ ∞ ωp2 ωv ∂f v⊥ dv⊥ dv Mz σ = −i2π0 ω ω − k v ∂v −∞ 0

∂f v⊥ ∂f + + k v⊥ − v ω 4 ∂v⊥ ∂v M− M+ + × ω − k v − ωΓc ω − k v + ωΓc where

0 0 0 Mz = 0 0 0 001

1 −i 0 M− = i 1 0 0 0 0

1 i M+ = −i 1 0 0

(3.11)

0 0 0

Assuming that the distribution function is an idealized ring-like function: f (v , v⊥ ) = (2πv0 )−1 δ(v )δ(v⊥ − v0 ), this becomes:

76

P. Louarn

0 1 − χ1 iχ2 0 −iχ2 1 − χ1 = 0 0 1 − χz with

(3.12)

δ(N2 − 1) 1 1 + ± , 1 + ωc /ω 1 − (1 − δ)ωc /ω [1 − (1 − δ)ωc /ω]

χ1,2

ωp2 =± 2 2ω

χz

= −(ωp /ω)2

With the additional assumption that the generation takes place close to the gyrofrequency or ∆ω = (ω − ωc )/ωc 1, one obtains the simpliﬁed dielectric tensor that will be systematically used here: 0 1 − χi,o iχi,o 0 −iχi,o 1 − χi,o i,o = (3.13) 0 0 1 − χzi,o where the abbreviations are 2

χi =

i ∆ω + δN , 2 (∆ω + δ)2

χo =

o 1 , 2 ∆ω

χzo,i = o,i

and the subscripts i, o refer to the internal or the external plasma. Let us now take into account the ﬁnite geometry. In each of the three homogeneous regions that constitute the slab geometry, the Maxwell equations read: ω2 ·E =0 c2 ∇ × (∇ × H) + iω0 ∇ × ( · E) = 0

∇ × (∇ × E) −

(3.14)

It can be shown that the components of the electromagnetic ﬁeld can be deduced from the parallel components only (E and H , with H = µ0 B): ∂E ∂E i Ex = − + iχ k (N2 − 1 + χ) ∆ ∂x ∂y ∂H ∂H + (N2 − 1 + χ) +ωµ0 −iχ (3.15) ∂x ∂y ∂E ∂E i + (N2 − 1 + χ) k −iχ Ey = − ∆ ∂x ∂y ∂H ∂H + iχ −ωµ0 (N2 − 1 + χ) (3.16) ∂x ∂y and Hx = −

1 ωµ0

k Ey + i

∂E ∂y

,

Here ∆ = (ω 2 /c2 )[(1 − χ − N2 )2 − χ2 ].

Hy =

1 ωµ0

k Ex + i

∂E ∂x

(3.17)

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

77

The parallel components of the electromagnetic ﬁeld are related by the equations χ ω 2 1 − χz µ0 2 2 (1 − χ + N ) E = i ω 2 N H ∇⊥ + 2 (3.18) c 1 − chi c 1−χ

χ(1 − χ2 ω 2 1 − 2χ 0 − N2 H E (3.19) ∇2⊥ + 2 = −i ω 2 N c 1−χ c 1−χ where ∇2⊥ is the “perpendicular” Laplacean. For perpendicular propagation (N = 0), these equations respectively correspond to the dispersion of the O and the extraordinary (X and Z) mode. They can be combined to obtain a fourth-order equation for either the parallel electric or magnetic components. This equation must be solved in the diﬀerent plasma regions. Using the continuity conditions for Ey , Hy , E and H , one gets the compatibility condition that determines the set of discrete acceptable solutions. At the leading order in ∆ω, one ﬁnds: N2 ∂ χ ∂ + + 1− H = icµ0 N (3.20) E ∂x ˜ ∂ y˜ 1−χ 1−χ

∂ ω 2 1 − 2χ ∂ + + 2 − N2 H = −ic0 N χ1 − χE (3.21) ∂x ˜ ∂ y˜ c 1−χ where (˜ x, y˜) = (ωc /c)(x, y). We assume that χz = 0 which is justiﬁed by the low value of (ωp /ωc )2 inside and near the sources of AKR. With this simpliﬁcation, the fourth-order diﬀerential operator can be combined into two second-order wave operators corresponding respectively to the extraordinary and O modes: 1 − 2χ − N2 ∂ ∂ ∂ ∂ H + + + + 1 − N2 = 0 . (3.22) E ∂x ˜ ∂ y˜ 1−χ ∂x ˜ ∂ y˜ At this order of simpliﬁcation, the “O mode” operator simply becomes the “vacuum” operator. The O mode then “sees” no diﬀerence between the source and the external plasma. In this limiting case, it is stable, it freely propagates, and it is not coupled with the X mode. Let us consider the extraordinary mode: 1 − 2χ − N2 ∂ ∂ + + H = 0 . (3.23) ∂x ˜ ∂ y˜ 1−χ Let us choose symmetric solutions in each portion of the slab geometry. Inside the source ! ˜ exp i(Ny y˜ + N z˜ − ωt) . (3.24) Hi (x, y, z, t) = cos (si x Outside the source (+ on the right side, – on the left side)

78

P. Louarn

! Ho (x, y, z, t) = A exp(±iso x ˜) exp i(Ny y˜ + N z˜ − ωt) .

(3.25)

A is a normalization factor. si,o are the transverse wave numbers outside and inside the source, respectively. They satisfy the equation 1 − 2χi,o − N2

s2i,o + Ny2 =

1 − χi,o

.

(3.26)

Now, using the condition of continuity of Ey , Hy , E and H , one gets a relation between the amplitude inside and outside the sources and a compatibility condition: A (N2

= cos(si L) exp[−isi,o L] si so − 1 − χi ) sin(si L) + i(N2 − 1 + χo ) cos(si L) ∆i ∆

o χo χi = Ny − cos(si L) . ∆o ∆i

(3.27)

(3.28)

Equation (3.28) can be considered as the dispersion equation that contains the physical eﬀects due to the ﬁnite geometry of the source region. 3.4.2 Solutions of the Dispersion Relation The main results concerning the solutions of (3.28) are summed up below: • The solutions are discrete. They are localized both in the complex frequency and Re(ω)/Re(k) dispersive plane along curves presenting strong analogies with those obtained in the homogeneous case. In particular, the frequency of the X mode cut-oﬀ and of the Z mode resonance are not modiﬁed. This is illustrated in Fig. 3.11, where the solutions (for N = 0 and Ny = 0) are plotted. Two sets of discrete solutions are obtained (i) supraluminous solutions (s < 1) with the same cut-oﬀ as the “homogeneous” X Re ∆ω

Re ∆ω

Instability

0 0.5

- 0.5

Im ∆ω

X-mode

2 N

- 0.5 X-mode

Z-mode - 1.0

1

0

Z-mode

- 1.0

Fig. 3.11. Repartition of the discrete solutions in the complex frequency and dispersion plane. Normalized frequencies are used (see text)

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

•

• •

•

79

mode and (ii) infra-luminous solutions (s > 1) that accumulate near the Z mode resonance. The cut-oﬀ and the resonance do not depend on the width of the source. The wave vectors of two successive solutions diﬀer by 2π/L, the quantization rule being si (n) ∼ 2nπ/L. For s ∼ 1, the frequency gap between two successive solutions is thus of the order of 2cπ/L which corresponds to 200 Hz for a source of 30 km width. The cyclotron maser instability presents the same regimes as in the homogeneous case. They correspond to parameter ranges independent on the source width: for δ > i /2, the most unstable solutions are on the X mode near fc and N ∼ 1. For 3i /8 < δ < i /2, the most unstable solutions are on the X mode near the cut-oﬀ and N ∼ 0, for 3i /8 > δ, the most unstable solutions are on the Z mode near the resonance and N 1. Nevertheless, compared to the homogeneous case, both the maximum growth rate and the domain of instability are slightly reduced. Whatever the value of the electron energy, there is no possibility of a direct connection between the internal unstable waves and the external X mode waves. From the simple approach (see Sect. 3.3.2), one could conclude that for very energetic sources, such that δ > o /2, some internal unstable waves might directly escape the source on the X mode. The complete analysis shows that this is not the case. When δ is increased, the instability domain is indeed shifted towards lower frequencies which further increases the frequency gap with the external X mode cut-oﬀ. In fact, the direct connection is only possible for overdense source regions (i > o ). Polarization of the discrete mode. Inside the source, for s ∼ 0, the polarization is circular (|Ex | ∼ |Ey |). For s ∼ 1 it is elliptic (|Ex | < |Ey |) and, for s 1 it becomes quasi-longitudinal (|Ex | > |Ey |). The external part of the solutions corresponds to Z mode wave. The amplitudes of the internal and of the external parts of the solutions are of the same order, the level of the external Z mode is similar to that of the internal X mode. This result was also obtained in the simulation studies of Pritchett [50, 52]. It is in contradiction with the observations since no Z mode escapes from the sources.

3.4.3 The Problem of Energy Escape The problem of a too large theoretical production of the Z mode can be solved by considering an oblique propagation (Ny = 0). The component of the Poynting ﬂux that corresponds to the energy escape is proportional to Sx = Ey H . As already mentioned, the polarization is strongly elliptic for N ∼ 1, and Ey can be decreased by rotating the wave vector towards y direction. The X/Z transmission coeﬃcient can then be reduced by a factor larger than 20, from Sx ∼ 0.9 if Ny = 0 to Sx < 0.05 for Ny ∼ 0.95. Wave generation at Ny ∼ 1 would thus explain the very low level of Z mode production. This implies that the radiation would be preferentially generated tangentially

80

P. Louarn

rather than perpendicularly to the source. This is precisely what was deduced from the analysis of the polarization. As discussed in Louarn and Le Qu´eau [36], the fact that waves escape from the sources on the X mode can be explained by considering the upward propagation inside the source. When it propagates upward, the electromagnetic energy generated at a given frequency f indeed “sees” a decreasing fc and thus fxo . At some altitude above the region of generation, f becomes larger than fxo and the connection with the external X mode is possible. For this mechanism to be eﬃcient, the connections between the internal X mode and the external O and Z modes must be as ineﬃcient as possible. Otherwise, the energy would be converted into O and Z modes prior to the possible escape on the X mode. From both the observations and theoretical considerations, we thus get a new scenario of generation of the diﬀerent components of the AKR. During the upward propagation inside the source, the internal X mode waves are connected to the external O and Z modes and a part of the initial energy is converted into Z and O modes. Nevertheless, the transmission coeﬃcients (internal X mode → external O or Z modes) being small, the electromagnetic energy remains conﬁned inside the source until the connection with the external X mode becomes possible (see Fig. 3.12). This indirect production of the observable radio emission by a mode conversion at the source frontiers is one of the speciﬁc properties of the laminar source model. Let us quantify this eﬀect. If W is the density of electromagnetic energy inside the source, L the width of the source, T the coeﬃcient of energy transmission across the interfaces, vg⊥ and vg the perpendicular and parallel group velocities, one

X-mode

O-mode

upward propagation

Source

X-mode Plasma density Production of O-mode

Plasma cavity Fig. 3.12. Schematic of wave refraction and propagation inside and outside the sources. Sources corresponding to deep plasma cavities produce more O mode

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

has: L

vg⊥ ∂W = −2T W ∂h vg

81

(3.29)

where h is the altitude. The energy radiated from the source, assuming an upward propagation from ho to h is: h dz v g⊥ (3.30) W (h) = W0 exp −2 T vg L ho

T can be split into three terms: Txx the X/X transmission, Txo the X/O and, Txz , the X/Z ones. No Z mode escapes from the source which, as already presented, could be explained if the radiation is produced tangentially to the source. Thus, with this polarization, one has Txz ∼ 0. Txx is zero from ho to the altitude of possible connection with the X mode (hx ). This altitude is such that: (3.31) fxo (hx ) = fc (hx )(1 + o /2) ≤ fc (ho ) , where fxo (h) is the cut-oﬀ of the external X mode at altitude h. Assuming a linear variation of the magnetic ﬁeld, h = hx − ho is of the order of Ro /6. This altitude range is ∼100 km if R = 15000 km and o = 4 × 10−2 . Txo can be calculated from the continuity of the electromagnetic ﬁeld components. Using the dispersion relation in Sect. 4.1, it is possible to perform a parametric study of Txo . One approximately gets: Txo = 10−5 o δ 3

(3.32)

Using the equation of transfer, one obtains the relative proportion of X and O mode produced by the source, assuming that the energy not converted into O mode escapes as X mode whenever this becomes possible: O/X = (1 − A)/A

with

A ∼ exp[−10−4 o δ 3 (o + δ/2)R/L]

(3.33)

This expression shows that the proportion of O mode increases if: • • • •

the energy of the emitting particles increases (role of δ), the density of the external plasma increases (role of o ), the distance to the planet increases (R), or the width of the source decreases (L).

In conclusion, narrow energetic sources, embedded in a dense plasma, produce a more powerful O mode radiation. Again, as explained in Sect. 3.3, this is an observational fact.

82

P. Louarn

3.5 Discussion and Conclusion 3.5.1 O and X Mode Production in Narrow Sources An important aspect of the laminar source model is to separate the initial generation of the electromagnetic energy (the maser inside the source) and the production of the observable radio emission (by conversion across the source frontiers). One may consider that the maser operates inside the sources with maximum eﬃciency: this produces X mode waves at the gyrofrequency in a fully relativistic plasma. The produced electromagnetic energy is conﬁned in the source, and mode conversion must be taken into account for explaining the energy escape. During this stage, radiation in the O mode is produced. To illustrate this scenario, we have performed a parametric study corresponding to terrestrial kilometric and Jovian decametric radiation respectively [Louarn, 37]. These two cases diﬀer by the distances from the center of the planet (R = 15000 km for the Earth and R = 70000 km for Jupiter) and the value of the gyrofrequency, a factor of 100 larger at Jupiter than at Earth. Since fc is used for normalization, one obtains normalized parameters (i,o , L . . .) much smaller in the Jovian than the terrestrial cases. For the terrestrial case, one shows that only relatively extreme conditions (width smaller than 10 km, strong density: o ∼ 0.3 and high energy: 10 keV) lead to the generation of a large fraction of O mode, as is actually observed with Viking. For the Jovian case, the expected very low values of o due to the strong magnetization of the plasma make O mode production even more diﬃcult. Even for very narrow sources and energetic particles, the X mode is thus expected to dominate. It is nevertheless possible that energetic very low altitude sources (so that o could be larger) emit in the O mode. This discussion could be extended to other types of radio emission. For example, in the solar corona, the laminar source model is consistent with the idea of fragmented energy release. During a ﬂare, the global energy release could well take place over small scale regions that could also be the sources of non-thermal emission. The polarization of the radiation coming from such a strongly structured ﬁbrous-like corona would simply result from the conversion processes at the frontiers of the sources, not from speciﬁc generation processes [see Sharma et al., 54]. 3.5.2 Fine Structure of Radiation What could be the radiating diagram of a laminar source? Could it explain some of the ﬁne structures observed in the radio spectra by Baumback and Calvert [4], Ergun et al. [17], and Pottelette et al. [45]? As already discussed, the ﬁlamentary geometry creates an asymmetry around the magnetic ﬁeld. The x and y directions are indeed not equivalent: ky is a free parameter when kx is quantized (kx = 2πn/L). For a given ky , the unstable waves are thus emitted at discrete frequencies, along preferential directions of propagation, the radiation diagram being a succession of

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

83

narrow beams in the plane perpendicular to B0 . At a diﬀerent ky , a slightly modiﬁed diagram will be obtained with diﬀerent quantized frequencies. Due to the possibility of continuous variation of ky , the narrow beams likely overlap and, at some distance from the source, a continuous dynamic spectrum would be observed. This point is discussed by Pritchett et al. [52]. Without other ingredients as, for example, a mechanism that would explain wave generation at preferred altitudes or an interaction with non-linear structures [see Pottelette and Treumann, 46, this issue], it seems unlikely that the laminar source model alone may explain the ﬁne structures. Nevertheless, the possibility of the formation of conics has still not been analyzed. Due to the quantization of kx , it is indeed not excluded that the energy radiated by the source concentrates along deﬁned ray-paths. A spacecraft crossing these ray-paths would see a more powerful emission at deﬁned frequencies. Another possibility would be that both ky and kx are quantized as it would be the case for sources limited in both the x and the y directions. This 3-D model also remains to be studied. 3.5.3 Conclusion and Pending Questions To conclude, the laminar source model has been precisely tested, both observationally and theoretically, and it is certainly not exaggerated to aﬃrm that the AKR generation is today one of the best-known magnetospheric phenomena. Given the similarity of the terrestrial auroral phenomena with those at Jupiter and Saturn, the laminar source model can also certainly be generalized. One fundamental point is still lacking: the prediction of the power of the emission, in relation with the energy and the number of the accelerated particles. An important diﬃculty here is the identiﬁcation of the process that saturates the maser. Is this just the convection of the energy out of the source? Is it some non-linear eﬀect, for example the relaxation of the free energy by non-linear wave/particle interactions? The full quantitative understanding of the coherent radio wave production remains an exciting goal for the future, with many potential applications regarding the interpretation of radio emission from distant astrophysical objects.

References [1] Aschwanden, M.J. and A. Benz: On the electron cyclotron instability: II. Pulsation in the quasi-stationary state, Astrophys. J. 332, 466, 1988. [2] Bahnsen, A., M. Jespersen, E. Ungstrup, and I.B. Iversen: Auroral hiss and kilometric radiation measured from the Viking satellite, Geophys. Res. Lett. 14, 471, 1987. [3] Bahnsen A., B.M. Pedersen, M. Jespersen, E. Ungstrup, L. Eliasson, J.S. Murphree, D. Elphinstone, L. Blomberg, G. Hohmgren, and L.J. Zanetti: Viking observations at the source of the AKR, J. Geophys. Res. 94, 6643, 1989.

84

P. Louarn

[4] Baumback, M.M. and W. Calvert: The minimum bandwidths of auroral kilometric radiation, Geophys. Res. Lett. 14, 119, 1987. [5] Bekeﬁ, G.: Emission processes in plasma, p. 229, Gordon & Breach, New York, 1963. [6] Benediktov, E.A., G.G. Getmantsev, Y.A. Sazonov, and A.F. Tarojov: Preliminary results of measurements of the intensity of distributed extraterrestrial radio frequency emission at 725 and 1525 kHz (in Russian), Kosm. Issled. 1, 614, 1965. [7] Benson, R.F. and W. Calvert: ISIS-1 observations at the source of auroral kilometric radiation, Geophys. Res. Lett. 6, 479, 1979. [8] Benson, R.F., W. Calvert, and D.M. Klumpar: Simultaneous wave and particle observations in the auroral kilometric radiation source region, Geophys. Res. Lett. 7, 959, 1980. [9] Benson, R.F. and S.I. Akasofu: Auroral kilometric radiation/aurora correlation, Radio Sci. 19, 527, 1984. [10] Calvert, W.: The auroral plasma cavity, Geophys. Res. Lett. 8, 919, 1981. [11] Calvert, W.: A feedback model for the source of auroral kilometric radiation, J. Geophys. Res. 87, 8199, 1982. [12] Carlson, C.W., R.F. Pfaﬀ, and J.G. Watzin: The fast auroral snapshot (FAST) mission, Geophys. Res. Lett. 25, 2017, 1998. [13] De Feraudy, H., B.M. Pedersen, A. Bahnsen, and M. Jespersen: Viking observations of AKR. from plasmasphere to night auroral oval source region, Geophys. Res. Lett. 14, 511, 1987. [14] Delory, G.T., R.E. Ergun, C.W. Carlson, L. Muschietti, C.C. Chaston, W. Peria, J.P. McFadden, and R.J. Strangeway: FAST observations of electron distributions within AKR source regions, Geophys. Res. Lett. 25, 2069, 1998. [15] Dulk, G.A.: Radio emission From the sun and the stars, Ann. Rev. Astron. Astrophys. 23, 169, 1985. [16] Eliasson, L., G.A. Holmgren, and K. R¨ onnmark: Pitch-angle and energy distributions of auroral electrons measured by the ESRO-4 satellite, Planet. Space Sci. 27. 87, 1979. [17] Ergun, R.E., et al.: FAST satellite wave observations in the AKR source region, Geophys. Res. Lett. 25, 2061, 1998. [18] Ergun, R.E., C.W. Carlson, J.P. McFadden, G.T. Delory, R.J. Strangeway, and P.L. Pritchett: Electron-cyclotron maser driven by charged particle acceleration from magnetic ﬁeld-aligned electric ﬁelds, Astrophys. J. 538, 456, 2000. [19] Ergun, R.E., et al.: The FAST satellite ﬁelds instrument, Space Sci. Rev. 98, 67, 2001. [20] Green, J.L., D.A. Gurnett, and R.A. Hoﬀman: Correlation between auroral kilometric radiation and inverted-V electron precipitations, J. Geophys. Res. 84, 5216, 1979. [21] Gurnett, D.A.: The Earth as a radio source: terrestrial kilometric radiation, J. Geophys. Res. 79, 4227, 1974. [22] Gurnett, D.A. and J.L. Green: On the polarization and origin of auroral kilometric radiation, J. Geophys. Res. 83, 689, 1978. [23] Gurnett, D.A. and R.R. Anderson: The kilometric radio emission spectrum; Relation to auroral acceleration processes, in Physics of Auroral Arc Formation, Geophys. Monogr. Ser., vol. 25, edited by S.-l. Akasofu, and J. R. Kan, p. 341, AGU, Washington, D. C., 1981.

3 Generation of Auroral Kilometric Radiation in Bounded Source Regions

85

[24] Hilgers, A., A. Roux, and R. Lundin: Characteristics of AKR sources: a statistical description, Geophys. Res. Lett. 18, 1493, 1991. [25] Hilgers, A., H. de Feraudy, and D. Le Qu´eau: Measurement of the direction of the auroral kilometric radiation electric ﬁeld inside the source with the Viking satellite, J. Geophvs. Res. 79, 8381, 1992. [26] Kaiser, M.L., J.K. Alexander, A.C. Riddle, J.B. Pierce, and J.W. Warwick: Direct measurements by Voyager 1 and 2 of the polarization of terrestrial kilometric radiation, Geophys. Res. Lett. 5., 857, 1978. [27] Lee, L.C. and C.S. Wu: Ampliﬁcation of radiation near cyclotron frequency due to electron population inversion, Phys. Fluids 22, 1348, 1980. [28] Le Qu´eau, D., R. Pellat, and A. Roux: Direct generation of the auroral kilometric radiation by the maser synchrotron instability: An analytical approach, Phys. Fluids 27, 247, 1984a. [29] Le Qu´eau, D., R. Pellat, and A. Roux: Direct generation of the auroral kilometric radiation by the maser synchrotron instability: Physical discussion of the mechanism and parametric study, J. Geophys. Res. 89, 2841, 1984b. [30] Le Qu´eau, D., R. Pellat, and A. Roux: The maser synchrotron instability in an inhomogeneous medium: Application to the generation of the auroral kilometric radiation, Ann. Geophys. 3, 273, 1985. [31] Le Qu´eau, D. and P. Louarn: Analytical study of the relativistic dispersion: application to the generation of the AKR, J. Geophys. Res. 94, 2605, 1989. [32] Louarn, P., D. Le Qu´eau, and A. Roux: A new mechanism for stellar radio bursts: the fully relativistic electron maser, Astron. Astrophys. 165, 211, 1986. [33] Louarn, P., A. Roux, H. de Feraudy, D. Le Qu´eau, M. Andr´e, and L. Matson: Trapped electrons as a free energy source for the auroral kilometric radiation, J. Geophys. Res. 95, 5983, 1990. [34] Louarn, P., D. Le Qu´eau, and A. Roux: Formation of electron trapped population and conics inside and near auroral acceleration region, Ann. Geophys. 9, 553, 1991. [35] Louarn, P. and D. Le Qu´eau: Generation of the auroral kilometric radiation in plasma cavities, I, Experimental study, Planet. Space Sci. 44, 199, 1996a. [36] Louarn, P. and D. Le Qu´eau: Generation of the auroral kilometric radiation in plasma cavities, II, The cyclotron maser instability in small size sources, Planet. Space Sci. 44, 211, 1996b. [37] Louarn, P.: Radio emissions from ﬁlamentary sources: a simple approach, Planetary radio emissions IV, Ed. Rucker H.O., Bauer S.J. and Lecacheux A., Verlag ¨ der Osterreichischen Akademie der Wissenschaften, 1997. [38] Mellott, M.M., W. Calvert, R.L. Huﬀ, D.A. Gurnett, and S.D. Shawhan: DE 1 observation of ordinary mode and extraordinary mode auroral kilometric radiation, Geophys. Res. Lett. 11, 1188, 1984. [39] Melrose, D.B. and G.A. Dulk: Electron-cyclotron masers as the source of certain solar and stellar radio bursts, Astrophys. J., 259, 844, 1982. [40] Melrose, D.B., K.G. R¨ onnmark, and R.G. Hewitt: Terrestrial kilometric radiation: The cyclotron theory, J. Geophys. Res. 87, 5140, 1982. [41] Omidi, N. and D.A. Gurnett: Growth rate calculations of auroral kilometric radiation using the relativistic resonance condition, J. Geophys. Res. 87, 2377, 1982. [42] Omidi, N. and D.A. Gurnett: Path-integrated growth of auroral kilometric radiation, J. Geophys. Res. 89, 10,801, 1984.

86

P. Louarn

[43] Perraut, S., H. de Feraudy, A. Roux, P.M.E. D´ecr´eau, J. Paris, and L. Matson: Density measurements in key regions of the Earth’s magnetosphere: Cusp and auroral region, J. Geophys. Res. 95, 5997, 1990. [44] Pottelette, R., M. Malingre, A. Bahnsen, L. Eliasson, and K. Stasiewicz: Viking observations of bursts of intense broadband noise in the source region of auroral kilometric radiation, Ann. Geophys. 6, 573, 1988. [45] Pottelette, R., R.A. Treumann, and M. Berthomier: Auroral plasma turbulence and the cause of auroral kilometric radiation ﬁne structure, J. Geophys. Res. 106, 8465, 2001. [46] Pottelette, R. and R.A. Treumann: Auroral Acceleration and Radiation, in: part 1 of this book, 2005. [47] Pritchett, P.L.: Relativistic dispersion, the cyclotron maser instability, and auroral kilometric radiation, J. Geophys. Res. 89, 8957, 1984. [48] Pritchett, P.L. and R.J. Strangeway: A simulation study of kilometric radiation generation along an auroral ﬁeld line, J. Geophys. Res. 90, 9650, 1985. [49] Pritchett, P.L.: Electron-cyclotron maser instability in relativistic plasmas, Phys. Fluids 29, 2919, 1986a. [50] Pritchett, P.L.: Cyclotron maser radiation from a source structure localized perpen dicular to the ambient magnetic ﬁeld, J. Geophys. Res. 91, 13,569, 1986b. [51] Pritchett, P.L. and R.M. Winglee: Generation and propagation of kilometric radiation in the auroral plasma cavity, J. Geophys. Res. 94, 129, 1989. [52] Pritchett, P.L., R.J. Strangeway, R.E. Ergun, and C.W. Carlson: Generation and propagation of cyclotron maser emissions in the ﬁnite auroral kilometric ra-diation source cavity, J. Geophys. Res. 107, 12, doi:10.1029/2002JA009403, 2002. [53] Roux, A., A. Hilgers, H. de Feraudy, D. Le Qu´eau, P. Louarn, S. Perraut, A. Bahnsen, M. Jespersen, E. Ungstrup, and M. Andr´e: Auroral kilometric radiation sources: in situ and remote sensing observations from Viking, J. Geophys. Res. 98, 11657, 1993. [54] Sharma, R.R., L. Vlahos, and K. Papadopoulos: The importance of plasma eﬀects on electron-cyclotron maser emission from ﬂaring loops, Astron. Astrophys. 112, 377, 1982. [55] Shawhan, S.D. and D.A. Gurnett: Polarization measurements of auroral kilometric radiation by DE-1, Geophys. Res. Lett. 9, 913, 1982. [56] Strangeway, R.J.: Wave dispersion and ray propagation in a weakly relativ-istic electron plasma: Implications for the generation of auroral kilometric radiation, J. Geophys. Res. 90, 9675, 1985. [57] Trubnikov, B.A.: In Plasma physics and the problem of controlled thermonuclear reactions, ed. M.A. Leontovich, Pergamon Press Inc., New york, 1959. [58] Ungstrup, E., A. Bahnsen, H.K. Wong, M. Andr´e, and L. Matson: Energy source and generation mechanism for AKR, J. Geophys. Res. 95, 5973, 1990. [59] Wu, C.S. and L.C. Lee: A theory of the terrestrial kilometric radiation, Astrophys. J. 230, 621, 1979. [60] Zarka, P.: The auroral radio emissions from planetary magnetospheres: what do we know, what don’t we know, what do we learn from them? Adv. Space Res. 12, 99. 1992.

4 Generation of Emissions by Fast Particles in Stochastic Media G.D. Fleishman National Radio Astronomy Observatory, Charlottesville, VA 22903 [email protected]

Abstract. We demonstrate the potential importance of small-scale turbulence in the generation of radio emission from natural plasmas. This emission, being reliably detected and interpreted, probes small-scale turbulence in remote sources in the most direct way. The radiation emitted is called “diﬀusive synchrotron radiation” because it is related to shaking the electron distribution when passing through randomly distributed small-scale plasma inhomogeneities caused in turbulence. The emissivity is calculated and shown to be in the observable range. A further eﬀect of inhomogeneities is transition radiation arising from fast particles which interact with small-scale density inhomogeneities. This emission process generates continuum emission below synchrotron and is applicable to some solar radio bursts. It serves for probing number densities. The eﬀect of inhomogeneities on coherent emissions is either broadening or splitting of the spectral peaks generated by the electron cyclotron maser mechanism.

Key words: Solar radio bursts, random inhomogeneities, turbulent radiation, diﬀuse synchrotron radiation

4.1 Introduction Interaction of charged particles with each other and/or with external ﬁelds results in emission of electromagnetic radiation. This article considers the emission arising as fast (nonthermal) particles move through media with random inhomogeneities. The nature of these inhomogeneities might be rather arbitrary. One of the simplest examples of inhomogeneities is a distribution of the microscopic particles (atoms or molecules) in an amorphous substance, so the medium is inhomogeneous at microscopic scales (of the order of the mean distance between particles), perhaps remaining uniform (on average) at macroscopic scales. More frequently, however, real objects are inhomogeneous on macroscopic scales as well. The irregularities might be related to the interfaces between G.D. Fleishman: Generation of Emissions by Fast Particles in Stochastic Media, Lect. Notes Phys. 687, 87–104 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

88

G.D. Fleishman

inhomogeneities, variations of the elemental composition, temperature, density, electric, and magnetic ﬁeld. Random inhomogeneities of any of these parameters may strongly aﬀect in various ways the generation of electromagnetic emission. For example, the presence of the density inhomogeneities implies that the dielectric permeability tensor is a random function, as well as the refractive indices of the electromagnetic eigen-modes. As a result, the eigen-modes of the uniform medium are not the same as the eigen-modes of the real inhomogeneous medium. The irregularities of the electric and magnetic ﬁelds aﬀect primarily the motion of the fast particle (although the eﬀect of the ﬁeld ﬂuctuations on the dielectric tensor exists as well). Below we consider one of many emission processes appearing due to or affected by small-scale random inhomogeneities, namely, diﬀusive synchrotron radiation arising as fast particles are scattered by the small-scale random ﬁelds. This emission process is of exceptional importance since current models of many astrophysical objects (see, e.g., [6, 7] and references therein) imply generation of rather strong small-scale magnetic ﬁelds. The eﬀect of the inhomogeneities on other emission processes is discussed brieﬂy as well.

4.2 Statistical Methods in the Theory of Electromagnetic Emission The trajectories of charged particles and the ﬁelds created by them are random functions as the particles move through a random medium. Thus, the use of appropriate statistical methods is required to describe the particle motion and the related ﬁelds. 4.2.1 Spectral Treatment of Random Fields For a detailed theory of the random ﬁelds we refer to a monograph [Toptygin, 20] and mention here a few important points only. To be more speciﬁc, let us discuss some properties of the random magnetic ﬁelds. Assume that total magnetic ﬁeld is composed of regular and random components B(r, t) = B 0 (r, t) + B st (r, t), such as B 0 (r, t) = B(r, t) and

B st (r, t) = 0, where the brackets denote the statistical averaging. Note, that the method of averaging depends on the problem considered. The statistical properties of the random ﬁeld might be described with a (inﬁnite) sequence of the multi-point correlation functions, the most important of which is the (two-point) second-order correlation function (2)

Kαβ (R, T, r, τ ) = Bst,α (r 1 , t1 )Bst,β (r 2 , t2 ) , where R = (r 1 + r 2 )/2, r = r 2 − r 1 , T = (t1 + t2 )/2, and τ = t2 − t1 .

(4.1)

4 Generation of Emissions by Fast Particles in Stochastic Media

89

Since the regular and random ﬁelds are statistically independent ( B0 Bst = B0 Bst = 0), each of them satisﬁes the Maxwell equations separately. In particular ∇ · B st = 0, so only two of three vector components of the random ﬁeld are independent. For a statistically uniform random ﬁeld the Fourier transform of the cor(2) relator Kαβ (r, τ ) over spatial and temporal variables r and τ gives rise to the spectral treatment of the random ﬁeld drdτ i(ωτ −kr) (2) e Kαβ (r, τ ) . (4.2) Kαβ (k, ω) = (2π)4 In the case of isotropic turbulence we easily ﬁnd

1 kα kβ Kαβ (k, ω) = K(k)δ(ω − ω(k)) δαβ − 2 , 2 k

(4.3)

which, in particular, satisﬁes the the Maxwell equation ∇ · B st = 0, since the tensor structure of the correlator is orthogonal to the k vector: kα Kαβ = 0. Although the spectral shape of the correlators is not unique and may substantially vary depending on the situation, we will adopt for the purpose of the model quasi-power-law spectrum of the random measures: " 2# ν−1 Γ (ν/2 + 1)kmin A Aν , Aν = , (4.4) K(k) = 2 2 ν/2+1 3/2 (kmin + k ) 3π Γ (ν/2 − 1/2) where ν is the spectral index of the turbulence, and the spectrum K(k) is normalized to d3 k: kmax " # K(k)d3 k = A2 , for kmin kmax , ν > 1 , (4.5) 0

where A2 is the mean square of the corresponding measure of the random 2 2 , Est , ∆N 2 etc. ﬁeld, e.g., Bst 4.2.2 Emission from a Particle Moving along a Stochastic Trajectory The intensity of the emission of the eigen-mode σ σ En,ω = (2π)6

ω 2 nσ (ω) |(eσ · j ω,k )|2 c3

(4.6)

depends on the trajectory of the radiating charged particle since the Fourier transform j ω,k of the corresponding electric current has the form: ∞ dt v(t) exp(iωt − ikr(t)) , (4.7) j ω,k = Q (2π)4 −∞

90

G.D. Fleishman

where Q is the charge of the particle. For the stochastic motion of the particle, we have to substitute (4.7) into (4.6) and perform the averaging of the corresponding expression: σ En,ω =

Q2 ω 2 nσ (ω) 4π 2 c3 T ×Re dt −T

(4.8) ∞

dτ eiωτ e−ik[r(t+τ )−r(t)] (e∗σ · v(t + τ ))(eσ · v(t)) ,

0

where 2T is the total time at which the emission occurs, and eσ is the polarization vector of the eigen-mode σ. It is convenient to perform the averaging denoted by the brackets with the use of the distribution function of the particle(s) F (r, p, t) at the time t and the conditional probability W (r, p, t; r , p , τ ) for the particle to transit from the state (r, p) to the state (r , p ) during the time τ . For statistically uniform random ﬁeld we obtain: σ En,ω = ×Re

Q2 ω 2 nσ (ω) (4.9) 4π 2 c3 T ∞ dt dτ eiωτ drdpdp (e∗σ · v )(eσ · v)F (r, p, t)Wk (p, t; p , τ ) .

−T

0

Then, the integration over τ gives rise to the temporal Fourier transform of W , so the spectrum of emitted electromagnetic waves is expressed via the spatial and temporal Fourier transform of the distribution function of the particle in the presence of the random ﬁeld. 4.2.3 Kinetic Equation in the Presence of Random Fields The conditional probability W , which substitutes the particle trajectory in the presence of random ﬁelds, can be obtained from the kinetic Boltzmannequation: ∂f ∂f ∂f +v· + FL · =0, (4.10) ∂t ∂r ∂p where F L = QE + Qc (v × B) is the Lorentz force, while the electric (E) and magnetic (B) ﬁelds contain in the general case both regular and random components. Let us express the Lorentz force as a sum of these two components explicitly: (4.11) F L = F R + F st . Accordingly, we’ll seek a distribution function in the form of the sum of the averaged (W ) and ﬂuctuating (δW ) components: f (r, p, t) = W (r, p, t) + δW (r, p, t) .

(4.12)

The equation for the averaged component W can be derived from (4.10) applying the Green’s function method [Toptygin, 20]:

4 Generation of Emissions by Fast Particles in Stochastic Media

91

2 Qc ∂W ∂W → − +v· − (Ω · O )W = (4.13) ∂t ∂r E ∞ × dτ Oα Tαβ [∆r(τ ), τ ]Oβ W [r − ∆r(τ ), p − ∆p(τ ), t − τ ] , 0

where

QBc (4.14) E is a vector pointing in the direction of the magnetic ﬁeld and whose magnitude equals the rotation frequency of the charged particle with energy E, Ω=

2 $

Bst rα rβ % . ψ(r)δαβ + ψ1 (r) 2 3 r (4.15) To derive (4.13) we transformed the terms with the magnetic ﬁeld using the following property of the scalar triple product: Q ∂ QB ∂ ∂ QBc (v × B) · =− · v× · O, O= v× =− , c ∂p c ∂p E ∂v (4.16) where O is the operator of the angular variation of the velocity. The equation (4.13) is rather general and can be applied to the study of both emission by fast particles and particle propagation in the plasma [Toptygin, 20]. Further simpliﬁcations of equation (4.13) can be done by taking into account some speciﬁc properties of the problems considered. The theory of wave emission involves a fundamental measure called the coherence length (or the formation zone) that refers to that part of the particle path where the elementary radiation pattern is formed. The coherence length is much larger than the wavelength for the case of relativistic particles, e.g., the coherence length for synchrotron radiation in the presence of the uniform magnetic ﬁeld is ls = RL /γ = M c2 /(QB), where RL is the Larmor radius, γ = E/M c2 is the Lorentz-factor of the particle. Length ls is by the factor of γ 2 larger than the corresponding wave length. The eﬀect of magnetic ﬁeld inhomogeneity on the elementary radiation pattern is speciﬁed by the ratio of the spatial scale of the ﬁeld inhomogeneity and the coherence length. If the scale of inhomogeneity is much larger than the coherence length, the eﬀect of the inhomogeneity is small and can typically be discarded. However, if the magnetic ﬁeld changes noticeably at the coherence length, the inhomogeneity aﬀects the emission strongly, so the spectral and angular distributions of the intensity and polarization of the emission can be remarkably diﬀerent from the case of the uniform ﬁeld. This means, in particular, that in the presence of magnetic turbulence with a broad distribution over the spatial scales, the large-scale spatial irregularities should be considered like the regular ﬁeld, while the small-scale ﬂuctuations should be properly taken into account as the random ﬁeld. Since the variation

Tαβ (r, τ ) = Bst,α (r 1 , t1 )Bst,β (r 2 , t2 ) =

92

G.D. Fleishman

of the particle speed (momentum) over the correlation length of the smallscale random ﬁeld is small, then we can adopt ∆p(τ ) = 0, ∆r(τ ) = vτ in the right-hand-side of (4.13). Then, the kinetic equation takes the form ∂W ∂W

B 2 → − +v· − (Ω · O )W = st ∂t ∂r 3

Qc E

2 ∞ O2 dτ ψ(vτ )W (r − vτ, p, t − τ ) , 0

(4.17) 4.2.4 Solution of Kinetic Equation Let us outline the solution of the kinetic equation (4.17) for the averaged distribution function W . First of all, the stochastic ﬁeld has to be split into large& and small-scale (B st ) components. To see this, consider a purely scale (B) sinusoidal spatial wave of the magnetic ﬁeld with the strength B0 and the wavelength λ0 = 2π/k0 . If the wavelength λ0 is less than the coherence length ls0 calculated for the emission in a uniform ﬁeld B0 , ls0 ∼ M c2 /(QB0 ), this wave represents the small-scale ﬁeld, whose spatial inhomogeneity is highly > ls0 , it is a large-scale important for the emission; in the opposite case, λ0 ∼ ﬁeld. The splitting is less straightforward when the random ﬁeld is a superposition of the random waves with a quasi-continuous distribution over the spatial scales. Let us consider the eﬀect provided by a random magnetic ﬁeld corresponding to a small range ∆k in the spectrum (4.4) on the charged particle trajectory. The energy of this magnetic ﬁeld is δEst ∼ K(k)k 2 ∆k .

(4.18) √ The corresponding non-relativistic “gyrofrequency” is δωst ∼ Q δEst /(M c). For a truly random ﬁeld, when the harmonics with k and k+dk are essentially uncorrelated, we can arbitrarily select the value ∆k to be small enough to satisfy δωst kc for any k, so all the independent ﬁeld components represent the small-scale ﬁeld. However, in a more realistic case the Fourier components of the random ﬁeld with similar yet distinct k are typically correlated, so they disturb the particle motion coherently and ∆k in estimate (4.18) cannot be arbitrarily small any longer. Accordingly, all components of the random ﬁeld < δωst /c (where δωst is calculated for the smallest allowable ∆k) must with k ∼ be treated as a large-scale ﬁeld. & together with the regular ﬁeld B 0 speciﬁes the The large-scale ﬁeld B vector Ω in the left-hand-side of equation (4.17): Ω=

& Q(B 0 + B)c . E

(4.19)

For the analysis of the emission process (and, respectively, for the solution of the kinetic equation (4.17)), we treat the large-scale ﬁeld (which is the sum of the regular and large-scale stochastic ﬁelds) as uniform, (Ω = const);

4 Generation of Emissions by Fast Particles in Stochastic Media

93

the actual inhomogeneity might be taken into account by averaging the ﬁnal expressions of the emission if necessary. Equation (4.17) has been solved in the presence of a uniform magnetic ﬁeld and small-scale random magnetic ﬁelds [cf. 13]: 2 ωpe 1 ωv 1− τ w(θ 0 , θ, τ ) , (4.20) Wk = 2 δ(p − p0 ) exp −i p c 2ω 2 where

x exp − x(θ2 + θ02 ) coth zτ π sinh zτ 2 τ (θ − θ 0 ) · (n × Ω) Ω⊥ − + 2xθθ 0 sinh−1 zτ − , (4.21) 2q 4q

θ2 θ2 v0 v −n 1− 0 , θ0 = θ = −n 1− , (4.22) v 2 v 2

1 1 ω 2 , z = (1 − i)(ωq) 2 , (4.23) x = (1 − i) 16q 2 Qc (4.24) dk K(k )δ[ω − (k − k )v] . q(ω, θ) = π E

w(θ 0 , θ, τ ) =

If there is only a random ﬁeld and no regular ﬁeld, the function w reads x −1 2 2 exp −x(θ +θ0 ) coth zτ +2xθθ 0 sinh zτ , (4.25) w(θ 0 , θ, τ ) = π sinh zτ while in the opposite case, when there is no random ﬁeld, we have iωτ 2 Ω2 τ 2 w(θ 0 , θ, τ ) = δ(θ − θ 0 + [n × Ω]τ ) exp θ0 − θ 0 · [n × Ω]τ + ⊥ . 2 3 (4.26) Calculation of the emission with the use of this distribution function leads evidently to the standard expressions of synchrotron radiation in the uniform magnetic ﬁeld. Finally, the distribution function of the free particle (moving without any acceleration), which does not produce any emission in the vacuum or uniform plasma, is iωτ 2 θ . (4.27) w0 (θ 0 , θ, τ ) = δ(θ − θ 0 ) exp 2

94

G.D. Fleishman

4.3 Emission from Relativistic Particles in the Presence of Random Magnetic Fields 4.3.1 General Case Let us consider the energy emitted by a single particle (regardless of polarization) based on the general expression (4.9), i.e., we take the sum of (4.9) over the two orthogonal eigen-modes: ' Q2 ω 2 ε(ω) (4.28) En,ω = 2π 2 c3 T ∞ × Re dt dτ eiωτ dr dp dp [n × v ] · [n × v]F (r, p, t)Wk (p, t; p , τ ) , −T

0

where we neglected the diﬀerence between 'the two refractive indices in the magnetized plasma and adopted nσ (ω) ≈ ε(ω). In the presence of a statistically uniform and stationary magnetic ﬁeld the emitted energy (4.28) is proportional to the time (on average), although the intensity of emission at a given direction n depends on time since the angle between the instantaneous particle velocity v(t) and n changes with time as described by the dependence of the function F (r, p, t) on time t. This kind of the temporal dependence is not of particular interest, e.g., it represents periodic pulses provided by the rotation of the particle in the uniform magnetic ﬁeld, so it is more convenient to proceed with time-independent intensity of radiation emitted into the full solid angle ∞ 2 2 γ ωpe Q2 ω 2 ' iωτ ε(ω) Re dτ exp 1+ (4.29) Iω = 2π 2 c 2γ 2 ω2 0 × d2 θ d2 θ (θθ )(w(θ, θ , τ ) − w0 (θ, θ , τ )) , where w0 (θ, θ , τ ) is the distribution function of the free particle (4.27), which does not contribute to the electromagnetic emission (since the VavilovCherenkov condition cannot be fulﬁlled in the plasma or vacuum). Then, calculation of (4.29), described in detail in [21], results in 2 γ 2 ωpe Q2 ω 8Q2 q (ω) γ 2 Φ1 (s1 , s2 , r) + 1+ Φ2 (s1 , s2 , r) , (4.30) Iω = 3πc [1 + γ 2 (ωpe /ω)2 ] 4πcγ 2 ω2 where Φ1 (s1 , s2 , r) and Φ2 (s1 , s2 , r) stand for the integrals: ∞ 6|s|4 Φ1 (s1 , s2 , r) = Im dt exp(−2st) × (4.31) s1 s2 0

t 1 × coth t exp −2rs3 coth t − sinh−1 t − − , 2 t

4 Generation of Emissions by Fast Particles in Stochastic Media

95

∞ cosh t − 1 × (4.32) Φ2 (s1 , s2 , r) = 2r|s|2 Re dt sinh t 0

t × exp −2st − 2rs3 coth t − sinh−1 t − − iφ , 2

which depend on the dimensionless parameters s1 , s2 , r: φ

12 π 2 2 ωpe γ ω e−i( 4 + 2 ) √ s = s1 − is2 = 1+ , |q(ω)| ω2 4 2γ 2 −3

2 2 2 ωpe γ Ω⊥ 6 r = 32γ . 1+ ω ω2

(4.33)

(4.34)

The parameter s depends on the rate of scattering of the particle by magnetic inhomogeneities q(ω), which has the form √ 2 ν−1 2 πΓ (ν/2)ωst ω0 α (ν − 1)ωst q(ω) = , (4.35) +i 2 2 2 2 2 ν/2 3γ ω0 (1 + (ν − 1)α2 /ω02 ) 3Γ (ν/2 − 1/2)γ (α + ω0 ) for power-law distribution of magnetic irregularities over scales: P (k) = 2 2 4πK(k)k 2 ∝ k −ν at k kmin = ω0 /c, where ωst = Q2 Bst /(M c)2 is the square of the cyclotron frequency in the random magnetic ﬁeld, α = 2 /ω 2 , and a is a factor of the order of unity. (aω/2) γ −2 + ωpe In the general case the integrals (4.31, 4.32) cannot be expressed in terms of elementary functions. However, there are convenient asymptotic expressions of these integrals. In particular, if r|s|3 1 and r|s|3 |s| we have 1 1 1 2 |s|4 (2π − 3φ), Φ2 ≈ 2 3 3 6 Γ (4.36) Φ1 ≈ − r3 , s1 s2 3 while for |s| 1, r 1, the functions Φ1 and Φ2 contain exponentially small terms: ( 1 1 3π 2 r 4 |s|4 8 2 , (4.37) Φ1 ≈ 1 − exp − 5 3 r 2 4 s1 s2 ( 3 1 1 r 8 2 . (4.38) Φ2 ≈ + 2 4 π 2 r 4 exp − 32|s|2 3 r Complementary, for s 1 and rs3 1 we obtain Φ1 ≈ 6s

1 − rs2 , 1 + r2 s4

Φ2 ≈

rs 1 + rs2 . 2 1 + r2 s4

(4.39)

4.3.2 Special Cases The radiation intensity (4.30) depends on many parameters, allowing many diﬀerent parameter regimes. It is clear that in the absence of the random ﬁelds we arrive at standard expressions for synchrotron radiation in a uniform magnetic ﬁeld. Let us consider here a few interesting cases when the presence of small-scale random ﬁeld results in a considerable change of the emission.

96

G.D. Fleishman

Weak Random Magnetic Inhomogeneities Superimposed on Regular Magnetic Field Consider the case of weak magnetic irregularities with a broad (power-law) 2 & 2 , so that distribution over spatial scales, with B⊥ B ωc ω &st ω0 .

(4.40)

Here ωc is the gyro-frequency, and ω &st is the gyro-frequency related to the & 2 12 . total random ﬁeld B Radiation from highly relativistic particles [Toptygin and Fleishman, 21] is mainly speciﬁed by the regular ﬁeld, since either |s| 1 or r|s|3 1. However, at high frequencies, where synchrotron radiation decreases exponentially, the spectrum is controlled by the small-scale ﬁeld: the spectral index of radiation is equal to the spectral index of the random ﬁeld, see Fig. 4.1.

Fig. 4.1. Spectra of radiation by a relativistic particle with γ = 104 for diﬀering value of the random magnetic ﬁeld (left) and with diﬀerent γ in the presence of weak 2 /B02 = 10−4 (right) random magnetic ﬁeld Bst

However, a more interesting regime, which has not been considered so far, takes place for moderately relativistic (and non-relativistic) particles moving in a dense plasma (the case typical for solar and geospace plasmas), when synchrotron radiation is known to be exponentially suppressed according to (4.37), (4.38) by the eﬀect of plasma density (Razin-eﬀect [5, 19]) at all frequencies. The contribution of the small-scale random ﬁeld, which we refer to as diﬀusive synchrotron radiation, in this conditions takes the form: 2 ν−1 2ν ωst ω0 γ Q2 2ν+1 Γ (ν/2)(ν 2 + 7ν + 8) . (4.41) Iω = √ 2 3 πΓ (ν/2 − 1/2)(ν + 2) (ν + 3) c ω ν 1 + ω 2 γ 2 /ω 2 ν+1 pe

It is important that this radiation decreases with the increase of the plasma −ν ) than density (plasma frequency) much more slowly (as a power-law, ∼ ωpe synchrotron radiation. As a result, the diﬀusive synchrotron radiation can

4 Generation of Emissions by Fast Particles in Stochastic Media

97

Fig. 4.2. Spectra of radiation by a relativistic particle with diﬀerent γ in a dense 2 /B02 = 10−6 (left), and plasma in the presence of weak random magnetic ﬁeld Bst 4 with γ = 10 in small-scale random magnetic ﬁeld (right). If ω0 is big enough (e.g., ω0 /ωce = 107 in the ﬁgure) the spectral region provided by multiple scattering, ω 1/2 , disappears

dominate the entire spectrum even if the random ﬁeld is much weaker than the regular ﬁeld, as is evident from the left part of Fig. 4.2: the emission by < 10 is deﬁned exclusively by the small-scale ﬁeld. particles with γ ∼ Small-Scale Magnetic Field Consider an extreme case, which might be relevant in the physics of cosmological gamma-ray bursts, when there is only a small-scale random magnetic ﬁeld but no (very weak) regular ﬁeld, so that ω0 ωst [4]. Now, the parameter q depends substantially on ω0 , the particle motion is similar to the random walk, so the radiation spectrum is similar to some extent to bremsstrahlung provided by multiple scattering of the fast particle by randomly located Coulomb centers. In particular, the spectrum of diﬀusive synchrotron radiation can 1 contain a ﬂat region (as standard bremsstrahlung) and a ∝ ω 2 region (like bremsstrahlung suppressed by multiple scattering), Fig. 4.2 right. However, at suﬃciently high frequencies (ω > ω0 γ 2 ), the ﬂat spectrum gives way to a power-law region ∝ ω −ν typical for the diﬀusive synchrotron radiation. We should note, that the spectrum depends signiﬁcantly on the energy of radiating particle (compare the left part in Fig. 4.3). For low-energy particles some parts of the spectrum (e.g., ﬂat region) might be missing. 4.3.3 Emission from an Ensemble of Particles The results presented in the previous section can be directly applied to monoenergetic electron distributions, which can be generated in the laboratory, but are rare exceptions in nature (e.g., in astro- and geo-plasmas). Natural particle distributions can frequently be approximated by power-laws, say, as function of the dimensionless parameter γ:

98

G.D. Fleishman

Fig. 4.3. Spectra of radiation by a relativistic particle with diﬀerent γ = 30, 3 · 103 , 106 in the presence of small-scale random magnetic ﬁeld (left). Emissivity by fast electron ensemble with diﬀerent energetic spectra (ξ = 2.5, 4.5, 6.5) for the case of dense plasma, ωce /ωpe = 3 · 10−3 (right)

dNe (γ) = (ξ − 1)Ne γ1ξ−1 γ −ξ , γ1 ≤ γ ≤ γ2 ,

(4.42)

where Ne is the number density of relativistic electrons with energies E ≥ mc2 γ1 , and ξ is the power-law index of the distribution. Evidently, the intensity of incoherent radiation produced by the ensemble (4.42) of electrons from the unit source volume is Pω = Iω dNe (γ) . (4.43) Hard Electron Spectrum As we will see, the radiation spectrum produced by an ensemble of particles diﬀers for hard (ξ < 2ν + 1) and soft (ξ > 2ν + 1) distributions of fast electrons over energy. Let us consider ﬁrst the case of hard spectrum [Toptygin and Fleishman, 21], which is typical, e.g., for supernova remnants and radio galaxies. Assuming the small-scale ﬁeld to be small compared with the regular ﬁeld, we may expect the contribution of diﬀusive synchrotron radiation to be noticeable only in those frequency ranges where synchrotron 'emission is small. 2 /ωc⊥ , ωpe ωpe γ1 /ωc⊥ ), synIn particular, at low frequencies ω max(ωpe chrotron radiation is suppressed by the eﬀect of density. Diﬀusive synchrotron radiation is produced by relatively low energy electrons at these frequencies, and each electron produces the emission according to (4.41), which peaks at ω ∼ ωpe γ. Evaluation of the integral (4.43) gives rise to

ν+1−ξ 2 ν−1 (ξ − 1)Γ ν2 (ν 2 + 7ν + 8) e2 Ne γ1ξ−1 ωst ω ω0 Pω √ ν c ωpe ωpe 3 πΓ ν2 − 12 (ν + 2)2 (ν + 3) (4.44) in agreement with Nikolaev and Tsytovich [14]. The spectrum can either increase or decrease with frequency depending on the spectral indices ν and ξ.

4 Generation of Emissions by Fast Particles in Stochastic Media

99

This expression holds for ω ωpe γ1 . If there are no particles with γ < γ1 , the spectrum at even lower frequencies drops as 2 ν−1 ν+2 ω0 ω e2 Ne ωst (ξ − 1)2ν+1 Γ (ν/2)(ν 2 + 7ν + 8) √ . (4.45) 2ν+2 3(ξ + 1) πΓ (ν/2 − 1/2)(ν + 2)2 (ν + 3) c γ12 ωpe ' 2 /ωc⊥ , ωpe ωpe γ1 /ωc⊥ ) ω ωc⊥ γ22 , where At high frequencies max(ωpe the eﬀect of density is not important, the spectrum is speciﬁed by standard synchrotron radiation. However, at higher frequencies, ω ωc⊥ γ22 , the intensity of synchrotron radiation decreases exponentially, and the contribution of diﬀusive synchrotron radiation dominates again. Adding up contributions from all particles described by (4.41) at these frequencies, we obtain

Pω =

2 ν−1 2ν−ξ+1 2ν+1 (ξ − 1)Γ ( ν2 )(ν 2 + 7ν + 8) e2 Ne γ1ξ−1 ωst ω0 γ2 ν . Pω = √ c ων 3 π(2ν − ξ + 1)Γ 2 − 12 (ν + 2)2 (ν + 3) (4.46) Thus, power-law spectrum of relativistic electrons with a cut-oﬀ at the energy E = mc2 γ2 produces diﬀusive synchrotron radiation at high frequencies, whose spectrum shape is deﬁned by the small-scale ﬁeld spectrum. Remarkably, the corresponding ﬂattening in the synchrotron cut-oﬀ region has recently been detected in the optical-UV range for the jet in the quasar 3C273 [8], which would imply the presence of a relatively strong small-scale ﬁeld there in agreement with the model of Honda and Honda [6]. Although formally the spectrum (4.46) is valid at arbitrarily high frequencies, there is actually a cut-oﬀ related to the minimal scale of the random ﬁeld lmin . Accordingly, the largest frequency of the diﬀusive synchrotron radiation is about ωmax ∼ (c/lmin )γ22 .

Soft Electron Spectrum Let us now turn to the case of suﬃciently soft electron spectra, ξ > 2ν + 1, which are typical, e.g., in many solar ﬂares. The contribution of synchrotron radiation is described by the standard expression, Pω ∝ ω −α , α = (ξ − 1)/2, which is steeper for soft spectra than the spectrum of diﬀusive synchrotron radiation, Pω ∝ ω −ν . Hence, for soft electron spectra, diﬀusive synchrotron radiation can dominate even at ω < ωc⊥ γ22 . The spectrum of diﬀusive synchrotron radiation has the same shape as before but its level is deﬁned by lower-energy electron contribution: 2 ν−1 2ν+1 (ξ − 1)Γ ν2 (ν 2 + 7ν + 8) e2 Ne γ12ν ωst ω0 Pω = √ . (4.47) ν 1 c ων 3 π(ξ − 2ν − 1)Γ 2 − 2 (ν + 2)2 (ν + 3) At low frequencies, ω ωpe γ1 , the radiation is still speciﬁed by expression (4.44), see the corresponding curves on the right in Fig. 4.3. One may note that in the case of soft electron spectra, the emission produced by the electron ensemble is similar to the emission from a mono-energetic electron distribution

100

G.D. Fleishman

with γ = γ1 , which is the main diﬀerence between the cases of hard and soft electron spectra. Let us estimate the ratio of the diﬀusive synchrotron radiation intensity to the synchrotron radiation intensity. For simplicity, we neglect factors of the order of unity, assume ω0 = ωce and γ1 ∼ 1, and introduce frequency 2 /ωce (ω ≡ (ω/ω∗ )ω∗ ), where synchrotron radiation has a peak, then ω∗ = ωpe Psh ω2 ∼ st 2 Psyn ωce

ωpe ωce

(

ω ω∗

ξ−2ν−1 .

(4.48)

Evidently, this ratio increases with frequency, so that diﬀusive synchrotron radiation can become dominant well before the frequency reaches ωc⊥ γ22 . Moreover, in the case of dense plasma, ωpe ωce , diﬀusive synchrotron radiation can dominate at all frequencies under the condition 2 ωst 2 ωce

ωpe ωce

ξ−2ν−1 >1,

(4.49)

even if the random ﬁeld is small compared with the regular ﬁeld ωst ωce . On top of this, the radiation spectrum produced from the dense plasma depends critically on the highest energy of the accelerated electrons. Indeed, if γ2 ωpe /ωce (e.g., γ2 = 4 in Fig. 4.4), then the radiation spectrum is entirely 2 /B02 = 10−4 in set up by the small-scale ﬁeld, in spite of its smallness ( Bst Fig. 4.4). Evidently, the standard synchrotron emission increases and becomes observable as far as γ2 increases.

Fig. 4.4. Left: Same as in Fig. 4.3, right, for less dense plasma, ωce /ωpe = 3 · 10−2 . The contribution from the uniform ﬁeld (synchrotron radiation) decreases for softer electron spectra (i.e., as ξ increases). Right: Emissivity by fast electron ensemble with (ξ = 6) from dense plasma (ωce /ωpe = 3 · 10−2 ) in the presence of weak magnetic 2 /B02 = 10−4 for diﬀerent high-energy cut-oﬀ values γ2 . When inhomogeneities Bst γ2 is small enough, the uniform magnetic ﬁeld does not aﬀect the radiation spectrum

4 Generation of Emissions by Fast Particles in Stochastic Media

101

Diﬀusive Synchrotron Radiation from Solar Radio Bursts? According to microwave and hard X-ray observations of solar ﬂares, the energetic spectra of accelerated electrons are frequently rather soft [9, 15]. Consequently, the diﬀusive synchrotron radiation can dominate the microwave emission for dense enough radio sources. Nevertheless, as a rule the microwave emission from solar ﬂares meets reasonable quantitative interpretation as synchrotron (gyrosynchrotron) radiation by moderately relativistic electrons (moving in non-uniform magnetic ﬁeld of the coronal loop). An example of microwave burst produced by gyrosynchrotron emission is given in Fig. 4.5, left.

Fig. 4.5. Two microwave bursts recorded by Owens Valley Solar Array in the range 1–18 GHz with 40 spectral channels and 4 sec temporal resolution. The ﬁrst one (left bottom) displays evident spectral hardening with time, while the second one shows remarkable constancy of the high-frequency spectral slope (courtesy of D.E. Gary)

Note, that the high-frequency spectral index δ (the radio ﬂux is ﬁtted by a power-law, F ∝ f δ , at high frequencies, Fig. 4.5, bottom left) decreases in value with time. Such spectral evolution typical for solar microwave bursts is well-understood in the context of the energy-dependent life time of electrons against the Coulomb collisions. Indeed, higher energy electrons have longer life times, which results in spectral hardening of the trapped electron population [Melrose and Brown, 12], and, respectively, hardening of the produced gyrosynchrotron radiation as observed by Melnikov and Magun [11]. However, if microwave emission is produced by diﬀusive synchrotron radiation as fast electrons interact with small-scale magnetic (and/or electric)

102

G.D. Fleishman

ﬁelds, then the radio spectrum is speciﬁed by the spectrum of random ﬁelds rather then of fast electrons. Thus, no spectral evolution (related to electron distribution modiﬁcation) is expected. Indeed, there is a minority of solar microwave bursts, which do not show any spectral evolution (e.g., no spectral hardening). An example of such a burst, demonstrating constancy in time of the high-frequency spectral index, is shown in Fig. 4.5 at the right. Curiously, the spectral index is δ = −1.5 to − 1.7 in agreement with standard models [Vainshtein et al., 22] and measurements of the turbulence spectra, e.g., in interplanetary [see, e.g., Toptygin, 20] and interstellar [Cordes et al., 1] space. Although it has not been ﬁrmly proven so far, such microwave bursts are possibly produced by diﬀusive synchrotron radiation mechanism. Since (to be dominant) this mechanism requires relatively dense plasma at the source site and soft spectra of accelerated electrons, the observational evidence can be found from analysis of simultaneous observations of soft and hard X-ray emissions from the same ﬂares.

4.4 Discussion The analysis presented demonstrates the potential importance of small-scale turbulence in the generation of radio emission from natural plasmas. This emission, being reliably detected and interpreted, provides the most direct measurements of small-scale turbulence in the remote sources. The diﬀusive synchrotron radiation is only one of the observable eﬀects of the turbulence on the radio emission. Indeed, the presence of density inhomogeneities aﬀects the properties of bremsstrahlung, because the Fourier transform of the square of the electric potential produced by background charges in a medium depends on spatial distribution of the charges through the double sum: * ) |∆N |2q 2 −iq(RA −RB ) 3 = N + (2π) , (4.50) e | ϕq0 ,q | ∝ V A,B

where RA and RB are the radius-vectors of the particles A and B, respectively, |∆N |2q is the spectrum of the inhomogeneity, and V is the volume of the system. In statistically uniform media the positions of various particles are uncorrelated and this double sum equals the total number of particles N . However, the macroscopic inhomogeneities make the positions correlated, so the double sum deviates from N . The second term in (4.50) gives rise to coherent bremsstrahlung, which in a certain spectral range dominates the incoherent bremsstrahlung [Platonov et al., 17]. Another important radiation process in the turbulent plasma is transition radiation arising when fast particles interact with small-scale density inhomogeneities of the background plasma (see Platonov and Fleishman [18], and references therein), whose potential importance for ionospheric conditions has been pointed out long ago by Yermakova and Trakhtengerts [23] (see also the

4 Generation of Emissions by Fast Particles in Stochastic Media

103

discussion in LaBelle and Treumann [10]). This emission process, giving rise to enhanced low-frequency (at frequencies lower than the accompanying synchrotron emission) continuum radio emission, has recently been reliably conﬁrmed in a subclass of two-component solar radio bursts [3, 16]. This ﬁnding is of particular importance for diagnostics of the number density, the level of small-scale turbulence, and the dynamics of low-energy fast particles in solar ﬂares. In addition, the turbulence can also aﬀect the coherent emissions from unstable electron populations [Fleishman et al., 2], e.g., providing strong broadening (or splitting) of the spectral peaks generated by electron cyclotron maser (ECM) emission. The typical bandwidth of the broadened ECM peaks and its distributions are found to be quantitatively consistent with those observed for narrowband solar radio spikes [Fleishman et al., 2].

Acknowledgements The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. This work was supported in part by the Russian Foundation for Basic Research, grants No. 03-02-17218, 04-02-39029. I am very grateful to T. S. Bastian for his numerous comments on this paper.

References [1] Cordes, J.M., M. Ryan, J.M. Weisberg, D.A. Frail, and S.R. Spangler: Nature 354, 121 (1991). [2] Fleishman, G.D.: Astrophys. J. 601, 559 (2004). [3] Fleishman, G.D., G.M. Nita, and D.E. Gary: Astrophys. J. 620, 506 (2005). [4] Fleishman, G.D.: Astro-ph/0502245 (2005). [5] Ginzburg, V.L. and S.I. Syrovatsky: Ann. Rev. Astron. Astrophys. 3, 297 (1965). [6] Honda, M. and Y.S. Honda: Astrophys. J. 617, L37 (2004). [7] Jaroschek, C. H., H. Lesch, and R. A. Treumann: Astrophys. J. 618, 822 (2005). [8] Jester, S., H.-J. R¨ oser, K. Meisenheimer, and R. Perley: Astron. Astrophys. 431, 477 (2005). [9] Kundu, M.R., S.M. White, N. Gopalswamy, and J. Lim: Astrophys. J. Suppl. 90, 599 (1994). [10] LaBelle, J. and R. A. Treumann: Space Sci. Rev. 101, 295 (2002). [11] Melnikov, V.F. and A. Magun: Solar Phys. 178, 153 (1998). [12] Melrose, D.B. and J.C. Brown: Monthly Notic. Royal Astron. Soc. London 176, 15 (1976). [13] Migdal, A.B.: Dokl. Akad. Nauk SSSR (in Russian) 96, 49 (1954); Phys. Rev. 103, 1811 (1956). [14] Nikolaev, Yu.A. and V.N. Tsytovich: Phys. Scripta 20, 665 (1979).

104

G.D. Fleishman

[15] Nita, G.M., D.E. Gary, and J. Lee: Astrophys. J. 605, 528 (2004). [16] Nita, G.M., D.E. Gary, and G.D. Fleishman: Astrophys. J. 629, L65 (2005). [17] Platonov, K.Yu., I.N. Toptygin, and G.D. Fleishman: Uspekhi Fiz. Nauk (in Russian) 160, 59 (Engl. transl.: Sov. Phys. Uspekhi, 33, 289) (1990). [18] Platonov, K.Yu. and G.D. Fleishman: Physics-Uspekhi 45, 235 (2002). [19] Razin, V. A.: Radioﬁzika 3, 584 (1960). [20] Toptygin, I.N.: Cosmic rays in interplanetary magnetic ﬁelds (DordrechtHolland, D. Reidel, 1985) 387 pp. [21] Toptygin, I.N. and G.D. Fleishman: Astroph. Space Sci. 13, 213 (1987). [22] Vainshtein, S.I., A.M. Bykov, and I.N. Toptygin: Turbulence, Current Sheets, and Shocks in Cosmic Plasma (The Fluid Mechanics of Astrophysics and Geophysics, Vol. 6, Langhorne: Gordon and Breach Science Publ., 1993) 398 pp. [23] Yermakova, E.N. and V.Yu. Trakhtengerts: Geomagn. Aeron. (Engl. Transl.) 21, 56 (1981).

5 Auroral Acceleration and Radiation R. Pottelette1 and R.A. Treumann2,3 1

2

3

CETP/CNRS, 4 av. de Neptune, 94107 St. Maur des Foss´es Cedex, France [email protected] Ludwig-Maximilians Universit¨ at M¨ unchen, Sektion Geophysik, Theresienstr. 37-41, 80333 M¨ unchen, Germany, [email protected] Department of Physics and Astronomy, Dartmouth College, Hanover, New Hampshire 03755, USA

Abstract. A brief review is given of the recent achievements in understanding the connection between processes in the generation of auroral acceleration and processes taking place at the tailward reconnection site. It is shown that most of the acceleration in the aurora is due to local ﬁeld-aligned electric potentials which are located in vertically narrow double layers along the magnetic ﬁeld of the order of ∼10 km and which are the site of preferential excitation of phase space holes of kilometer size extension along the magnetic ﬁeld which by themselves sometimes represent local potential drops and accelerate electrons and ions antiparallel to each other such that the energy modulation of the electron and ion energy ﬂuxes are in antiphase. Auroral kilometric radiation observations suggest that these structures may be the elementary radiation sources which build up the entire spectrum of the auroral kilometric radiation. Leaving open the very generation mechanism of the double layer whether produced by locally applied shear ﬂows as recently suggested in the literature (and reviewed here as well) we argue that the ﬁeld aligned current generator responsible for the production of the initial auroral current system is non-local but is related to reconnection in the tail. The ﬁeld aligned currents are interpreted as the closure currents required to close the recently observed electron Hall current system in the ion diﬀusion region at the tail reconnection site. Such a model is very attractive as it does not need any other secondary current disruption mechanism. Coupling to the ionosphere may be provided by kinetic Alfv´en waves emanating from the Hall reconnection region as surface waves and generating local shear ﬂow when focussing close to the ionosphere and transforming into shear-kinetic Alfv´en waves. A main problem still remains in how the decoupling of the two hemispheres observed in the aurora is produced at reconnection site. Multiple reconnection would be one possible solution.

Key words: Auroral processes, auroral particle acceleration, double layers, auroral kilometric radiation, elementary radiation sources, electron holes, reconnection, generator region

R. Pottelette and R.A. Treumann: Auroral Acceleration and Radiation, Lect. Notes Phys. 687, 105–138 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

106

R. Pottelette and R.A. Treumann

5.1 Introduction For many years it has been a mystery why electrons and ions are accelerated in the auroral zone. The idea of St¨ ormer [40] that cosmic ray particles spiral in along Earth’s magnetic ﬁeld to excite the atoms and ions of the upper atmosphere and ionosphere and generate the aurora has readily been proven wrong. Cosmic rays are too energetic to cause anything like an aurora. For them the Earth’s magnetic ﬁeld is practically invisible. They impinge on the ionosphere isotropically causing showers of secondary elementary particles and become absorbed in the atmosphere at altitudes way below those of the aurora. Solar energetic particles on the other hand are not energetic enough to reach anywhere deep enough into the ionosphere. According to their magnetic hardness they are excluded from the inner magnetosphere, while when ﬂowing in from their forbidden zone boundary they are still too energetic to cause substantial aurora. Moreover, they appear at too low latitudes to be associated with auroral phenomena. In addition, aurora is located most frequently on the antisolar side of the magnetosphere even though there is also aurora on the dayside under the cusp and driven by particles accelerated in magnetopause reconnection. These dayside aurorae complete the auroral oval. The energy range required for electrons to become eﬀective in generating aurorae lies between several 100 eV and several keV. The electron distribution in the magnetospheric tail, i.e. the plasma sheet, is at the lower end of this range. It thus can serve as the source for the auroral electron distribution. However, it must become additionally accelerated by some secondary acceleration process which up till now has only been badly identiﬁed. There is a high probability that this process does not consist of one unique step but that the acceleration of electrons takes place in a primary step followed by a sequence of a few or several secondary acceleration steps until the electrons have enough energy to play their role in the aurora. Presumably these two steps are located in spatially separate regions of the magnetosphere. Since aurorae occur during magnetic substorms and storms which are believed to be the result of violent reconnection going on in the magnetospheric tail current sheet, it is reasonable to assume that the ﬁrst step of acceleration is directly related to the tailward reconnection site. The second step, on the other hand, has been found to occur much closer to the ionosphere at altitudes of ∼2000–8000 km. Both steps act in concert but from the point of view of the aurora the second step is the more interesting one. It is also related to the generation of the famous auroral kilometric radiation which, under special plasma conditions, is emitted from the acceleration region. In the present paper we brieﬂy review the processes in the second step acceleration region adding some founded speculations on the processes in the ﬁrst step acceleration region.

5 Auroral Acceleration and Radiation

107

5.2 Morphology of the Auroral Acceleration Region During the past decade various spacecraft have passed the auroral zone at diﬀerent altitudes. The lowest ﬂying spacecraft was the Swedish satellite Freja (at a nominal altitude of ∼2000 km), followed by the NASA satellite FAST (at nominal altitude ∼4000 km) and the Swedish satellite Viking (at ∼8000 km altitude). Also, Geotail and Polar have passed magnetic ﬁeld lines connected to the auroral zone at much larger distances from Earth thus providing occasional information about auroral region-magnetosphere connections. According to these observations we may divide the auroral-magnetosphere connection into two diﬀerent regions: the lower magnetospheric auroral region and the tail auroral region. We will ﬁrst discuss the former as the processes taking place there have recently been illuminated best. Traditionally this region is divided into a sequence of ﬁeld-aligned current elements: the upward current and the downward current regions, respectively. Since the auroral ﬁeld-aligned currents are carried mainly by electrons with the ions contributing only little, the upward current region is the region where electrons from the magnetosphere ﬂow down into the ionosphere. Traditionally this region is called the “inverted-V” a name adapted from the Λ-like shape of the electron energy ﬂuxes versus time – or space – during satellite crossings of the auroral upward current region. An example is shown in Fig. 5.1 for a relatively long-lasting “inverted-V” event which in the low time resolution looks rather like a step function with sharp onset in the electron energy ﬂux around a few keV at the low latitude boundary of the event followed by a long-lasting plateau and a correspondingly sharp dropout of the ﬂux at the high latitude boundary. In high time resolution this increase is rather gradual indicating that the potential drops on a somewhat longer spatial distance. The whole event lasts for roughly 30 s, a very short time which for a spacecraft velocity of ∼6 km s−1 corresponds to the short horizontal distance of only ∼180 km in the corresponding northern hemispherical auroral passage of FAST. We will return to this particular fact in Sect. 5.6 below. In fact, the length of the “inverted-V” event is marked even better by the ion energy ﬂuxes than by the electron ﬂuxes. This is very clearly seen in Fig. 5.1 where the sharp increase in the ion energy accompanies the changes in energetic electron energy ﬂux. Before entering into a discussion of the entire sequence of data during this short passage we note that this short “inverted-V” event is a section of a longer auroral disturbance shown in Fig. 5.2 in very low time resolution. This entire event lasts for ∼5 min, corresponding to a horizontal distance of ∼1800 km on the northern hemispherical auroral region in the altitude range between 4000 km and 3500 km. It covers a latitudinal range of roughly 5 degrees in invariant latitude and a longitudinal range of roughly 1 hour in magnetic local time (or 15 degrees) in the pre-midnight sector. Thus the entire event is located fully on the northern hemisphere. This is important to remember for our later discussion. However, the self-sustained short time section shown in Fig. 5.1 which is an “inverted V” on its own shows that the full event consists

108

R. Pottelette and R.A. Treumann

Fig. 5.1. An overview of data from FAST orbit 1843 during a passage of the region above an aurora on the northern hemisphere

of a sequence of several sub-events more or less well covered by the spacecraft orbit, each of them a separate “inverted V”. Inspection of the overall event thus suggests that it consists of a mixture of a sequence of more or less well developed upward and downward current regions, and in order to obtain a true impression of its nature it would be necessary to investigate its higher

5 Auroral Acceleration and Radiation

109

Fig. 5.2. The whole auroral disturbance on FAST orbit 1843 as seen in the electron ﬂux. Shown are the downward (upper panel), the perpendicular (second panel), the upward electron ﬂux (third panel), and pitch angle distributions selected for two electron energy ranges (fourth and ﬁfth panels) in very low time resolution overview (courtesy C. W. Carlson, UCB)

110

R. Pottelette and R.A. Treumann

time resolution in order to separate it into parts which physically represent regions of either upward or downward currents. With this idea in mind we return to Fig. 5.1. The ﬁrst panel in Fig. 5.1 shows the perpendicular component of the magnetic ﬁeld which is caused by ﬁeld aligned currents. The positive slope of the ﬁeld indicates downward currents corresponding to upward electron ﬂux, while negative slopes occur in upward currents. The former are narrow in time and space and highly variable, while the latter are less variable and broader in time and space, corresponding to the dilute downward energetic auroral electron beam emanating from somewhere in the magnetosphere. These electron ﬂuxes are shown in the ﬁfth panel. Combined with the pitch angle distribution in the sixth panel one indeed realizes that intense upward highly variable electron ﬂuxes correlate with the downward current region, and relatively stable energetic electron ﬂuxes correlate with the upward current region. In this particular event the downward electron energies are close to 10 keV, while the upward electron energies cover a broad energy range from 10 eV upward up to 10 keV. Figure 5.3 shows a schematic of the auroral current system inferred from a sequence of upward and downward electron ﬂuxes [Elphic et al., 11]. Such a ﬁgure suggests that the upward and downward currents form one closed current system. If one assumes that the generator of the currents is in the magnetosphere, a reasonable assumption, then the upward currents are the primary currents while the downward currents are the return currents which close the system. The connection between the two is given by ionospheric Pedersen currents ﬂowing perpendicular to the magnetic ﬁeld ﬂux tubes in the ionosphere. Depending on the direction of the perpendicular ionospheric electric ﬁeld these currents may ﬂow in the plane of the upward and downward electron ﬂuxes or deviate from it forming a three-dimensional current system. In fact the direction of the electric ﬁeld will be imposed by the mapping of the magnetospheric electric potential along the magnetic ﬁeld into the ionosphere which in the most general case will be 3d as the transport of the potentials is done by shear Alfv´en waves. Figure 5.3 thus is an idealization which assumes that everything happens in the plane containing the upward and downward currents. It is, however, clear that in order to maintain both upward and downward currents in a closed current system the electrons, which carry the current must have been accelerated both from the magnetosphere down into the ionosphere and from the ionosphere up into the magnetosphere in order to arrive at closed and divergence free currents. 5.2.1 Upward Current Region The upward current region or downward electron acceleration region is in our view the primary acceleration region. Ignoring for the moment the primary magnetospheric acceleration processes, the data suggest that it contains a warm energetic primary electron beam of a few keV energy and several 100 eV velocity spread or temperature. This beam is not necessarily Maxwellian.

5 Auroral Acceleration and Radiation

Subauroral Latitude

Auroral Zone

111

Polar Cap

FAST Orbit

ionospheric density

Fig. 5.3. Schematic of the auroral ﬁeld-aligned current system inferred from a path of FAST across a sequence of downward and upward electron ﬂuxes [Elphic et al., 11] (Reprinted with permission of the American Geophysical Union)

Figure 5.4 shows on its right a cut through the electron distribution function (for FAST orbit 1773) along the magnetic ﬁeld. The auroral beam sits on a broad plateau in the distribution function. The increase at low velocities (energies) is due to electrons 0 which is the condition for the presence of free energy and a “lifted” to higher energy level distribution function, and the particle resonance condition

Perpendicular velocity v

v (km/s)

10 5

0

los s 0 cone

-105 10 5

0 Parallel velocity v

=

=

0 v (km/s)

Fig. 5.18. Left: Measured [after Delory et al., 9, with permission of the American Geophysical Union] and Right: model horseshoe electron distribution in the auroral region. The measured distribution consists of the spread electron beam moving down along the ﬁeld line and a plateau which is produced by the quasilinear interaction of the distribution with the self-generated VLF waves. The loss cone on the left in both distributions is practically empty

5 Auroral Acceleration and Radiation

131

Fig. 5.19. Left: Schematics of the mechanism of generation of horseshoe distributions in a converging magnetic ﬁeld. The converging ﬁeld reﬂects the particles while the parallel electric potential accelerates the particles. These gain in this way perpendicular energy. The result is the distribution on the Right: which is a horseshoe distribution. The formal analytical treatment of this process has been given by Chiu and Schulz [8]

k vr /2π = f − fce /γr

(5.4)

where k is the parallel wave number, f the wave frequency, fce the local 1 electron cyclotron frequency, and γr = [1 − v 2 /c2 ]− 2 the relativistic energy factor. Clearly, for strictly perpendicular radiation k = 0, the resonance line becomes a circle in the (v , v⊥ )-plane which can be positioned along the nearly circular horseshoe distribution where the derivative in the above equation is positive and maximum. This is the simple idea of emission in the x-mode. A theory like this generates very broadband emission unless the distribution is deformed locally very strongly. Inspection of the high-resolution emission spectrum however demonstrates that strongest emission is generated in very narrow bands which by themselves drift on variable speed across the frequency-time spectrogram. Such situations are shown in Figs. 5.20 and 5.21 for two diﬀerent cases: when the narrow band emission is standing and when it is moving. Figure 5.21 in addition shows the observation of two such interacting narrow band AKR sources of high intensity. In this case the two move initially at diﬀerent velocities. When interacting, they obviously repel each other and move together at the same speed to higher frequencies downward along the magnetic ﬁeld. This can be interpreted as both emission regions being of same polarity and thus not mixing.

132

R. Pottelette and R.A. Treumann

Fig. 5.20. An example of a narrow emission band in AKR which in this case is stationary in space and frequency FAST orbit 1761 kHz 445

Frequency

440

435

430

1997-01-31/06:44:56.002 UT

425 0

2

4 Time (s)

6

Fig. 5.21. Left: Two narrow band AKR emission lines moving across the frequencytime spectrogram approach each other, collide and move together down along the magnetic ﬁeld at same speed. Right: The explanation of this case as two electric ﬁeld structures of similar amplitude and equal polarity approaching each other [adapted from Pottelette et al., 32] (Reprinted with permission of the American Geophysical Union)

Another very interesting observation can be made from these two ﬁgures. They show that the emission of AKR is not continuum but consists of the superposition of a large number of “elementary radiation events” (ERE’s) the emitted radiation of which superimposes to generate the apparently

5 Auroral Acceleration and Radiation

133

continuous AKR spectrum. The nature of these elementary radiators has not yet been determined. However, in light of the above discussion of the dynamics of the auroral plasma it seems quite natural to assume that the elementary radiators are the electron holes which are generated in the plasma and move on the background of the horseshoe distribution. If this is true then the same theory applies as before with two modiﬁcations: • the electron holes must generate steep gradients on the distribution in perpendicular velocity which implies that they must be bent in phase space; • the emission from each electron hole is then not necessarily perpendicular as this is no longer required by the resonance condition. It will, however, be inclined, and the strongest emission is detected when the spacecraft will be in the emission cone of the elementary radiators. The existence of strong emission lines which move together may suggest that many such radiators have become attracted to a certain region and move approximately together over relatively long times. This is possible when realizing that the electron holes are eﬀectively positive charges on the electron background which implies that they will interact in a certain way with the main electron component becoming attracted by the bulk of the electrons and thus trapped in the electron distribution while at the same time growing in amplitude as their absolute depth will be conserved. This can lead to a collection of holes in a more narrow space region which then acts as the radiation source. The questions related to these problems have not yet been solved and are subjects of ongoing research.

5.5 The Tail Acceleration Region We now come to the discussion of the ultimate source of the auroral electron beam. This discussion will be rather brief as there is no consensus about this hot problem yet. It seems, however, reasonable to assume that processes in the tail of the magnetosphere provide the primary energy source since aurora are most prominent during substorms which are directly related to reconnection in the tail current sheet. It is thus natural to consider the structure of the reconnection site and to speculate about the reconnection site being the ultimate energy source of aurorae. It has recently been shown by Øieroset et al. [30] that the reconnection diﬀusion site in the tail current sheet is the site of separate electron and ion dynamics. On the small scale of the ion diﬀusion length c/2πfpi the motion of electrons and ions becomes diﬀerent. The ions in the plasma carrying the tail current are eﬀectively non-magnetic on this scale while the electrons are still tied to the magnetic ﬁeld. This diﬀerence gives rise to the generation of Hall currents since the magnetized electrons follow the convective motion of the magnetic ﬁeld from the

134

R. Pottelette and R.A. Treumann

Fig. 5.22. Schematic of the earthward half of the ion diﬀusion region in the magnetospheric tail current sheet during reconnection and its relation to the auroral electron ﬂux regions

tail lobe into the tail current sheet. These Hall currents must close somewhere. However, since such a closure cannot be done locally, the only possibility for them to close is through ﬁeld-aligned currents. Inﬂow of electrons on the boundary far from the current sheet and outﬂow from the boundary in the current sheet are required. This is demonstrated in Fig. 5.22 which shows a schematic of the connection of a part of the ion diﬀusion region in the tail and the auroral electron ﬂuxes which have been measured during one particular path of FAST. In this ﬁgure the electron inﬂows and outﬂows in the ion diﬀusion region in reconnection are seen as large arrows, while the corresponding Hall current is indicated in oppositely directed lines. The electron speed is essentially the inward-convection and outward reconnection-jet speed, respectively, to both

5 Auroral Acceleration and Radiation

135

Fig. 5.23. The schematic connection between tail reconnection and the auroral region. Tail reconnection is shown as a simulation result for the magnetic ﬁeld xpoint region, when one dominant x-point has evolved and several secondary x-points are visible in the tail current sheet. In this case the current sheet is very thin, of the order of the ion skin depth, such that the entire current sheet is aﬀected by reconnection. The Hall coupling of the reconnection region to the auroral zone causes the coupling of the reconnection to the processes in the aurora. It is not yet clear in which way this really happens

sides of the diﬀusion region. The electron outﬂow is related to the downward auroral electron beam ﬂuxes, while the electron inﬂow is fed by or at least related to the ionospheric outﬂow of electrons in the downward current region. In this picture the downward currents are the Hall currents emanating from the near-Earth reconnection site in the tail. Figure 5.23 shows the speculative geometry how activity in the tail connects to the auroral ionosphere. This picture does not take into account fast particles generated during reconnection in the tail since these particles are only a fraction of the entire electron distribution. It also suppresses the real process of coupling between the reconnection site and the ionosphere. This is a process which depends on the real conditions, on the propagation speed of the currents and particles and how fast particles can be extracted from the ionosphere to feed the Hall current in the tail reconnection region. The coupling might be produced by kinetic Alfv´en waves which are generated in the Hall region with transverse scales just of the ion inertial scale. When these waves, which are surface waves, move down to the ionosphere they constitute a current pulse as well as a transverse electric ﬁeld pulse which may cause the required shear ﬂow potential at the topside ionosphere which, as has been discussed here, will drive a U-shaped potential in the ionosphere, causing the double layer and, depending on its polarity, extracting electrons out of the ionosphere. An important problem in this connection is that in all available models the magnetospheric tail reconnection site is so large that it should aﬀect both hemispheres in the same way. Hence aurorae should be symmetrical with

136

R. Pottelette and R.A. Treumann

respect to the equator if a coupling like the one proposed here will exist. This cannot be the case for the auroral processes. The observations of FAST and other spacecraft show that in most cases the closure is nearly local. At ﬁrst glance this suggests that the auroral ﬁeld-aligned current system should be local and should have little in common with a tail current system generated in reconnection. One might therefore argue that the above picture is incorrect as it requires a global closure of the magnetic ﬁeld-aligned currents in the ionosphere similar to those closure processes which have been predicted for decades in the literature [for a review of the various auroral ﬁeld-aligned current system generator mechanisms see Borovsky, 4]. Though this may sometimes be indeed the case, there are possibilities for small scale reconnection in the tail being restricted solely to one hemisphere and even to part of it when reconnection is multiple [as has recently been argued by Pottelette and Treumann, 34] or when, as has been observed recently with Cluster [as can be concluded from observations by Runov et al., 36, 37], the current sheet in the tail bifurcates into narrow current layers. The mechanism of such a bifurcation is not understood yet. It is probably related to the preference of the current layer to generate conditions which are in favor of reconnection to develop on a very fast time scale. For this to happen it is required that the current sheet is very thin. This, however, causes again problems with the topology of the magnetic ﬁeld which becomes very complicated. In principle bifurcation can become possible only when the initially two-dimensional magnetic ﬁeld and current structure develops into a three-dimensional conﬁguration. If this is the case, as most recent numerical simulations of reconnection have shown, then a model of the connection between tail reconnection in the near Earth tail and auroral processes can be developed which is restricted solely to small scale auroral phenomena taking place on one hemisphere only in the auroral region.

5.6 Conclusions We have given a brief overview of the current state of the art of our knowledge in the fundamental acceleration, radiation, and source processes in the nighttime auroral ﬁeld-aligned current system. After a quarter of a century in situ observations from S3-3 to Viking, Freja, Polar, FAST, Geotail and ﬁnally Cluster a stage has been reached on which we can ﬁrmly conclude that our understanding of the auroral processes has stepped up from a purely descriptive one to a semi-quantitative understanding of the dominant acceleration processes of charged particles, both electrons and ions, in aurorae, of the structure of the auroral current system that consists of upward and downward ﬁeld-aligned currents which close in the ionosphere through Pedersen currents on surprisingly small scales, on the radiation mechanism in auroral kilometric radiation, and on the generation of the parallel potential drops. We know by now that indeed such parallel potential drops are generated which cause

5 Auroral Acceleration and Radiation

137

tremendous density decreases in the ionosphere while being highly dynamic. Usually in an auroral acceleration region not only one such drop exists but several are present along the magnetic ﬁeld. Their longitudinal scale is of the order of not more than few 10 km. They thus comprise quasi-stationary but small-scale double layers containing potential drops of order 100 eV to ∼1 keV, in rare cases few keV. These double layers move along the magnetic ﬁeld and can interact with each other. They are the sources of electron holes, small scale structures in the electric ﬁeld and electron phase space distribution of enormous dynamics. These electron holes may themselves contain smaller potential drops which erase the large double layer potential and at the same time cause out of phase variability in the acceleration of electrons and ions. Moreover, these electron holes seem to be the very sources of the auroral kilometric radiation, serving as elementary radiation sources. Thus they are of enormous dynamical importance in the auroral processes and in analogous applications under astrophysical conditions. Nevertheless, a number of open questions still remain which in near future will have to be attacked by multi-spacecraft missions and numerical simulations of the auroral and magnetospheric processes. One of the most interesting of these problems is the relation of the auroral processes and the current generator processes in the geomagnetic tail. These are most probably related to reconnection in the tail current sheet under thinning conditions during substorms. A direct relation has been argued for in this review which is based on the realization that the reconnection Hall current system in the tail current sheet must be closed by ﬁeld aligned currents which connect down to the ionosphere. Many aspects of such a model are in agreement with observation. However, one basic property is still barely understood: this is the fact that aurorae are very obviously local phenomena on one hemisphere only. How this can happen when reconnection is the driving force has still to be clariﬁed. Since reconnection is the most probably generator one may thus ask how reconnection can be imagined being restricted to one hemisphere only. We have suggested ideas on a possible resolution of this puzzle only without going into more detail since models of this kind are still under evolution.

References [1] Bale, S. J., et al.: Astrophys. J. Lett. 575, L25 (2002). [2] Bernstein, I.B., J.M. Greene, and M.D. Kruskal, Phys. Rev. 108, 546 (1957). [3] Berthomier, M., R. Pottelette, and M. Malingre: J. Geophys. Res. 103, 4261 (1998). [4] Borovsky, J.E.: J. Geophys. Res. 98, 6101 (1993). [5] Carlson, C.W., et al.: Geophys. Res. Lett. 25, 2017 (1998). [6] Cattell, C., et al.: Geophys. Res. Lett. 26, 425 (1999). [7] Cattell, C., et al.: J. Geophys. Res. 110, A01211 (2005). [8] Chiu, L. and M. Schulz: J. Geophys. Res. 83, 629 (1978). [9] Delory, G.T., et al.: Geophys. Res. Lett. 25, 2069 (1998).

138 [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42]

R. Pottelette and R.A. Treumann Dupree, T.: Phys. Fluids 26, 2460 (1983). Elphic, R., et al.: Geophys. Res. Lett. 25, 2033 (1998) Ergun, R.E., et al.: Geophys. Res. Lett. 25, 2041 (1998a). Ergun, R.E., et al.: Geophys. Res. Lett. 25, 2025 (1998b). Goldman, M.V., M.M. Oppenheim, D.L. Newman: Geophys. Res. Lett. 13, 1821 (1999). Goldman, M.V., D.L. Newman, and R.E. Ergun: Nonlin. Proc. Geophys. 10, 37 (2003). Gray P., et al.: Geophys. Res. Lett. 17, 1609 (1990). Kindel, J.F. and C.F. Kennel: J. Geophys. Res. 76, 3055 (1971). LaBelle, J. and R.A. Treumann: Space Sci. Rev. 101, 295 (2002). Louarn, P.: this volume (2005). Lysak, R.L. and C.T. Dum: J. Geophys. Res. 88, 365 (1983). Marklund, G., T. Karlsson, and J. Clemmons: J. Geophys. Res. 102, 17509 (1997). Marklund, G., et al.: Nature 414, 724 (2001). McFadden, J.P., et al.: Geophys. Res. Lett. 25, 2045 (1998). McFadden, J.P., C.W. Carlson, and R.E. Ergun: J. Geophys. Res. 104, 14453 (1999). Muschietti, L., et al.: Geophys. Res. Lett. 26, 1093 (1999). Muschietti, L., et al.: Nonlin. Proc. Geophys. 9, 101 (2002). Newman, D.L., et al.: Phys. Rev. Lett. 87, 255001 (2001). Newman, D.L., M.V. Goldman, and R.E. Ergun: Phys. Plasmas 9, 2337 (2002). Newman, D.L., et al.: Comp. Phys. Comm. 164, 122 (2004). Øieroset, M., et al.: Nature 412, 414 (2001). Pickett, J.S., et al.: Ann. Geophys. 22, 2515 (2004). Pottelette, R., R. A. Treumann, and M. Berthomier: J. Geophys. Res. 106, 8465 (2001). Pottelette, R. and R. A. Treumann: Geophys. Res. Lett. 32, L12104, doi:10.1029/2005GL022547 (2005a). Pottelette, R. and R. A. Treumann: Geophys. Res. Lett. 32, submitted (2005b). Pottelette, R., R. A. Treumann, and E. Georgescu: Nonlin. Proc. Geophys. 11, 197 (2004). Runov, A., et al.: Geophys. Res. Lett. 30, 1036, doi: 10.1029/2002GL016136 (2003). Runov, A., et al.: Ann. Geophys. 22, 2535 (2004). Sagdeev, R.Z.: Rev. Mod. Phys. 51, 1 (1979). Singh, N., et al.: Nonlin. Proc. Geophys. 12, in press (2005) St¨ ormer C.: Ergebn. kosm. Physik 1, 1 (1931). Treumann, R. A. and W. Baumjohann: Advanced Space Plasma Physics (Imperical College Press, London, 1997). Wu, C.S. and L.C. Lee: Astrophys. J. 230,621 (1979).

Part II

High-Frequency Waves

6 The Inﬂuence of Plasma Density Irregularities on Whistler-Mode Wave Propagation V.S. Sonwalkar Department of Electrical and Computer Engineering, University of Alaska Fairbanks, Fairbanks, AK, 99775, USA [email protected] Abstract. Whistler mode (W-mode) waves are profoundly aﬀected by FieldAligned Density Irregularities (FAI) present in the magnetosphere. These irregularities, present in all parts of the magnetosphere, occur at scale lengths ranging from a few meters to several hundred kilometers and larger. Given the spatial sizes of FAI and typical wavelength of W-mode waves found in the magnetosphere, it is convenient to classify FAI into three broad categories: large scale FAI, large scale FAI of duct-type, and small scale FAI. We discuss experimental results and their interpretations which provide physical insight into the eﬀects of FAI on whistler (W) mode wave propagation. It appears that FAI, large or small scale, inﬂuence the propagation of every kind of W-mode waves originating on the ground or in space. There are two ways FAI can inﬂuence W-mode propagation. First, they provide W-mode waves accessibility to regions otherwise not reachable. This has made it possible for W-mode waves to probe remote regions of the magnetosphere, rendering them as a powerful remote sensing tool. Second, they modify the wave structure which may have important consequences for radiation belt dynamics via wave-particle interactions. We conclude with a discussion of outstanding questions that must be answered in order to determine the importance of FAI in the propagation of W-mode waves and on the overall dynamics of wave-particle interactions in the magnetosphere.

Key words: Field aligned irregularities, whistler propagation, ducting, plasma inhomogeneity, ionosphere

6.1 Introduction The whistler mode is a cold plasma wave mode with an upper cutoﬀ frequency at the plasma frequency (fpe ) or cyclotron frequency (fce ), whichever is lower. Waves propagating in whistler mode (W-mode) are found in all regions of the Earth’s magnetosphere. They are also found in the magnetospheres of other planets. These waves may originate in sources residing outside the magnetosphere, such as lighting or VLF transmitters, or they may originate within the magnetosphere as a result of resonant wave-particle interactions. W-mode V.S. Sonwalkar: The Inﬂuence of Plasma Density Irregularities on Whistler-Mode Wave Propagation, Lect. Notes Phys. 687, 141–191 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

142

V.S. Sonwalkar

waves have been detected on every spacecraft carrying a plasma wave receiver and at numerous ground stations [see, e.g., 48, 51, 54, 74, 107, 115]. W-mode waves are important because they inﬂuence the behavior of the magnetosphere and partly because they are used as experimental tools to investigate the upper atmosphere. W-mode waves and their interactions with energetic particles have been a subject of interest since the discovery of the radiation belts. These interactions establish high levels of ELF/VLF waves in the magnetosphere and play an important role in the acceleration, heating, transport, and loss of energetic particles in the magnetosphere via cyclotron and Landau resonances [see, e.g., 56, 67, 78, 104]. Propagation, including reﬂection, refraction, guiding, and scattering, determines the extent to which whistler mode energy can remain trapped in the magnetosphere and inﬂuence the nature of wave-particle interactions. Plasma density irregularities in turn determine to a large extent the nature of whistler mode propagation. The role of plasma density irregularities on whistler mode propagation and their eﬀects on the overall dynamics of the magnetosphere is the subject of this paper. The magnetosphere is a highly structured magnetoplasma containing ﬁeld aligned irregularities (FAI) ranging in size from meters to hundreds of kilometers in the cross-B0 direction [30, 39, and references therein] where B0 is the geomagnetic ﬁeld. These irregularities are present in all parts of the magnetosphere. Plasma density and density structures are believed to play an important role in many physical processes at low and high latitudes: wave particle interactions, mode conversion, particle acceleration and precipitation, including auroral precipitation, visible and radar aurora, and substorm activity [e.g., 66, 105, 113]. On the practical side, these irregularities are important because they contribute to the fading of high frequency trans-ionospheric signals and to the degradation of ground-satellite communication. Forecasting and speciﬁcation of these irregularities is a major component of space weather programs. Though the subject of this paper is to understand the role of FAI on whistler mode propagation, the knowledge of these irregularities themselves is incomplete and is a subject of ongoing research [e.g., 30, 41]. Moreover, a wealth of information about FAI has actually been obtained by studying their eﬀects on plasma wave propagation. Thus the research on W-mode waves, and in general on plasma waves, and plasma density irregularities are intimately tied to one another. Evidence that FAI play an important role in the propagation of W-mode waves has been accumulating since the 1950s and 1960s [e.g., 26, 58, 60, 65, 112, and references therein]. In these earlier studies, W-mode propagation in the presence of FAI was considered mainly to explain the occurrence rates of lightning-generated whistlers. In recent years it has been realized that FAI may play a role more important in the overall physics of the magnetosphere than initially envisioned. For example, lighting-generated whistlers refracting through large scale ionospheric horizontal density gradients in the topside ionosphere may lead to generation of plasmapheric hiss, believed to be

6 Density Irregularities and Whistler Propagation

143

primarily responsible for the slot region in the radiation belts [44, 79, 118]. On the other hand, experimental observations of burst electron precipitation by lightning generated whistlers, which propagate through FAI called ducts, indicate that discrete whistlers may play a signiﬁcant role in the loss rates for the radiation belts in the mid-to-low latitudes within the plasmasphere [20, 134]. The general subject of whistler mode wave generation, propagation, and their interaction with energetic particles is a broad subject and is discussed in various books [40, 54] and review papers [48, 51, 74, 115]. Plasma density irregularities also aﬀect waves propagating in free space R-X and L-O modes and in Z-mode, which are discussed elsewhere [e.g., 14, 15, 30, 31, 123]. We focus in this review paper on particular aspects of whistler mode propagation motivated by the following questions: • How is whistler mode propagation aﬀected by FAI of various scale sizes? • What is the signiﬁcance of FAI inﬂuence on W-mode propagation from the magnetosphere to the ground and vice versa? • What are the implications of W-mode wave refraction or scattering by FAI for the generation of new kinds of waves? • What are the implications for mode conversion? • What are the implications for wave-particle interaction? We shall ﬁnd that: • FAI are responsible for providing W-mode waves accessibility to various regions in the magnetosphere otherwise not reachable in a smooth magnetosphere; • FAI guide, reﬂect, and scatter W-mode waves and, in general, modify wave structure, which may have profound eﬀects on wave-particle interactions and the general dynamic equilibrium of the radiation belt particles. It is impossible in this brief review to provide a comprehensive review of the vast amount of literature available on these topics. Our approach is to provide physical insight into various ways FAI may aﬀect W-mode propagation with the help of qualitative explanations and a few illustrative examples. This chapter is organized as follows: Section 6.2 brieﬂy describes magnetospheric plasma and ﬁeld aligned irregularities; Section 6.3 provides a theoretical background on the propagation of whistler mode waves; Section 6.4 presents ground and spacecraft observations of whistler mode waves illustrating the role played by ﬁeld aligned irregularities in the propagation of these waves, and Sect. 6.5 follows with a discussion and concluding remarks.

6.2 Magnetospheric Plasma Distribution: Field Aligned Irregularities The Earth’s magnetosphere is described in many texts and monographs [e.g., 66, 69]. Here we brieﬂy review aspects of magnetospheric cold plasma

144

V.S. Sonwalkar

Fig. 6.1. Schematic showing in the noon-midnight meridian various regions and features of the magnetosphere. Also shown are the locations where whistler-mode waves of various types are observed

distribution important for understanding how various magnetospheric boundaries and density irregularities aﬀect the whistler mode propagation. Figure 6.1 shows schematically various regions of the magnetosphere in the noonmidnight meridian and the locations where whistler mode waves of various types are observed. Immediately surrounding the Earth is the non-conducting atmosphere roughly 60–80 km thick, transparent to the propagation of radio waves. The next layer above the neutral atmosphere is called the ionosphere which extends up to ∼1000 km, completely encircles the Earth, and then merges into the magnetosphere. The boundary between the ionosphere and the neutral atmosphere below it is called the Earth-ionosphere boundary and the region between this boundary and the Earth is called the Earth-ionosphere waveguide. The ionosphere is a conductive medium, in which a small fraction ( fce/2 c)

Fig. 6.3. The refractive index as a function of wave normal direction for (a) f < flhr , (b) flhr < f < fce /2, and (c) f > fce /2. The ray direction is perpendicular to the refractive index surface and in general is in the direction diﬀerent from the wave normal direction as illustrated in (a) and (b)

150

V.S. Sonwalkar

frequency increases in value from below fLHR to a value higher than fLHR . In a closed refractive index surface (f < fLHR ), shown in Fig. 6.3a, the refractive index values remain ﬁnite for all values of the wave normal angle θ. In an open refractive index surface, the refractive index tends to inﬁnity at a certain angle called the resonance angle, θRES ; the corresponding surface of revolution is called resonance cone. For frequencies fLHR < f < fce /2, as shown in Fig. 6.3b, the refractive index surface is concave downward for smaller values of θ and then concave upward for large values of θ. Refractive index surface undergoes another topological change as the frequency increases beyond fce /2 and becomes concave upward for all wave normal angles. These changes in refractive index surface topology have important consequences for wave propagation in the magnetosphere, as discussed in the next subsection. As Fig. 6.3b and Fig. 6.3c show, for f > fLHR , the refractive index n → ∞ as θ → θRES . At wave normal angle close to resonance cone angle, the refractive index becomes large and W-mode waves become short wavelength quasielectrostatic waves and may suﬀer Landau damping. For cold magnetospheric plasma with temperatures fLHR,max , where fLHR,max is the maximum lower hybrid frequency along the ray path, will propagate to the other hemisphere and will undergo total internal reﬂection at the Earth-ionosphere boundary. A ray such as B or B at a low frequency will undergo magnetospheric reﬂection. In general, W-mode waves injected from the ground will remain trapped in the magnetosphere. Also, rays can reach any given location by more than one path after undergoing magnetospheric reﬂections, such as the rays B and B arriving at the satellite as shown in the Fig. 6.4a. Figure 6.4b shows eﬀects of large scale and small scale FAI on the W-mode propagation. A large scale FAI, satisfying condition (1) and thus W.K.B., can bend the upgoing ray such that at a given location rays may arrive by more than one path, each starting at a slightly diﬀerent latitude (rays A and A in Fig. 6.4b). A W-mode wave (ray B in Fig. 6.4b) incident on small scale irregularity can scatter into W-mode waves of both small and large wave normal angles (rays collectively labelled C in Fig. 6.4b). FAI with scale sizes comparable or smaller than W-mode wavelengths can scatter electromagnetic W-mode waves into quasi-electrostatic W-mode propagating with wave normal angles close to resonance cone angle θRES . In general, when condition (2) is satisﬁed, W.K.B. fails and strong scattering of W-mode waves occurs. The excited quasi-electrostatic waves are cut oﬀ at fLHR and are a type of lower hybrid (LH) wave. A number of mechanisms, both linear and nonlinear, have been proposed for producing LH waves from W-mode waves [6, 7, 8, 45, 86, 133]. Figure 6.6 schematically shows the linear mode conversion mechanism ﬁrst proposed by Ngo and Bell [1985]. This mechanism is attractive because it is simple and because it explains most experimental observations better than others do, as shown by Bell and Ngo [7]. Ducted Propagation through the Magnetosphere Figure 6.4c shows W-mode propagation guided by a duct (ray B) and by the plasmapause (ray D). Some of the energy injected from the ground (rays A in Fig. 6.4c) can couple into whistler mode duct or other duct type FAI such as plasmapause and be guided to the other hemisphere with relatively small wave normal angles. At the other hemisphere, part of the energy can propagate down to the Earth and part may be reﬂected back into the magnetosphere, some of it getting coupled into the same duct. Because the refractive index surface goes through a topological change (see Fig. 6.4) at f /fce = 12 , conditions for ducting also change at that frequency [22, 54, 112]. Snell’s construction shows that the density gradients on both sides of a crest tend to rotate the wave normal toward the geomagnetic ﬁeld direction. For frequencies below fce /2, the refractive index surface is concave downward. This geometry requires a density enhancement (crest) for ducting. For frequencies above fce /2, the refractive index surface is concave upward. This geometry requires a density depletion (trough) for ducting. However,

6 Density Irregularities and Whistler Propagation

155

Fig. 6.6. A geometric description of linear mode conversion of whistler mode waves. An incident small wave-normal (θi1 < θR ) whistler mode wave (labelled INC WM in region 1) is partly reﬂected as a small wave normal angle wave (labelled REF WM in region 1) and as a quasi-electrostatic wave with a large wave-normal angle close to resonance cone (labelled REF ES in region 1) and partly transmitted as a small wave-normal whistler mode wave (labelled TRANS WM in region 2), and as a quasi-electrostatic wave with a large wave normal angle close to resonance cone (labelled TRANS ES in region 2). The linear mode conversion is possible because W-mode refractive index has multiple solutions for the parallel component (nzi ) of the refractive index in the fLHR ≤ f < fce range. Thus W-mode waves scattered by this mechanism exhibit a low frequency cutoﬀ at fLHR . The mechanism of mode conversion also works when the incident wave is at large wave normal angle as shown by dotted arrow (θi 1 ∼ θRES ) [adapted from Bell and Ngo, 7] (Reprinted with permission of American Geophysical Union)

since this mode of ducting requires f > fce /2, the wave frequency must be ∼0.5 MHz or higher in order to be ducted all the way down to the ionosphere. Thus this type of ducting does not apply to whistlers. We conclude that whistlers received on the ground require enhancement ducts and that ducting should be eﬀective up to one half of the minimum electron gyrofrequency along the path. Alternatively, a ducted whistler propagating along a certain geomagnetic ﬁeld line will show an upper cutoﬀ at half the gyrofrequency at a point where the ﬁeld line crosses the equator. Both ground and spacecraft observations of whistlers conﬁrm this upper frequency cutoﬀ [Carpenter et al., 28]. Whistler dispersion, time delay as a function of frequency, depends on both the path length and the electron density along the path. Several factors including duct L shell, width, and enhancement/depletion determine whether or not the duct will trap waves incident from below the ionosphere. Helliwell [54] has discussed in detail the trapping of rays in ducts as a function of density enhancement/depletion, the scale size (gradients), and the initial wave normal angle. The coupling of wave energy into and out of a duct may also depend on how low the duct extends downward in altitude [e.g., James, 65]. Using

156

V.S. Sonwalkar

numerical simulations, Bernhardt and Park [16] have estimated that ducts may extend down to 300 km altitude at night but usually terminate above 1800 km during the day. In summer, ducts terminate above 1000-km altitude at all local times. As a result of ducts ending above the topside ionosphere, large scale FAI may play an important role in coupling wave energy from the ground to the duct and vice versa [James, 65]. Depending on the conditions at the exit points of the duct a wave trapped in a duct can undergo reﬂections at the Earth-ionosphere boundary and then propagate back to the other hemisphere in the same duct. Such reﬂections lead to multiple traces of whistler, called echo train, each showing increasing dispersion. After reﬂection, a whistler may not be trapped back in the duct and can propagate subsequently in the non-ducted mode [Rastani et al., 93]. In the case of several ducts, lightning energy from the same lightning discharge may propagate along several ducts and can be detected on the ground station as a multi-component whistler. Ducts and duct-type large scale ﬁeld aligned density drop oﬀs in the magnetosphere can also aﬀect non-ducted propagation [Edgar, 37]. In such cases the ducts are not able to trap the waves, but can signiﬁcantly modify their ray paths such that signals such as MR whistlers show distinctive signatures. Equatorial anomaly and plasmapause are other examples of duct-type irregularities. Equatorial anomaly, in which electron concentration during the daytime is depleted near the equatorial ionosphere and is enhanced in two regions on either side of the equator, also acts like a duct [Sonwalkar, 115]. It can be shown that ducts need not have density gradients on both sides to guide the waves from hemisphere to hemisphere. Thus plasmapause, where the density decreases sharply with increasing radial distance, provides an excellent one-sided whistler duct [61]. Plasmapause can also trap nonducted W-mode waves in certain cyclic trajectories [Thorne et al., 131]. The ducts need not go from one hemisphere to another. They may start and terminate in the same hemisphere. Such ducts have been found at high latitude and have been used to explain the dispersion of impulsive auroral hiss [Siren, 109]. Propagation of W-Mode Waves in the Magnetosphere: Waves Originating in a Source in the Magnetosphere Whistler mode emissions generated near the magnetic equator including midlatitude hiss, discrete, quasi-periodic, periodic emissions, and chorus are also believed to be generated by gyroresonance instabilities, sometimes called whistler mode instabilities [Sonwalkar, 115, and references therein]. The energy sources in most of these cases are the energetic electrons (∼1–100 keV) trapped in the magnetosphere. In the case of chorus, the energy sources are 5–150 keV electrons injected in the dayside region beyond the plasmapause during the period of heightened magnetic activity. The W-mode emissions in the auroral region include auroral hiss which may be produced via Cherenkov mechanism by precipitating beams of electrons with energies greater than

6 Density Irregularities and Whistler Propagation

157

10 keV [LaBelle and Treumann, 74]. Recently, for the ﬁrst time, we have a manmade source of W-mode waves on-board the IMAGE satellite [94, 123]. The Radio Plasma Imager (RPI) on the IMAGE satellite is capable of radiating in W-mode during the low altitude portion of its orbit. The general principles of W-mode propagation for a source in the magnetosphere are similar to those discussed for the case of a ground source. The principal diﬀerence between these two types of sources is that the waves injected from the ground enter the ionosphere vertically, whereas the waves injected into the magnetosphere from space may start with a wide range of initial wave normal angles. Figure 6.7a schematically illustrates the propagation of whistler mode signals from a source location in the equatorial plane within the magnetosphere. Most of the rays, such as A and B, from a magnetospheric source cannot reach a ground receiver because they either undergo total internal reﬂection at the Earth-ionosphere boundary or they undergo LHR reﬂections in Propagation of Plasma Waves - Source of Waves in Space a)

b)

c)

B Large

A

Rx

Earth

C

B 15kHz

Small

Tx

C‘

C‘’ L=5 fLHR=1kHz

ducted propagation

CAH Generation 10 - 100 m Irregularties

Transmission Cone o o (5 - 10 )

PP L=4

Small

10 - 100 m Irregularties

1kHz

1kHz

B Large

IAH Generation

Transmission Cone o o (5 - 10 )

Earth

Earth Diffuse ZM Echo

d)

e)

B

Discrete Non-Ducted Multipath Echo

Earth

B

Diffuse WM Echo

Duct Width [~10 - 100 km]

1 - 10 km Scale Irregularities

Earth

B

f)

Discrete Ducted WM Echo

Earth

10 m - 1 km Scale Irregularity

Fig. 6.7. An illustration of various ray paths for whistler mode wave propagation in the magnetosphere from a source in the magnetosphere: (a) source at the equator, ducted and nonducted propagation; (b) source in the auroral region, nonducted propagation followed by scattering by small scale FAI and propagation to the ground; (c) source in the auroral region, ducted propagation followed by scattering by small scale FAI and propagation to the ground; (d) source on a low altitude satellite, reﬂection from the ionosphere leading to an echo; echoes arriving from multiple paths resulting from propagation through a large scale FAI; (e) source on a low altitude satellite, ducted echoes, (f) source on a low altitude satellite, W-mode and Z-mode diﬀuse echoes resulting from scattering by small scale FAI [adapted from Sonwalkar et al., 123] (Rerpinted with permission of American Geophysical Union)

158

V.S. Sonwalkar

the magnetosphere. However, in the presence of ducts, some of the rays, such as C, can be guided to the low altitude ionosphere with their wave normal at small angles with respect to the local vertical; these rays can be observed at a ground receiver after propagating within the Earth-ionosphere wave guide [Helliwell, 54]. The ducted signals that have backscattered from the ionosphere-waveguide boundary can propagate back into the duct or can propagate as nonducted signals [Rastani et al. 93]. In the topside ionosphere, FAI may refract or scatter the downcoming W-waves, which may play a crucial role in determining subsequent paths of these waves. The waves can also reach the ground directly from the source if the source is at low altitudes and radiates such that some of the wave energy reaches the Earth-ionosphere boundary with suﬃciently small wave normal angles with respect to the local vertical [Thompson and Dowden, 132]. Other modes of guiding energy from a magnetospheric source to the ground include propagation along the plasmapause [61] and a sub-protonospheric mode [92, 110]. Figures 6.7b and 6.7c show propagation from a source located on auroral ﬁeld lines and generating waves at large wave normal angles. It is believed that auroral hiss (AH) is generated in this manner. As shown in the ﬁgure, these W-mode waves can propagate down in ducted or nonducted mode and, in general, will be reﬂected back by magnetospheric or total internal reﬂection. However, if there are FAI present in the regions where these reﬂections occur, some of the large wave normal angle W-mode waves may be refracted or scattered into small wave normal angle waves which then can be observed on the ground. Figures 6.7d-f show various ways W-mode waves can propagate from a low altitude source, assumed here to be a transmitter on a satellite, towards the Earth-ionosphere boundary and be seen on the satellite as an echo or be seen on the ground. An echo can reach the satellite by single or multiple non-ducted paths (Fig. 6.7d), or propagate down to the Earth-ionosphere boundary in a duct and return back in the same duct (Fig. 6.7e), or be scattered by small scale FAI (Fig. 6.7f). The scattered waves generally show a range of time delay giving echo a diﬀuse appearance on a spectrogram.

6.4 Observations and Interpretations The literature describes numerous observations of W-mode waves, from ground stations and from low and high altitude spacecraft [e.g., 1, 48, 51, 54, 56, 74, 115]. We present here a few examples to illustrate the key features of W-mode propagation in the presence of ﬁeld-aligned irregularities. W-mode observations are frequently categorized as those found on the ground and those found on spacecraft. The reason for dividing ground and spacecraft observations into separate categories is that somewhat distinct and apparently uncorrelated activity is detected at ground stations and on satellite [Sonwalkar, 115]. However, in this paper, for the reasons discussed in the

6 Density Irregularities and Whistler Propagation

159

previous section, we classify observations of W-mode waves of magnetospheric origin into two main categories: • the observations of W-mode waves with sources on the ground, • the observations of W-mode waves with sources in the magnetosphere. 6.4.1 W-Mode Observations When the Source is Below the Ionosphere It is convenient to categorize W-mode observations into two subsections: (1) non-ducted propagation, (2) ducted propagation. Ducted propagation can be seen both on the spacecraft and on the ground, whereas non-ducted propagation can only be seen on the spacecraft. Historically ducted propagation was discovered ﬁrst (on the ground) and non-ducted later. It is, however, easier to start with observations of non-ducted signals as non-ducted propagation is the natural mode of propagation of W-mode waves in a smooth magnetosphere. Nonducted Propagation: Eﬀects of Large-scale FAI, Ducts, and Density Drop-oﬀs We begin with the simplest possible propagation scenario: propagation of a single frequency signal reaching a satellite as a plane wave in a smooth magnetosphere as illustrated by ray B in Fig. 6.4a. Figure 6.8a shows an example of near plane wave received on the DE 1 satellite. A 20-s-long, 4.0 kHz continuous wave Siple transmitter signal was received by the 200-m long electric antenna on the DE 1 satellite. The satellite was located outside the plasmasphere at L ∼ 4.5 and λm ∼ 7◦ N. The local electron density was about 15 el/cc corresponding to fpe ≈ 35 kHz and the local magnetic ﬁeld was about 0.0034 G corresponding to fce ≈ 9.5 kHz. The electric ﬁeld amplitude shows well deﬁned spin fading at half the spin period (6 s), indicating reception of a single plane wave. Reception of plane W-mode waves is relatively rare, but when detected, it can provide testing of some of the most fundamental ideas of W-mode propagation and antenna properties in space plasmas [114, 116]. The depth and the phase of spin fading pattern can be used to obtain the wave normal direction of about 55◦ with respect to geomagnetic ﬁeld. Ground transmitter signals received on the satellite often show fading patterns which cannot be explained as a result of satellite spin motion alone. In fact fading patterns have been observed on satellites that were not spin stabilized. Heyborne [59] and Scarabucci [102] observed amplitude fading at low altitude (∼1000 km) OGO 1, OGO 2, and OGO 4 satellites. Cerisier [32]

160

V.S. Sonwalkar

Fig. 6.8. Nonducted whistler mode propagation of Siple transmitter signals: (a) An example of near plane wave received on the DE 1 satellite. A 20-s-long, 4.0 kHz continuous wave Siple transmitter signal was received by the 200-m long electric antenna on the DE 1 satellite. (b) An example of two plane waves received on the ISEE 1 satellite. A 20-s-long, 4.0 kHz continuous wave Siple transmitter signal was received by the 215-m long electric antenna on the ISEE 1. (c) Fourier transform of the wave amplitude shown in (b) [adapted from Sonwalkar et al. 114, 120] (Reprinted with permission of American Geophysical Union)

observed two Doppler shifts in the FUB signal observed on the FR 1 (altitude ∼750 km). Neubert et al. [85] and Sonwalkar et al. [120] have observed pulse elongations and amplitude modulations in Omega and Siple transmitter pulses received on the GEOS 1 and ISEE 1 satellites, respectively. These

6 Density Irregularities and Whistler Propagation

161

authors have interpreted their observations as indicating the presence of two or more plane waves arriving from diﬀerent directions. Figure 6.8b shows an example of a 20-s-long, 4.0 kHz continuous wave Siple transmitter signal received by the 215-m long electric antenna on the ISEE 1 satellite. The fading pattern now contains frequencies other than twice the spin frequency of the ISEE 1 satellite (spin period = 3 s). This fading pattern can be explained as that arising from the satellite receiving two plane waves arriving from two diﬀerent directions. The Doppler frequency for each plane wave is diﬀerent, and thus the satellite receiver measures the beat pattern resulting from this diﬀerence. It can be easily shown that the number of spectral components in the mean square received voltage is given by (3N 2 −3N +4)/2, where N is the number of multiple paths [Sonwalkar and Inan, 114]. Figure 6.8c shows the amplitude of the Fourier transform of the electric ﬁeld amplitude envelope shown in Fig. 6.8b. We clearly notice peaks at ﬁve frequencies, which are identiﬁed as the d.c., diﬀerential doppler shift (∆ωd ), twice the spin frequency (ωs ), and the sum and diﬀerence of diﬀerential Doppler shift with twice the spin frequency. A detailed analysis of Siple transmitter signals received on the ISEE 1 satellite showed that, in general, at any point in the magnetosphere the direct signals (before reﬂections) transmitted from the ground arrive, within few hundred milliseconds of each other, along two or more closely spaced multiple paths, as illustrated by rays A and A in Fig. 6.4b . The multiple paths can be explained by assuming propagation through 1–10 km cross-B scale size FAI in the topside ionosphere with a few percent enhancement or depletion in the plasma density. Such FAI can refract the wave normal direction by ∼5–10◦ , giving rise to multiple path propagation as illustrated in Fig. 6.4b [Sonwalkar et al., 120]. The diﬀerential Doppler shift ∆f resulting from the observation of multipath W-mode ground signals on a satellite is of the order of ∆nVs f /c, where ∆n is the diﬀerence in the refractive index values of two closely spaced multiple paths, Vs is the satellite speed and c the velocity of light in vacuum. Assuming ∆n ∼ 1, f ∼ 4 kHz, Vs ∼ 8 km, we obtain ∆f ∼ 0.1 Hz, a fraction of Hz. Clearly to resolve such multiple paths we need either long duration ﬁxed frequency signals, such as the one discussed above, or signals originating in impulsive (duration ∼ ms) lightning discharges which can be resolved in time domain when they arrive at the satellite by more than one path, or we need measurements of multiple spacecraft separated by distances of the order of 100 km (e.g. Cluster). It is possible to misinterpret or have more than one interpretation of wave normal direction measurements of a short duration (∼1 s) ground transmitter signals received on a satellite. For example, Lefeuvre et al. [76] determined wave normal directions of seven Omega transmitter pulses observed on the GEOS 1 satellite. They found a 0.2- to 0.4-s periodicity in the observed wave normal directions, which Sonwalkar et al. [120] interpreted as an indication of two closely spaced multiple paths. Whistlers propagating in the non-ducted mode also show evidence of propagation along single direct paths in a smooth magnetosphere and that of

162

V.S. Sonwalkar

Fig. 6.9. Non-ducted whistler mode propagation of whistlers : (a) Spectrogram of a multi-component MR whistler received on OGO 1 satellite via in a smooth magnetosphere, (b) propagation in the presence of FAI [adapted from Smith and Angerami, 111] (Reprinted with permission of American Geophysical Union)

propagation along multiple paths in a magnetosphere containing large scale FAI. Figure 6.9a shows the spectrogram of a magnetospherically reﬂected or MR whistler received on OGO 1 satellite [Smith and Angerami, 111]. The MR whistler usually consists of a series of traces or components, each exhibiting a frequency of minimum travel time or nose frequency. The multi-component feature of MR whistlers can be explained in terms of propagation of lightning energy in relatively smooth magnetosphere by non-ducted paths which have undergone multiple MR reﬂections as illustrated by rays B and B in Fig. 6.4a [Kimura, 68]. MR whistlers, also called unducted or non-ducted whistlers, show a wide variety of frequency-time signatures, depending on the location of the receiver (spacecraft) with respect to the causative lightning discharge,

6 Density Irregularities and Whistler Propagation

163

and on the density distribution of electrons and ions in the magnetosphere. By the nature of their propagation mode, these unducted whistlers cannot be detected on ground. The spacecraft observations of unducted whistlers have led to a new understanding of the many subtle features of whistler-mode propagation and to the deduction of important plasma parameters in space [17, 35, 36, 37, 38, 111, 118, 129]. Edgar [37] has shown that ray tracing simulations in a smooth magnetosphere can explain the observed spectrogram shown in Fig. 6.9a. In general, MR spectrogram contain features that cannot be explained by propagation in a smooth magnetosphere. For example Fig. 6.9b shows the spectrogram of MR whistler observed on 18 March 1965 on OGO 1 satellite. This whistler has a spectral signature quite diﬀerent than that of the MR whistler shown in Fig. 6.9a. The traces O+ , 2− , and 2+ are expected for an MR whistler propagating in a smooth magnetosphere for the given location of the satellite and the hemisphere of the lightning. Smith and Angerami [111] speculated that extra traces, labelled 2∗− and 2∗+ , probably resulted from multi-path propagation, possibly due to large-scale irregularities in the magnetosphere. Using ray tracing simulations, Edgar [37] showed that the extra traces 2∗− and 2∗+ could be explained by assuming multi-path propagation resulting from a 30% density drop oﬀ at L ∼ 1.8. He also demonstrated that the distortion in the 2− , and 2+ components near 6 kHz can be explained by considering a duct with ∼ 20% enhancement in density near L = 2.4. In general Edgar [37, 38] and more recently Bortnik et al. [17] have shown that large and small scale FAI in the topside ionosphere and density drop oﬀs and ducts in the magnetosphere can explain speciﬁc observed features in MR whistler spectrograms. Some features such as the frequency cutoﬀ of MR whistlers may be explained in terms of their trapping in density gradients, Landau damping, and D-region absorption [17, 38, 131]. It is evident that the same mechanism, viz. bending of rays due to the presence of large scale FAI has been used to explain multi-path propagation of the ground signals (Fig. 6.8b) as well as multi-path propagation of MR whistlers (Fig. 6.9b). This bending permits more than one ray paths of similar type, such as two direct ray paths or two or more MR paths of similar type (e.g. 2− and 2∗− ), to reach the satellite for a given location in the magnetosphere. It is clear that this type of multi-path propagation (Rays A and A in Fig. 6.4b), resulting from the presence of large scale FAI in the magnetosphere or topside ionosphere, is diﬀerent from the one resulting from the multiple magnetospheric reﬂections in the smooth magnetosphere (Rays B and B in Fig. 6.4a). Non-Ducted Propagation: Eﬀects of Small-Scale FAI Small scale irregularities satisfying condition (2) can scatter W-mode waves into directions which may be drastically diﬀerent than their initial wave normal directions. For W-mode, with typical wavelengths of the order of a few km,

164

V.S. Sonwalkar

the condition (2) can be easily satisﬁed by small scale FAI with LF AI ∼10– 100 m and ∆Ne /Ne ∼ a few percent. The scattered waves are generally Wmode waves with both small and large wave normal angles. The large wave normal angle waves are quasi-electrostatic and show a lower cutoﬀ at fLHR and are therefore generally call lower hybrid (LH) waves. Earliest evidence, albeit indirect, of inﬂuence of small scale FAI on Wmode propagation came from a strong correlation of whistler activity at midlatitude with HF backscatter [Carpenter and Colin, 26] and the correlation of auroral hiss activity with HF backscatter [Hower and Gluth, 60]. HF waves are scattered by FAI with scale lengths comparable to HF-wavelength, a few tens of meters. A dramatic evidence of scattering of W-mode waves by small scale FAI came when the “spectral broadening phenomenon” was discovered for ground transmitter signals received on the low altitude satellites [Bell et al., 11]. Figure 6.10 shows an example of Omega transmitter signals received on the DE 1 satellite. The pulses of ∼ 1 s duration from the Omega transmitter in North Dakota are present at 10.2 kHz, 11.05 kHz, and 11.333 kHz. Spectrally broadened Omega transmitter pulses are seen both in Fig. 6.10a, showing a spectrogram of the electric ﬁeld measured by a 200 m long dipole antenna, and in Fig. 6.10b, showing spectrogram of the electric ﬁeld measured by a short 8 m long dipole antenna. The spectrally broadened pulses are direct pulses which have propagated from the ground along a ∼2000 km path directly to the spacecraft. Following the direct pulses are echoes of the direct pulses reﬂected from the conjugate hemisphere with a delay of ∼2 s. In both panels the apparent bandwidth of the direct pulses is larger than 100 Hz, ∼100 times the 1 Hz bandwidth of the transmitted pulses.

Fig. 6.10. Observations of lower hybrid waves excited by signals from the Omega (ND) transmitter. (a) and (b) respectively show reception by the Ex and Ez antennas. The Ex antenna consists of two 100 m wires deployed in the spin plane. The Ez antenna consists of two 4 m tubes deployed along spin axis [adapted from Bell et al., 10] (Reprinted with permission of American Geophysical Union)

6 Density Irregularities and Whistler Propagation

165

The LH waves can also be excited by non-ducted whistlers. Figure 11a shows an example of the excitation of LH waves above local fLHR by a series of non-ducted whistlers observed on the ISIS-2 satellite. Note that LH waves are excited for f ≥ fLHR , consistent with generation by linear mode conversion method (Fig. 6.6). Observations from many satellites and theoretical analysis show that the excited waves are short wavelength (5 m < λ < 100 m) electrostatic lower hybrid (LH) waves excited by electromagnetic whistler mode waves through scattering from magnetic ﬁeld-aligned irregularities located in the topside ionosphere and magnetosphere [e.g., Bell et al., 12, and references therein]. These LH waves exhibit very large Doppler shifts which cause the observed spectral broadening (∼300–1000 Hz) of the signal as received at the moving satellite [7, 10, 11, 62]. Earlier observations of this phenomenon occurred at altitudes ≤7000 km [Bell and Ngo, 6]. Recently, Bell et al. [12] report new observations from the Cluster spacecraft of strong excitation of lower hybrid (LH) waves by electromagnetic (EM) whistler mode waves at altitudes ≈20,000 km outside the plasmasphere. These observations provide strong evidence that EM whistler mode waves are continuously transformed into LH waves as the whistler mode waves propagate at high altitudes beyond L ∼ 4. This process may represent the major propagation loss for EM whistler mode waves in these regions, and may explain the lack of lightning generated whistlers observed outside the plasmapause [Platino et al., 90]. The excited LH waves represent a plasma wave population which can resonate with energetic ring current protons to produce pitch angle scattering on magnetic shells beyond L∼4. Thus linear mode coupling provides a new mechanism by which lightning generated whistler mode waves can aﬀect the lifetimes of energetic ring current protons. We discuss another interesting manifestation of the eﬀects of FAI, perhaps both small and large scale, on the propagation of whistlers. Figure 6.11b shows an example of whistler-triggered hiss emissions. Using data from the DE 1 satellite, Sonwalkar and Inan [118] showed that lightning generated whistlers

Fig. 6.11. Examples of eﬀects of large and small scale FAI on non-ducted whistler propagation. (a) ISIS-2 satellite observations of lower hybrid waves near 10 kHz excited by whistlers. (b) DE 1 satellite observations of hiss band excited by whistler (courtesy of VLF group, Stanford University)

166

V.S. Sonwalkar

often trigger hiss emissions that endure up to 10- to 20-s periods and suggested that the lightning may be an embryonic source of plasmaspheric hiss. They recognized that direct multiple path propagation resulting from FAI irregularities may play an important role in the generation of whistler triggered hiss. Subsequently using ray tracing simulations Draganov et al. [35, 36] showed that the combined contribution from whistler rays produced by a single lightning ﬂash entering the magnetosphere at a range of latitudes and with a range of wave normal angles can form a continuous hiss-like signal. It was assumed that at 400 km altitude the initial wave normal angles of the rays are spread around the nominal vertical orientation. The spread in the wave normal angle was presumed to be generated as a result of W-mode propagation through large and small scale FAI in the topside ionosphere. It may be noted that H. C. Koons ﬁrst suggested the idea in 1984 that plasmaspheric hiss might simply be the accumulated waves of many non-ducted whistlers [See p. 19,486 of 126]. Recently, Green et al. [44] found that the longitudinal distribution of the hiss intensity (excluding the enhancement at the equator) is similar to the distribution of lightning: stronger over continents than over the ocean, stronger in the summer than in the winter, and stronger on the dayside than on the nightside. These observations strongly support lightning as the dominant source for plasmaspheric hiss, which, through particle-wave interactions, maintains the slot region in the radiation belts. Ducted Propagation: Observations on the Ground The importance and role of ducts in the propagation of W-mode waves cannot be overemphasized. As discussed in Sect. 6.3, without ducts W-mode waves would not be observable on the ground. The earliest observations of W-mode waves were ducted whistlers observed on the ground. Helliwell [54] in his now classic monograph has given an early history of W-mode wave observations. He gives examples of many types of W-mode waves including lightning-generated whistlers, ground transmitter signals, and VLF emissions of magnetospheric origins. Helliwell et al. [58] were the ﬁrst to suggest that whistlers might become trapped within ﬁeld-aligned density irregularities (ducts) and propagate from one hemisphere to another. The duct hypothesis was prompted by the observation that most whistlers observed on the ground show several discrete cases with varying frequency-time dispersion indicative of propagation over several distinct paths. Figure 6.12 shows examples of ducted propagation of Siple transmitter and lightning-generated whistlers, observed on the ground at Lake Mistissini, Canada and Palmer Station, Antarctica. Figure 6.12a describes the essential idea behind the active experiments performed from Siple Station, Antarctica, during the 1973–1998 period. These experiments provided new results and insight on the eﬀects of cold and hot plasma on W-mode propagation [25, 56, and references therein]. Figure 6.12b shows a 1-s pulse from Siple Station received at Lake Mistassini after propagation along two ducts in which wave

6 Density Irregularities and Whistler Propagation

167

Fig. 6.12. Examples of ducted W-mode propagation observed on the ground: (a) Schematic showing basics of ﬁeld-aligned whistler mode wave injection experiments between Siple Station, Antarctica (SI), and Lake Mistissini, Quebec, Canada (LM). (b) A 1-s pulse (lower right) from SI is received at LM (upper right) after propagation along two ducts in which wave growth and triggering of rising emissions occurs. A VLF receiver located at Palmer (PA) provides data on low L shell whistler mode paths and on subionospheric propagation from Siple Station. (c) Spectrogram of a multi-component ducted whistler received at Palmer Station, Antarctica, and the associated triggered emissions [adapted from Helliwell, 56] (Reprinted with permission of American Geophysical Union)

growth and triggering of rising emissions occur. Emission triggering is very common for ducted signals but rarely observed for non-ducted signals. Figure 6.12c shows an example of multi-component whistler and triggered emissions observed at Palmer station, Antarctica. The spheric originating from the same lightning discharge that caused the whistler was also detected and is marked by an arrow. Spatial and temporal occurrence patterns of ducted and non-ducted whistlers have been a subject of great interest [Helliwell, 54]. Many aspects of the occurrence rates, such as the local time, season, and locations could be related to the occurrence rates of causative lightning, and propagation conditions such as high absorption in the daytime ionosphere and the latitudinal dependance of whistler mode transmission cone. The occurrence of lightning activity reduces at high latitude and the transmission cone is very small at low latitudes. Consequently the whistler activity peaks at nighttime at mid-latitude (∼50◦ ). Many studies show that large and small scale FAI in the topside ionosphere and plasmapause regions play an important role in the propagation and

168

V.S. Sonwalkar

occurrence of ducted whistlers [see, e.g., 26, 27, 65, and references therein]. Carpenter and Colin [26] found that day-to-day variations of W-mode waves at mid-latitude are strongly correlated with the occurrence of small scale FAI in the F-region as observed by HF backscatter. They interpreted these F-region FAI as being the feet of the large scale FAI – whistler ducts. Using wave normal direction data from FR-1 satellite (altitude = 750 km) and ray tracing, James [65], showed that in addition to factors considered above, large scale FAI in the topside ionosphere inﬂuence the rate of occurrence of ˘ c [27] showed that ionospheric processes such as whistlers. Carpenter and Suli´ wave damping, defocusing within mid-latitude trough, focusing within large scale (50–100 km) FAI, and scattering by 10 to 100-m FAI can explain many features of the observed spatial distributions of whistler paths beyond the plasmapause and also the relatively low amplitude of the associated whistlers as compared to that of whistlers found inside the plasmasphere. They also conﬁrmed that ducted whistler propagation outside the plasmasphere occurs at comparatively low rates in comparison to activity within, and tend to occur at certain distances from the plasmapause. W-mode propagation can also be guided by ﬁeld aligned density irregularities other than ducts, such as the plasmapause. Carpenter [23] gives examples of whistler and VLF noises propagating just outside the plasmasphere. These whistlers and emissions exhibited features not observed in usual ducted whistlers, including extension of signal frequencies into the range 0.5−0.8fHeq , where fHeq is the equatorial gyro-frequency of the path, and the echoing or repeated propagation over the path at frequencies above 0.5fHeq . The VLF noise bands and bursts tended to occur within the frequency range 0.4−0.8fHeq . This kind of propagation is not well understood. Using ray tracing, Inan and Bell [61] have shown that rays can be guided along the plasmapause outer surface, but only for the waves leaving with wave normal angles close to relatively large Gendrin angle. Inan and Bell [61] suggest that large scale FAI with strong density gradients can tilt the vertical wave normal angles into Gendrin angle. Cairo and Cerisier [21] have shown that signiﬁcant departures of upgoing wave normals from the vertical are observed within the plasmasphere, apparently due to large scale FAI. There are many aspects of ducted whistler mode propagation that are not well understood. For example, analysis of ground whistler data reveals that a single whistler can contain several hyperﬁne structures [Hamar et al., 49]. These ﬁner structures in whistlers can be interpreted either in terms of electron density ﬂuctuations of the order of ∼ 1% and spatial scale sizes of the order of 50 km, or in terms of excitation of multiple propagating modes within a duct [Hamar et al., 50]. It is possible that the ﬁner details of wave particle interactions that lead to a variety of triggered emissions are related to the details of the ducted propagation.

6 Density Irregularities and Whistler Propagation

169

Ducted Propagation: Observations on the Spacecraft Ducted W-mode waves are rarely observed on satellites in the magnetosphere due to the apparently small volume (0.01%) occupied by the ducts [Burgess and Inan, 20], but are most commonly observed on the ground after they have exited from a duct and propagated in the Earth-ionosphere wave guide to a receiving site [Helliwell, 54]. While the duct hypothesis has been extensively used in interpreting ground observations of VLF waves, only a limited set of experimental data are available that provides for the determination of the properties of magnetospheric ducts based on in situ observations [2, 24, 71, 90, 103, 111, 122]. Detection of ducts using spacecraft observations was based in most cases on measurements of whistler dispersion [2, 24, 90, 103, 111]. In one case it was based on enhancements of whistler mode hiss correlated with density enhancements [Koons, 71]. In one case discussed below, wave normal direction measurement was used to establish ducted propagation and electron density measurement was used to determine duct parameters. These authors provided the following measurements of ducts: ∼ 400 km near L = 3 by Smith and Angerami [111], 223–430 km and 6–22% for L = 4.1 − 4.7 by Angerami [2], 68–850 km and 10–40 % for L = 3.1 − 3.5 by Scarf and Chappell [103], 630– 1260 km and ≤30% between L = 4 and L = 5 by Carpenter et al. [29], and ∼500 km and ∼40% by Koons [71]. Observations listed above, excepting those by Platino et al. [90], were all made in the plasmasphere. Recently, using data from the Cluster spacecraft, Platino et al. [90] have shown that whistlers observed outside the plasmasphere in the low density regions occur only in the presence of large scale irregularities within which the waves are “ducted”. They found that dispersion characteristics of observed whistlers are only matched by ray tracing simulations if the whistlers are ducted. They also propose a possible explanation why whistlers outside the plasmasphere are rarely observed, based on wave conversion from whistler mode to lower hybrid mode [Bell et al., 12]. A direct conﬁrmation of many features of both ducted and non-ducted propagation was obtained in a study involving measurement of the VLF transmitter (15 kHz, 4s ON/4s OFF, 800 kW radiated power) at Khabarovsk, Russia (48◦ N, 135◦ E; λm ∼ 38◦ , φm ∼ −158◦ ) received on the low altitude COSMOS 1809 satellite Sonwalkar et al. [Sonwalkar et al., 121]. Figure 6.13 shows an example of signals from the VLF transmitter at Khabarovsk, Russia (48◦ N, 135◦ E; λm ∼ 38◦ , φm ∼ −158◦ ) received on the low altitude COSMOS 1809 satellite (near-circular orbit at an altitude of ∼960 km and inclination of ∼82.5◦ ). Figure 6.13a top and bottom panel display the DE 1 and COSMOS 1809 satellite orbits in the magnetic meridional and magnetic equatorial planes, respectively, when Khabarovsk signals were observed on the two satellites on 23 Aug 89. Simultaneous observations of ground transmitter signals and electron density on COSMOS 1809 provided unique opportunity to

170

V.S. Sonwalkar (b)

(a)

Fig. 6.13. (a) Orbits of DE 1 and COSMOS 1809 satellites in magnetic meridional and equatorial planes on 23 Aug 1989 during the period when the Khabarovsk transmitter was operating. The thickened parts of the orbits indicate the time period during which the transmitter signals were detected on the satellites. (b) Amplitude of electric (top panel) and magnetic ﬁelds (second panel) of Khabarovsk signals on COSMOS 1809 satellites for 23 Aug 89. Also shown are the mean square root variation in the electron density dNe (third panel) and electron density Ne (fourth panel) [adapted from Sonwalkar et al., 122] (Reprinted with permission of American Geophysical Union)

measure the parameters of density irregularities responsible for direct multiple path propagation and for ducted propagation. Figure 6.13b shows COSMOS 1809 observations of Khabarovsk signals in the 15 kHz-channel of the spectrum analyzer when the satellite passed over the transmitter at ∼1600 UT and when the satellite passed over a region roughly magnetically conjugate to the transmitter location in the southern hemisphere at ∼1622 UT. The ﬁgure presents electric ﬁeld Ey , magnetic ﬁeld Bx , mean square root variation in the electron density dNe , and the total electron density Ne along the satellite orbit. Since the satellite altitude during

6 Density Irregularities and Whistler Propagation

171

this interval was approximately constant, dNe is a measure of horizontal gradients in the ionospheric electron density. The top two panels of Fig. 6.13b show the amplitude fading of Khabarovsk transmitter signals due to multiple path propagation eﬀects and the third panel shows the ﬂuctuations ∆Ne in the electron density believed to be responsible for the multiple path propagation. The typical horizontal extent of these irregularities was ≤100 km and the density ﬂuctuations were of the order of ≤5%, consistent with previous theoretical estimates obtained using wave observations and ray tracing simulations [Sonwalkar et al., 120]. When COSMOS 1809 was in the southern hemisphere, the magnetic ﬁeld of the transmitter signal showed a large value near 1625 UT. The value of the ratio (cBx )/Ey gives an indication of the wave normal angle of the wave. Based on ray tracing simulations, for non-ducted wave propagation of ground transmitter signals injected in the northern hemisphere, a large wave normal angle corresponding to a lower value of the ratio (cBx )/Ey is to be expected in the southern hemisphere. From Fig. 6.14b second and third panel, we ﬁnd this to be the case in general except for a signal observed between 1624:57 and 1625:14 UT when rather large values of Bx and (cBx )/Ey were measured indicating low wave normal angles (< 30◦ ). Sonwalkar et al. [122] interpreted this low wave normal angle signal to be a case of ducted propagation. To further test this assumption, the electron density data were examined. It was found that between 1625:04 and 1625:12 there was a 3000 to 3800 el/cc enhancement over the background electron density of 29500 el/cc, consistent with the duct hypothesis (other variations in electron density seen in Fig. 6.14b third panel are presumed to be local ﬁeld aligned irregularities which are commonly present in the ionosphere). Further, the equatorial gyrofrequency corresponding to L =2.85, the duct location, is 34.4 kHz. Thus the 15 kHz Khabarovsk signal is well below the half gyrofrequency cutoﬀ of fHeq /2 = 17.2 kHz for ducted propagation [54]. The L shell thickness of the duct was 0.06 and corresponds to a duct cross section of 55 km at 981 km altitude and 367 km at the geomagnetic equator. The density enhancement was 10-13% over the background electron density of 29500 el/cc. Only a lower limit of ∆φ ≥ 0.2◦ can be placed on the longitudinal extent of the duct, consistent with the 3-4◦ of width in longitude estimated by Angerami [2]. The eﬃciency of signal coupling between the Earth-ionosphere waveguide and magnetospheric ducts is of importance to studies of VLF propagation and wave ampliﬁcation in the magnetosphere. This coupling is critically dependent on the duct end points in the ionosphere [16]. The above study indicates that duct end points can extend down to at least ∼980 km at ∼0130 local time in the southern hemisphere during austral winter. This result is consistent with theoretical studies of Bernhardt and Park [16] who predict a duct endpoint as low as 300 km at night during winter and equinoxes. The ducted signals were observed over an L shell range of 0.13, about twice the L shell width of the duct. This could be the result of ducted signals leaking from the duct as previously noted in both experimental and theoretical works

172

V.S. Sonwalkar

Fig. 6.14. (a) DE 2 observations of impulsive signals near the geomagnetic equator in the 512- to 1024-Hz channel of the VEFI instrument. (b) Langmuir probe observations of electron density illustrating equatorial (Appleton) anomaly. (c) Ray tracings through a horizontally stratiﬁed ionosphere. Trajectories of rays (750 Hz frequency) injected at 200-km altitudes with vertical wave normal angles. (d) Ray tracings in the presence of an equatorial anomaly. Trajectories of rays (750 Hz frequency) injected at 200-km altitudes with vertical wave normal angles [adapted from Sonwalkar et al., 121] (Reprinted with permission of American Geophysical Union)

[2, 127, 128]. The peak electric and magnetic ﬁeld were detected inside the duct near the duct center. Both the peak electric (>520 µV/m) and magnetic ﬁeld (36 pT) intensities of the ducted signals were comparable to those observed for the non-ducted signals over the transmitter in the northern hemisphere, but were about 20 dB higher than those of the non-ducted signals observed in the southern hemisphere in the vicinity of the duct, consistent with observations of Koons [71]. Inside the duct electric and magnetic ﬁeld intensities show a ﬁne structure, consistent with recent reports of whistler ﬁne structure by Hamar et al. [49, 50]]. Ducted whistlers are rarely observed below L = 2. An unusual observation of lightning energy reaching the low altitude equatorial regions was made by Sonwalkar et al. [Sonwalkar et al., 121]. They show that equatorial anomaly can focus W-mode waves from lightning into a region near the equator. The equatorial anomaly is a tropical ionospheric eﬀect arising from equatorial electrodynamics, which essentially leads to the formation of a ductlike structure with enhanced density along a ﬁeld line near ±20◦ geomagnetic latitude [66]. The Vector Electric Field Instrument (VEFI) on the Dynamic Explorer 2 (DE 2) satellite observed impulsive ELF/VLF electric ﬁeld bursts on almost every crossing of the geomagnetic equator in the evening hours (Fig. 6.14a). These signals were interpreted as originating in lightning discharges. These signals that peak in intensity near the magnetic equator were observed within

6 Density Irregularities and Whistler Propagation

173

5-20◦ latitude of the geomagnetic equator at altitudes of 300–500 km with amplitudes of the order of ∼ millivolts/m in the 512-1024 Hz frequency band of the VEFI instrument. The signals are observed in the same region near the equator where the equatorial anomaly, as detected by Langmuir probe, was found (Fig. 6.14b). Whistler-mode ELF/VLF wave propagation through a horizontally stratiﬁed ionosphere predicts strong attenuation of sub-ionospheric signals reaching the equator at low altitudes. However, ray tracing analysis showed (Figs. 6.14c and 6.14d) that the presence of the equatorial density anomaly, commonly observed in the upper ionosphere during evening hours, leads to the focusing of the wave energy from lightning near the geomagnetic equator at low altitudes, thus accounting for all observed aspects of the observed phenomenon. The observations presented here indicate that during certain hours in the evening, almost all the energy input from lightning discharges entering the ionosphere at < 30◦ latitude remains conﬁned to a small region (in altitude and latitude) near the geomagnetic equator. The net wideband electric ﬁeld, extrapolated from the observed electric ﬁeld values in the 512–1024 Hz band, can be ∼10 mV/m or higher. These strong electric ﬁelds generated in the ionosphere by lightning at local evening times may be important for the equatorial electrodynamics of the ionosphere. 6.4.2 W-Mode Observations When the Source is in the Magnetosphere As shown in Fig. 6.1, a variety of W-mode waves, collectively called VLF emissions, are observed in the magnetosphere. All of them, excepting VLF transmitter signals and lightning generated whistlers, have apparent origins in the magnetosphere. A general discussion of these naturally occurring VLF emissions, beyond the scope of this paper, can be found in several review papers on the subject [1, 48, 53, 54, 74, 101, 115]. This paper brieﬂy reviews those aspects of the propagation of VLF emissions which are aﬀected by the presence of FAI of all types, including plasmapause. The generation and propagation of the VLF emissions is closely related to the properties of the cold and hot plasma distributions found in the regions where these emissions are generated. W-Mode Observations when the Source is Near the Equator: Ducted Propagation It is believed that plasmaspheric hiss, mid-latitude hiss, chorus, periodic and quasi-periodic emissions, and triggered emissions are generated in the equatorial region of the magnetosphere via some kind of gyroresonance related instability [e.g., 1, 115, and references therein]. The energy sources in most of these cases are the energetic electrons (∼1–100 keV) trapped in the magnetosphere. Figure 6.15 shows dynamic spectra illustrating ground observations

174

V.S. Sonwalkar (b)

(a) dB kHz SIPLE STATON 0 4.0

02 MAY 86

17 20 UT

EIGHTS STATON

16 OCT 63

3.0 -10 2.0

2.0 -20 -30

1.0 0

5

10 Time (Sec)

ref 52 dB ∆ f 31 Hz

0

5

10

15

ref 67 dB ∆ f 15 Hz

Time (Sec)

Fig. 6.15. (a) Double banded ELF/VLF chorus observed at Siple, Antarctica (λg = 76◦ S, φg = 84◦ W, λm = 60◦ S). In this case chorus bands are accompanied by weak bands of midlatitude hiss. (b) Periodic emissions observed at Eights, Antarctica (λg = 75◦ S, φg = 77◦ W, λm = 64◦ S) (courtesy of VLF group, Stanford University)

of chorus (Fig. 6.15a, discrete rising tones), mid-latitude hiss (Fig. 6.15a, diffuse bands between 500–1500 Hz and 2–3 kHz), and periodic emissions (Fig. 6.15b). Excepting plasmaspheric hiss, VLF emissions generated near the equator are observed on the ground. As shown in Fig. 6.7a, rays A and B, in general W-mode waves generated near the equator will be reﬂected back into the magnetosphere by either MR reﬂection or by total internal reﬂection. The fact that we routinely observe these emissions on the ground indicates that these emissions are somehow trapped inside one of the nearby ducts and brought down to the ground, as illustrated by rays C and C in Fig. 6.7. As an example of ducted propagation of VLF emissions with the source in the magnetosphere, consider periodic emissions. Figure 6.15b shows a spectrogram of periodic emissions recorded at Eights, Antarctica. A sequence of discrete emissions or clusters of discrete emissions showing regular spacing is called periodic emissions [54, 101]. Usually their period is constant, but on occasion it changes slowly. They generally occur at frequencies below 5–10 kHz with a few kHz bandwidth and a period in the range from two seconds to six seconds. It was established that the period of the emissions is identical to the two-hop whistler transit time [Dowden, 33]. Observations of periodic emissions at conjugate stations Byrd-Hudson Bay showed that the emissions appeared at the two stations with a delay of about 0.8 s [Lokken et al., 77]. Periodic emissions are observed at latitudes that correspond to closed magnetic ﬁeld lines and are rarely observed on satellites [Sazhin et al., 100]. These characteristics indicate that the periodic emissions, like other VLF emissions observed on the ground, have propagated in ﬁeld aligned ducts. The fact that they are rarely observed on the satellite hints that probably they are also generated at small wave normal angles within ducts and thus remain trapped inside the ducts. Thus a satellite has to be inside a duct to record these emissions, the probability of which is very small as discussed earlier.

6 Density Irregularities and Whistler Propagation

175

Chorus is routinely observed both on the ground and on spacecraft and many features, though not all, of chorus observed on the ground and spacecraft are similar [e.g., 53, 101, 115, and references therein]. This indicates that chorus may be propagating in both ducted and non-ducted modes. Recent analysis of chorus observed on the POLAR [Lauben et al., 75] and Cluster spacecraft [Santolik et al., 95, 96] support this viewpoint. Lauben et al. [75] ﬁnd that the upper-band chorus waves (f ≥ 0.5fHeq , where fHeq is the equatorial gyrofrequency) are emitted with wave normal θs 0, where θs is the wave normal angle at the source region near the equator, while the lower-band chorus waves ( f ≤ 0.5fHeq ) are emitted with θs θG , where θG is the Gendrin angle, giving minimum value of refractive index parallel to B0 . For both frequency bands, these respective θs values lead to wave propagation paths which remain naturally parallel to the static magnetic ﬁeld in the source region over a latitude range of typically 3◦ −5◦ , providing an ample opportunity for cumulative wave/particle interaction and thus rapid wave growth, notably in the absence of ﬁeld-aligned cold plasma density enhancements (i.e., ducts). Chorus generated at low wave normal angle can easily be trapped in a nearby duct and brought down to the Earth. Chorus at wave normal angles near θG can be guided to the Earth by the plasmapause [Inan and Bell, 61]. Mid-latitude hiss is frequently accompanied by whistlers echoing in the same path (duct) and ampliﬁed in the same frequency band. Dowden [33] suggested that at least part of the midlatitude hiss may be generated by the superposition of highly dispersed, unresolved, overlapping whistlers continuously echoing from hemisphere to hemisphere along a duct. However to overcome geometric loss of 10–20 dB corresponding to the small fraction of reﬂected energy being re-ducted, a magnetospheric ampliﬁcation of 10–20 dB is required to maintain steady hiss intensity over time scales of tens of minutes [Thompson and Dowden, 132]. On the other hand, ground based direction ﬁnding measurements suggest that midlatitude hiss is generated just inside the plasmapause [Hayakawa, 51]. There are a few examples in the literature where a more direct correlation between W-mode emissions and FAI is made. Koons [71], using data from VLF/MR swept frequency receiver on AMPTE IRM spacecraft, found strong enhancement of W-mode waves (identiﬁed as hiss or chorus) correlated with density enhancement in the outer plasmasphere, in the vicinity of plasmapause, between L = 4 and L = 6, which he identiﬁed as ducts. The ducts consisted of more than 40% density enhancement and had a typical half width of 250 km. The wave intensity inside the duct was an order of magnitude larger than that outside. On Cluster spacecraft Moullard et al. [83] found that electron density ﬂuctuations are regularly observed near the plasmapause together with hiss or chorus emissions at frequencies below the electron cyclotron frequency. Instruments on board Cluster spacecraft often observe two such emission bands with ﬂuctuating wave intensities that suggest wave ducting in density enhancements as well as troughs. During a plasmapause crossing on June 5, 2001 (near the geomagnetic equator, L = 4 − 6,

176

V.S. Sonwalkar

afternoon sector), density ﬂuctuations up to hundreds cm−3 were found while whistler mode waves were observed in two separate frequency bands, at 100– 500 Hz (correlated to the density ﬂuctuations) and 3–6 kHz (anti-correlated). Using the electron density and wave data from Cluster spacecraft, Masson et al. [80] found that in the vicinity of the plasmapause, around the geomagnetic equator, the four Cluster satellites often observe banded midlatitude hiss, typically 1–2 kHz bandwidth and center frequency between 2–10 kHz. They found that the location of occurrence of the hiss was strongly correlated with the position of the plasmapause with no MLT dependence. Plasmaspheric hiss is a broadband and structureless whistler-mode radiation that is almost always present in the Earth’s plasmasphere and is commonly observed by magnetospheric satellites, but not observed on the ground [53, 130]. It is observed in the frequency range extending from a few hundred Hz to 2–3 kHz with a peak below 1 kHz. Plasmaspheric hiss should be distinguished from auroral hiss which is observed at low altitudes in the auroral regions and covers a much wider frequency band (∼100 Hz – 100 kHz) and from mid-latitude hiss that is observed from ground stations and on satellites in 2–10 kHz range. Plasmaspheric hiss is found throughout the plasmasphere and is stronger in the daytime sector compared to the midnight-to-dawn sector, and generally peaks at high (>40◦ ) latitudes; it often shows a sharp cutoﬀ at the plasmapause for frequencies below ∼1 kHz, though it is also observed outside the plasmasphere at higher frequencies. W-mode waves generated in the magnetosphere cannot reach the ground (Fig. 6.7a, rays A and B). VLF emissions such as mid-latitude hiss and chorus, discussed above, are observed on the ground, presumably they are generated at low wave normal angles, and get trapped in a duct and thereby propagate down to the Earth. A question naturally arises: why is plasmaspheric hiss not observed on the ground? The answer to this question may come from studies devoted to determining the wave normal directions of plasmaspheric hiss [e.g., 117, 126, and references therein]. The general conclusion from these studies is that plasmapheric hiss most frequently propagates with large wave normal angles. As a result it cannot be trapped inside a duct and be seen on the ground. On the other hand, recent results from POLAR spacecraft show that plasmaspheric hiss often propagates at wave normals that make small angle with the geomagnetic ﬁeld [Santolik et al., 97]. It thus appears further work is warranted to understand why plasmaspheric hiss is not observed on the ground. It appears that FAI may play an important role in the generation of plasmaspheric hiss. In one generation mechanism, proposed by Thorne et al. [131], plasmapause plays a central role. In this mechanism, hiss propagates along certain cyclic trajectories, made possible by reﬂections from plasmapause. This allows hiss to reach the equatorial region repeatedly, each time with a small wave normal angle, thus permitting maximum possible cyclotron resonance interaction with energetic electrons. Alternatively, hiss may be generated by

6 Density Irregularities and Whistler Propagation

177

lightning-whistlers as discussed in Sect. 6.1.2 [35, 118]. In this mechanism, topside ionospheric FAI play a central role. W-Mode Observations when the Source is in the Auroral Region: Non-Ducted Propagation W-mode emissions commonly observed in the auroral and polar regions include auroral hiss, lower hybrid emissions, and Lion’s roar [74, 115, and references therein]. Auroral hiss (AH), shown in Fig. 6.16, is one of the most intense whistler mode plasma wave phenomenon observed both on the ground at high latitudes and on spacecraft in the auroral zone [e.g., 74, and references therein]. The apparent diﬀerence in the spectral form of continuous AH (Fig. 6.16a) and impulsive AH (Fig. 6.16b) has been attributed to diﬀerences in non-ducted versus ducted propagation [108, 109, 119]. The continuous structure-less spectra of continuous hiss can be explained in terms of non-ducted propagation and the impulsive, or sometimes falling tone spectra, as in Fig. 6.16b, of impulsive hiss in terms of ducted propagation. There has been some diﬃculty in understanding the propagation of AH from its source region to the ground and to a high altitude satellite. There is strong experimental evidence as well as theoretical analysis that indicates that AH is generated at large wave normal angle by Cerenkov resonance mechanism [74, 119]. The standard whistler mode propagation in a smooth magnetosphere, discussed in Sect. 6.3, predicts that auroral hiss generated at large wave normal angle along the auroral ﬁeld lines by Cerenkov resonance cannot penetrate to the ground. Two solutions for this problem have been suggested. Matsuo et al. [81] proposed that large scale FAI refract AH with θ ∼ θRES into AH with small wave normal angle which can fall into the transmission cone. Sonwalkar and Harikumar [119] argue that if the waves start with θ within a degree or so of the resonance cone angle, as the theory demands, then the relatively small tilts of horizontal gradients in the auroral ionosphere

Fig. 6.16. Examples of auroral hiss (AH) observed at South Pole, Antarctica (λg = 90◦ S, λm = 79◦ S). (a) Example of continuous auroral hiss (CAH). (b) Example of impulsive auroral hiss (IAH) showing dispersion [adapted from Sonwalkar and Harikumar, 119] (Reprinted with permission of American Geophysical Union)

178

V.S. Sonwalkar

are insuﬃcient to bend the wave normals enough for them to intercept the transmission cone. They propose that small scale FAI scatter AH propagating close to the resonance cone into small angle W-mode waves with small enough wave normal angles such that they fall into the transmission cone and thus propagate down to the Earth. This mechanism is shown schematically in Fig. 6.7f. There is some experimental evidence supporting this mechanism. Radar observations show a strong correlation between the occurrence of VLF hiss in the 1–10 kHz band and 18 MHz radar echoes from the F-region ﬁeldaligned irregularities [However and Gluth, 60]. Thirty two of the thirty three hiss events observed were associated with 18 MHz radar events. On occasion the hiss and radar events showed very close time correlation. Assuming that 18 MHz radar echoes result from scattering of radar signals from FAI with scale sizes equal to half the radar wavelength, the radar observations indicate presence of ∼15 m scale size FAI. Auroral cavity, a region of reduced electron density, may aﬀect propagation of AH upward from its source region. Snell’s law when applied to a sharp density boundary slanted with respect to the geomagnetic ﬁeld predicts that large wave normal angle W-mode waves can refract into small wave normal angle waves. Morgan et al. [82] suggest that through this mechanism, upward propagating auroral hiss converts from the short wavelengths ( m) observed at low altitudes to longer wavelengths (1–3 km) inferred at high altitudes by undergoing multiple reﬂections from the tilted sides of an auroral cavity which spatially diverges with increasing altitude. Satellite observations have often shown quasi-electrostatic plasma wave bands with a sharp cutoﬀ near the local lower hybrid resonance (fLHR ) frequency [4, 5, 18, 72, 73]. These bands are observed at mid- to high latitudes in the altitude range of ∼1000 km to a few thousand kilometers. These waves are often excited by whistlers, or they could simply be hiss emissions with a cutoﬀ at fLHR . The cold plasma theory predicts a resonance at the lower hybrid frequency for whistler mode waves propagating perpendicular to the static magnetic ﬁeld. At frequencies higher than the local LHR frequency, the resonance takes place at wave normal angles smaller than 90◦ . This explains the sharp lower cutoﬀ at fLHR as well as the quasi-electrostatic nature of the band [Brice and Smith, 18]. Several mechanisms including wave-particle resonances, parametric instabilities, and linear mode conversion have been proposed to explain the generation of lower hybrid waves. Bell and Ngo [7] have proposed that lower hybrid waves are excited when whistler mode waves are scattered by 10–100 m scale plasma density irregularities (Fig. 6.6). Propagation guided by FAI irregularities along open ﬁeld lines at high latitudes has been used to explain Lion’s roar observations at low altitudes in the polar cusp. Lion’s roar are intense sporadic whistler mode wave bursts observed in the Earth’s magnetosheath [1, 115, and references therein]. They are found in the 100–200 Hz frequency range and typically last for ∼10 s, though at times they can last for ≥5 minutes.

6 Density Irregularities and Whistler Propagation

179

Gurnett and Frank [46, 47], using Injun 5 data, detected lion roars in the high latitude magnetosheath near the polar cusp at 70–80◦ geomagnetic latitude and at ≤3000 km propagating parallel to the geomagnetic ﬁeld. They suggested that these waves were generated in the magnetosheath and trapped there in open tubes of force of the geomagnetic ﬁeld. These waves propagate along B0 , being guided by ﬁeld-aligned irregularities, and reach polar zone at ∼3000 km altitude. W-Mode Observations when the Source is at Low Altitude ( fp , presumably Langmuir waves, and in CMA3 in the form of slow-Zmode waves [James, 14]. The explanation of the slow-Z-mode observations

7 Dipole Measurements of Waves in the Ionosphere

197

called upon hot-plasma theory and an intermediate process of radiation by sounder-accelerated electrons to explain the characteristics of the waves at the OC receiver subpayload. 7.2.2 Whistler-Mode Propagation Near the Resonance Cone The two lowest pulse carrier frequencies transmitted in the OC two-point propagation experiment were 25 and 100 kHz, each in a receiver bandwidth of 50 kHz. Figure 7.2 is a detailed spectral display of the whistler-mode waves received in these two bands on the ﬂight downleg [James, 13]. These two greyscaled displays considerably expand the lowest part of Fig. 7.1, and show the spectra of received signals for selected portions of the history. The lower ﬁgure is for the bandwidth centred at 25 kHz and contains evidence of strong signals at that band centre. The same is true for the upper panel showing some of the 100-kHz band history. The two panels have been positioned in the diagram to have both the time after launch (TAL) and frequency scales maintained between the two. It is therefore easy to see that in addition to the spectral strength at the 25- and 100-kHz carriers, there is a broad swath of signal enhancement centred ﬁrst near the carrier in the bottom frame, then running diagonally toward its upper right corner. This same swath is seen to continue diagonally up through the top frame. It was suspected that the enhancement swath was a signature of loweroblique-resonance eﬀects [James, 12]. The “resonance cone frequency” frc is the whistler-mode frequency that puts the group velocity direction, at angle θgr with respect to B, along the transmitter-receiver direction, which is at angle δ with respect to B. Applying an expression for tan θgr as a function of f, fp and fc [Stix, 28] and equating it to tan δ leads to the relation tan2 δ = tan2 θgr =

2 −1 + fp2 /(frc − fc2 ) 2 2 1 − fp /frc

(7.3)

Inverting this equation for frc as a function of all the known variables throughout the total period in Fig. 7.2 produced the thin continuous line in that diagram labelled “frc ”. The frc locus is found to track the centre of the swath, conﬁrming the suspicion about the resonance-cone nature of the waves. This conclusion was supported by the observation of spin modulation of the signal in Fig. 7.2. Seen in both panels, spin modulation is especially clear in the lower frame in 700 < T AL < 750 as a series of pixels at 25 kHz that are alternatively strong and weak. In theory, the electric ﬁeld is polarized linearly along the wave vector direction for waves near the resonance condition. Fortuitously on OC, the major-frame duty cycle of HEX-REX is 3 seconds and the spin period of the receiving subpayload is very close to four times greater, 12 s. During each spin half rotation, HEX-REX executes two duty cycles; on one cycle, the receiving dipole is aligned close to the transmitted wave E; 3 s later, the alignment is perpendicular. Hence the clear 1-on /1-oﬀ pattern emerges.

198

H.G. James

Fig. 7.2. Summary plot of the detailed signal spectra of the 25- and 100-kHz pulses, for those portions of the downleg ﬂight when the resonance-cone frequency frc , as deﬁned in (7.2) and traced with the overlaid line, lies within either of the bandwidths centred on those two frequencies. The two panels have been positioned in the diagram to have both the time and frequency scales maintained between the two

The resonance cone signals at the carrier frequencies in Fig. 7.2 were much stronger than the electromagnetic whistler-mode signals at higher carrier frequencies in Fig. 7.1. These absolute signal strengths have been evaluated in a link calculation by Chugunov et al. [4]. Their detailed computation for the25kHz case took into consideration the geometry of the double-V dipoles on both ends of the link. A radiating theory for the transmitting dipoles that is pertinent to propagation directions near the cone was used [Mareev and Chugunov, 22]. For the receiving dipoles, a new application of the reciprocity principle [Chugunov et al. 3] was employed to estimate the receiving dipole’s Leﬀ , which was found to be 30 times its physical length. Good agreement between observed and theoretical resonance-cone signals was obtained.

7 Dipole Measurements of Waves in the Ionosphere

199

The good agreement obtained has implications for the interpretation of whistler-mode waves in space plasma. Sects. 7.3.2 through 7.3.4 following will illustrate this with a retrospective examination of some past observations of oblique-resonance-cone waves in the ionosphere.

7.3 Retrospective on Past Observations Section 7.1 reviewed analyses of OC data that conﬁrmed the theory for EM waves. The success of the classic short-dipole theory, represented here by the Kuehl [18] CW theory, has motivated a re-examination of reports about dipole Leﬀ for EM cases; this is in Subsect. 7.3.1. The success of the novel Chugunov [3] theory of dipole Leﬀ for quasi-electrostatic resonance-cone waves has prompted a return to published analyses of the amplitudes of such waves. Discussion along this line for both lower-oblique resonance waves (whistler mode) and upper-oblique resonance-cone waves (slow-Z mode) follows in Sects. 7.3.2 through 7.3.4. 7.3.1 Dipole Eﬀective Length for EM Propagation Jordan’s [16, pp. 336] deﬁnition of a generalized Leﬀ is taken as a point of departure. His (10-81) when applied to a cold plasma theory implies that the complex eﬀective-length vector can be written Leﬀ = ek Lkeﬀ + eθ Lθeﬀ + e φLφ eﬀ

(7.4)

This refers to a wave-vector space speciﬁed in spherical coordinates with the directions corresponding to radial direction k, polar-angle direction θ and azimuthal direction φ. The E polarization in magnetoionic theory in general has components in all three directions, as opposed to the vacuum case of Jordan’s (10–81) where the polarization ellipse is perpendicular to the wave vector. Under the reciprocity principle, it is assumed that an expression of the same form as Jordan’s (10–82) can be used to represent the vector ﬁeld E of an active dipole in a magnetoplasma as E=

60π I(0) Leﬀ , rλ0

(7.5)

where I(0) is the current injected at the antenna terminal, r is the distance and 0 is the vacuum wavelength. An expression for E was obtained from the Kuehl [18] theory used in the above-described tests of the OC radiation ﬁeld. His (14) can be rewritten as E=

k02 As e−iφ (eks Bs + eθs Cs + eφ Ds ) , 4π0 r

(7.6)

200

H.G. James

in which k0 is the vacuum wave number, φ is the phase, and we concentrate on a speciﬁc mode and solution of the dispersion relation, labelled with s. Tests of theory for radiated ﬁelds or Leﬀ must be carried out with due regard for two complications: First, the possibility that in some modes like the whistler, a single group-velocity direction can correspond to between one and three diﬀerent propagation directions; second, in space measurements, waves can take diﬀerent paths from a source to a receiver [Sonwalkar et al. 25]. The coeﬃcients As , Bs , Cs , Ds depend on the plasma parameters, the angle θ between the direction of propagation and B, and the dipole moment of the antenna current. When a triangular current distribution is assumed on the antenna, the dipole moment has a magnitude I(0)L/(4πf ). With this and the dipole-moment direction cosines inserted into the coeﬃcients, (7.6) can be rearranged to read E=

60πI(0) L (eks As Bs + eθs As Cs + eφ As Ds ) . rλ0 2

(7.7)

The Kuehl [18] expression is seen to be of the same form as Jordan’s (10– 82) in (7.5). The resulting coeﬃcients of the unit vectors are identiﬁed with the corresponding Leﬀ values in (7.4). It thus appears that the cold-plasma theory is consistent with the Jordan concept of general Leﬀ involving a vector description. For conformity with the principle of reciprocity, the open-circuit voltage induced on a receiving dipole is presumed to be Voc = E · =

L (eks As Bs + eθs As Cs + eφ As Ds ) 2

L (A B Ek + As Cs Eθ + As Ds Eφ ) . 2 s s

(7.8)

The coeﬃcients Bs , Cs , Ds are consistent with the polarization ratios given by the cold-plasma theory for the propagation direction. The coeﬃcients also depend on dipole orientation. Let us now see where the hypothesis of (7.8) leads when its predictions are compared with observations. Sonwalkar and Inan [26] scaled electric and magnetic ﬁelds measured on the DE-1 spacecraft to determine Leﬀ of the 200-m wire dipoles ﬂown on that satellite. Two measurement sets were found to give Leﬀ = 219.8 and 224.6 m. These evaluations were based on their expression (1) for Leﬀ , an assumption that has had wide use in space radio science. Often, Leﬀ as used in Sonwalkar and Inan [26] has been assigned a value of L/2, half the physical length of the dipole, appropriate for vacuum propagation detected by dipoles with L λ, the wavelength. The speciﬁc plasma parameters and geometry reported in Sonwalkar and Inan [26] allow Leﬀ to be evaluated from present (7.8) as L/2 times the multiplier factors As Bs , As Cs , As Ds for the three respective components. In the case of the whistler mode, the k and θ electric ﬁeld components are in phase with each other and in quadrature with the φ component. One axis of the E ellipse is the vector addition of the Ek and Eθ components and the

7 Dipole Measurements of Waves in the Ionosphere

201

other is the Eφ . All three components are generally needed to determine Voc . The magnitude of the eﬀective length of the dipole is !1/2 ≡ (L/2)M . (7.9) |Leﬀ | = (L/2) (As Bs )2 + (As Cs )2 + (As Ds )2 The multiplier M thus deﬁned for the Sonwalkar and Inan [26] cases is plotted as a function of the spin angle ξ in Fig. 7.3 with continuous and broken line. The inset ﬁgure shows that the spin axis of the DE-1 spacecraft is approximated to lie along the y axis. The wave vector direction and the plasma parameters take the values reported by the authors. The results are M values that oscillate between about four and zero. The theory thus conﬁrms that a short dipole receiving EM whistler-mode propagation can have |Leﬀ | ≈ L as reported by SI. Further comparison is not possible since Sonwalkar and Inan [26] assume that Leﬀ is isotropic whereas it is a vector here, depending on the orientations of both the wave vector and the dipole axis. Imachi et al. [8] determined Leﬀ of wire dipoles on the Geotail spacecraft by analyzing electric- and magnetic-ﬁeld amplitudes of chorus emissions. This result has been checked against the present method assuming that these whistler-mode waves propagate along B. The stated plasma and working frequencies and a reasonable electron gyrofrequency, 10 kHz, provide the variation of M (ξ) in the dot-dash line in Fig. 7.3. Although the details of the method of determining Leﬀ are not given by Imachi et al. [8], it is found that the range of M for all ξ brackets 1, the value deduced by those authors. Finally it is noted that (7.5) above simpliﬁes to the familiar expression Eθ =

60πI(0) L cos θ rλ0 2

(7.10)

for short dipoles and for working frequencies much higher than the electron characteristic frequencies, i.e., for the vacuum propagation conditions of (7.1). To check this, the multipliers of (7.8) were calculated for the same plasma characteristic frequencies as for Imachi et al. [8] but for the O mode, at a working frequency of 30 MHz. The result in Fig. 7.3 is the dot-dot-dot-dash curve, labelled “HF”, showing M when the θ term is the only term. The k has been put along the z axis. For the particular case of ξ = 0◦ , θ = 0◦ , for which the propagation direction is at right angles to this vacuum dipole axis, the r, θ, and φ multipliers have values of 0, −0.5i and −0.5, respectively. This same set becomes 0, −0.5i and 0.5 when run for the X-mode. The total Leﬀ magnitudes found from the sum of the two solutions are then (L/2)(0., 1., 0), corresponding to the expected linear polarized ﬁeld at θ = 0◦ from a simple dipole in vacuum, which radiates and detects the θ component only. 7.3.2 Intensity of Auroral Hiss Correct methodology for the dipole measurement of the absolute magnitudes of EM radio emissions in space includes a correct knowledge of the dipole Leﬀ .

202

H.G. James

kuehlef6.for, kuehl2.pro, Kuehl2.PS

Effective length multiplier, M

4 Sonwalkar and Inan (1986)

3

2

Imachi et at (2000) 1

0

0

45

90 135 Spin phase angle, ξ (˚)

180

Fig. 7.3. The multiplier M in (7.9) giving the magnitude of Leﬀ as a function of the dipole spin angle ξ, for comparison for three diﬀerent situations. The continuous and broken lines are the computed M for the two cases reported by Sonwalkar and Inan [26], the dot-dash curve for the Imachi et al. [8] conditions, and the lowest curve (HF) for vacuum conditions

This is illustrated in the history of auroral hiss. In the 1970s, there was interest in understanding the generation of hiss. Earlier work in space and astrophysical contexts had drawn attention to the idea that charged particles moving through magnetoplasma may produce EM radiation through the Cherenkov eﬀect – a particle radiates when its velocity matches the phase speed of EM plasma waves. An important question to be answered was whether a singleparticle theory suﬃced, wherein the total emission of a hiss source region could be obtained by considering all the particles therein as single incoherent radiators. Or were the absolute hiss levels high enough to require a more powerful process, e.g., a coherent beam-plasma interaction? Tests were carried out on diﬀerent satellite data sets. One example was ISIS-I observations of hiss in the dayside cusp at LF and MF [James, 9]. As illustrated in Fig. 7.4, that investigation dealt with passes of the spacecraft through L shells with soft electron precipitation. The observed evolution of hiss spectra was quantitatively consistent with spreading out of radiation

7 Dipole Measurements of Waves in the Ionosphere

203

Fig. 7.4. The trajectory of the ISIS-I satellite takes it through magnetic ﬁeld lines where its soft-particle spectrometer measures soft electron ﬂuxes [James, 9]. Low-frequency hiss is also measured by the sounder receiver as the spacecraft approaches, passes through and recedes from the precipitation. The calculation of total noise power ﬂux spectral density is computed within a bandwidth f ± ∆f by tracing resonance-cone group rays from the spacecraft location to establish 2 three-dimensional group cones that intersect the ﬂux sheet and so determine the total volume of energetic particles irradiating the spacecraft for that location and bandwidth (Reprinted with permission of the American Geophysical Union)

from the source ﬂux tubes on resonance-cone group paths. In one case study applying ISIS-I energetic-particle, ambient-density and wave measurements, incoherent-radiation theory was used to calculate theoretical electric-ﬁeld magnitudes. These were compared with values scaled from receiving-dipole voltages when the dipole was assumed to have an Leﬀ equal to its physical length, L. The ratio of the observed to theoretical ﬁeld strengths was about 20 dB, which led to the conclusion that the test-particle theory was not applicable. Similar discrepancies in the same sense were reported in other work, mostly at very low frequencies in the nightside ionosphere. Jorgensen [17] modelled the hiss generation region in the lower ionosphere and computed resulting

204

H.G. James

power ﬂux densities as high as 10−14 Wm−2 Hz−1 at 10 kHz. Gurnett and Frank [6] and Mosier and Gurnett [23] described the morphology of hiss with respect to other auroral-latitude phenomena observed with the Injun-V spacecraft. Power ﬂux densities for strong events were of the order 10−11 Wm−2 Hz−1 . Lim and Laaspere [20] also computed hiss levels at VLF and LF for typical energetic-ﬂux and ambient parameters. For comparison with the foregoing VLF observations, their computed ﬂux intensity was less than 10−13 Wm−2 Hz−1 . Taylor and Shawhan [29] carried out a case study of VLF hiss and energetic-particle spectra observed on Injun V. They came to similar conclusions, that the shape of the hiss spectrum is as expected based on ray paths and geometry but that the intensity of VLF hiss levels calculated was 2 orders of magnitude below observations. Maeda [21] reﬁned the model of the magnetosphere and calculated dayside hiss levels of 10−14 Wm−2 Hz−1 for strong electron ﬂuxes. For a wider summary of observed hiss intensities, see the Table 1 in LaBelle and Treumann [19]. Table 7.1 summarizes the preceding paragraph. The upshot of several parallel investigations of hiss power ﬂux was that observed values exceeded by at least 100 times values obtained from calculation using the incoherent theory [LaBelle and Treumann, 19]. This held true for a variety of diﬀerent observational circumstances. The power-ﬂux ﬁnding was typically followed by the conclusion that the incoherent test-particle theory for radiation was not applicable and that one should look to more powerful wave-particle interaction theory. However, the recent OC work at 25 kHz has now found that for resonance cone waves Leﬀ = 30 L. The ISIS-I data originally were interpreted assuming Leﬀ = L. This implies that electric ﬁeld strengths deduced were 30 times too large, or that the power ﬂux values were 20 log10 30 = 29.5 dB too high. Since this ﬁgure is similar to the disparity between calculated and observed hiss intensities, the rejection of the incoherent theory in the 1970s should now be reconsidered. Table 7.1. Auroral Hiss Field Strength Measurements and Calculations Observation VLF (0–100 kHz) Gurnett and Frank (1972) 0.5–1 ×10−11 Mosier and Gurnett (1972) >1, 5 × 10−11

LF-MF (100–500 kHz) James (1973) 0.81–4.6 ×10−17

Incoherent Radiation Theory Jorgensen (1968) 10−14 Lim and Laaspere (1972) 10−13 Taylor and Shawhan (1972) 0.5 − 3.2 × 10−13 Maeda (1975) 10−14 Ratio 11–32 dB James (1973) 0.21 − 8.8 × 10−19 Ratio: 17–26 dB

7 Dipole Measurements of Waves in the Ionosphere

205

That auroral hiss generation in the low-altitude auroral oval is correctly described with an incoherent theory may gain some credence when recent work on particle acceleration is consulted. Auroral electron acceleration is thought to result from the action of the parallel electric ﬁeld of shear Alfv´en waves and to occur at altitudes between about 3500 km and a few earth radii [Paschmann et al., 24, , pp. 187-189]. The low-earth-orbit-satellite observations cited above were all made below 3100 km, that is, below the acceleration region. Electron ﬂuxes in that region are expected to have lost their Alfv´en-imparted coherence through beam-plasma interactions. Various instability mechanisms like those producing hiss or auroral kilometric radiation may have the dominant role. In the former case, wave growth and saturation will happen over vertical scales of the order of ten plasma periods multiplied by particle velocities, say, several kilometers. What would then emerge from the bottom of the acceleration region would be turbulent electron ﬂuxes with spatial coherence scales corresponding to the saturation spectrum of the created waves. Paschmann et al. [24] state that the beam-plasma process causes quasilinear plateauing of the beam at the intermediate altitudes studied by James [9] and Taylor and Shawhan [29]. The eﬀect of plateauing is presumed to be one of shifting beam electrons to lower energies. However, at LF and MF, the hiss ray directions at frequencies not far below fp are not far from horizontal. Hence it can be argued that the particles measured by satellite are indeed responsible for the hiss observed on the same spacecraft, especially inside the precipitation L shells. The presence of turbulence would mean that overall the descending ﬂux would radiate incoherently but that the individual radiating entities would not be individual electrons but a three-dimensional ensemble of micro-structures. What might be the dimensions of a hiss-radiating microstructure? First of all, it is noted that an upper-limit on the cross-ﬁeld dimension of the emission source is given by the perpendicular scale size of the Alfv´en wave, which is of the order of the electron inertial length, 1 km. Therefore already the acceleration concept implies incoherence because the observed source regions in the auroral oval must contain a large number of such sources. Further, if thermal damping of whistler-mode waves aﬀects resonance-cone waves at refractive indices >1000, then saturated MF hiss at 300 kHz will have wavelengths of around 1 m. 1 m is probably a lower limit on structure sizes because the Debye length has approximately this order of magnitude in the acceleration region. Phase-space electron holes as small as a Debye length have been observed [Ergun et al., 5]. Assume that the hiss source is a ﬂux of 1-m structures. With a diﬀerential ﬂux density of 107 cm−2 s−1 ster−1 eV−1 in the dayside cusp, the density of particles in the beam is about 107 m−3 . This implies that the electric charge q in the incoherent radiated power expression should be 107 times a single-electron charge and that the number of the structures irradiating any observer must be reduced by 107 . In incoherent emission theory, the spectral power density varies as the square of the charge of the radiating “particle”. Hence to suppose that one observes radiating microstructure leads to a total

206

H.G. James

volume emission of (107 )2 /107 , a +70 dB correction in the absolute emission. The hypothesis of microstructure is discarded because the observation-theory disparity remains as large as before, albeit in the opposite sense. Hence the view is maintained that hiss in the topside ionosphere is incoherent radiation from a downward electron ﬂux below the acceleration region in the auroral magnetosphere. 7.3.3 Intersatellite Whistler-Mode Propagation Near a Resonance Cone Results from ISIS intersatellite propagation studies resemble the strong transmissions in the OC two-point experiment shown in Fig. 7.2. Whistler-mode transmission at 480 kHz was achieved between ISIS-I and ISIS-II 73-m sounder dipoles during a rendezvous campaign [James, 10]. ISIS I was in a 570 × 3520km polar orbit, while ISIS II had a nearly circular polar orbit around 1400 km altitude. The satellites were commanded into the “Alternate Mode” wherein the sounders cycled between conventional swept-frequency sounding and periods of about 20 s when the sounder pulse carrier frequency was held ﬁxed at 480 kHz. The rendezvous occurred at high latitude where the two orbital planes intersected. Transmissions in both directions were obtained on several rendezvous. In Fig. 7.5, a 4-s data excerpt from a rendezvous on 4 April 1974 displays representative pulses received. On the left are pulses emitted by ISIS II and received on ISIS I. The roles are reversed on the right side. Both satellites had 400-W emitters and produced input voltages at the other receiver of about 70 µV. Pulses about 100 µs long upon emission are seen to be dispersed spectacularly by 100 times and more at reception. All pulses in the 20-s interval from which this set is taken exhibit components with delays of the order of 10 ms. The pulses in 28 to 30 s also have sharp rising edges, on account of favourable antenna orientation at that time. Pulse stretching of this order was clearly maintained when the satellite separation lay well inside the group resonance cone for 480 kHz. Ray-optics studies were carried out based on three-dimensional electron density distributions obtained from the the real-height analysis of the sweptfrequency ionograms recorded during the rendezvous. Iterative techniques were used to ﬁnd the delays associated with direct inter-satellite propagation. Good agreement was found for the time of the sharp rising edges of pulses that had such, like those in Fig. 7.5. It was demonstrated that the pulse stretching was not caused by magnetoionic dispersion of the diﬀerent frequency components of the pulse. The signal delays required propagation near the resonance cone. Hence, the conclusion was that wave packets could take a great variety of scattering paths between the transmitter and the receiver, because the stretching was seen preferentially in the high-latitude irregular ionosphere. It was concluded that highly dispersed pulses arose from

7 Dipole Measurements of Waves in the Ionosphere

207

Fig. 7.5. Amplitude scans of whistler-mode pulses received reciprocally during a rendezvous of the ISIS-I and -II satellites [James, 10]. The pulse carrier RF was 480 kHz and the spacecraft separation was about 200 km (Reprinted with permission of the American Geophysical Union)

scattering by density irregularities along and inside the 480-kHz resonance cones through both spacecraft. An additional feature that supported the existence of quasi-electrostatic short-wavelength waves was the appearance of “primary” and “secondary” components of pulses. It was argued that this structure appeared because both the emitting and receiving dipoles had current distributions that favoured the emission or reception of certain wavelengths at a given frequency. These were only observed when each satellite lay near the 480-kHz group resonance cone of the other. The range of refractive indices, up to about 30, needed to explain the observed group delays correspond to wavelengths down to 20 m. This pointed to resonant- and anti-resonant-length relations between the now-ﬁnite-length dipoles and the waves, favouring the transmission of certain wave numbers and blocking others, hence the pulse envelope. 7.3.4 Intersatellite Slow-Z-Mode Propagation Near a Resonance Cone There have been observations of apparently intense waves propagating near the upper-oblique resonance cone in CMA 3. Another two-point experiment during the ISIS rendezvous campaign produced observations of strong, highly dispersed Z-mode pulses that appeared to carry an unusually large total energy [James, 11]. In fact the Z pulses resembled the inter-satellite whistlermode pulses just described, with comparable length and strength. Threedimensional ray-optics was applied to ﬁnd the phase paths of many pulse spectral components at the receiver. The received pulse envelope was then

208

H.G. James

constructed by an inverse Fourier transform, and found to have a very different shape from that measured. The conclusion again was that ionization irregularities eﬃciently quasi-electrostatically scatter waves and that the ISIS 73-m distributed dipole is sensitive to such waves. It was noted that highly dispersed slow-Z pulses were observed in ISIS monostatic ionograms when the satellite passed through high-latitude regions of density irregularities. These Z results are taken as further evidence of the combined eﬀect of the high Leﬀ , eﬀectively sensitizing the receiving dipoles, and of scattering ionospheric irregularities, providing long-delay paths along the group resonance cone. Benson et al. [2] in Chap. 1 of this volume have reviewed the Z mode as investigated using various spaceborne RF sounders. This includes new insights from the Radio Plasma Imager instrument on the IMAGE spacecraft. Pulse elongations similar to those found for whistler and Z- mode pulses during intersatellite propagation experiments between ISIS I and ISIS II were also found in recent whistler and Z-mode sounding experiments on IMAGE [Sonwalkar et al., 27]. These authors have put forward an interpretation similar to that given by [James, 10, 11], i.e. scattering by irregularities.

7.4 Conclusion The success of the cold-plasma CW theory for explaining dipole results from OEDIPUS commends wider use of that theory for supplying the eﬀective lengths of distributed dipoles for space-borne reception. The application of the principle of reciprocity brought over from the vacuum short-dipole theory appears useful for EM propagation. However the particular results about resonance-cone quasi-electrostatic propagation indicate a need for caution in the interpretation of signal levels. The importance in space physics of the correct dipole Leﬀ concept remains. It appears possible that hiss in the topside ionosphere is correctly described, after all, as incoherent radiation from a thermalized ﬂux left over from acceleration in the auroral magnetosphere. However, the interpretation of whistler-mode absolute intensities today is complicated by the clear evidence for scattered resonance-cone propagation. In the computation of power ﬂux densities from sources of emissions like hiss, which can have horizontal dimensions of hundreds of kilometers, there is now the added geometrical complication of dealing with scatter by irregularities. The scattering of EM VLF whistler-mode waves by ﬁeld-aligned small-scale irregularities into waves propagating near the cone has been explained as a consequence of either linear mode coupling or nonlinear mechanisms [Bell et al., 1, and references therein]. Whichever the case, the complete account of radiative transfer between source and observer brings a new challenge in understanding RF emissions of the atmosphere. This may be true not only for whistler-mode hiss but for other phenomena like hiss in the slow-Z mode.

7 Dipole Measurements of Waves in the Ionosphere

209

Given the state of knowledge about dipoles in 1978, the sustained pulses observed with topside sounders seemed to imply that great power was broadcast by the sounders along the oblique resonance cones. This inference is now tempered by the notion of large sensitivity of the receiving dipole to quasi-electrostatic resonance-cone waves, as represented by a big Leﬀ in the distributed dipole theory.

References [1] Bell, T.F., R.A. Helliwell and M.K. Hudson: Lower hybrid waves excited through linear mode coupling and the heating of ions in the auroral and subauroral magnetosphere, J. Geophys. Res. 96, 11,379–11,388 (1991). [2] Benson, R.F., P.A. Webb, J.L. Green, D.L. Carpenter, V.S. Sonwalkar, H.G. James and B.W. Reinisch: Active wave experiments in space plasmas: the Z mode. In: Proceedings Volume for the Ringberg Workshop, edited by J. LaBelle and R. A. Treumann, Springer Lecture Notes in Physics, Springer New YorkHeidelberg (2005). [3] Chugunov, Yu.V.: Receiving antenna in a magnetoplasma in the resonance frequency band, Radiophys. Quantum Electron. 44, 151–160 (2001). [4] Chugunov, Yu.V., E.A. Mareev, V. Fiala and H.G. James: Transmission of waves near the lower oblique resonance using dipoles in the ionosphere, Radio Sci. 38, 1022, doi:10.1029/2001RS002531 (2003). [5] Ergun, R.E., C.W. Carlson, J.P. McFadden, F.S. Mozer, L. Muschietti, I. Roth and R.J. Strangeway: Debye-scale plasma structures associated with magneticﬁeld-aligned ﬁelds, Phys. Rev. Lett. 81, 826–829 (1998). [6] Gurnett, D.A. and L.A. Frank: VLF hiss and related plasma observations in the polar magnetosphere, J. Geophys. Res. 77, 172–190 (1972). [7] Horita, R.E. and H.G. James: Two point studies of fast Z mode waves with dipoles in the ionosphere, Radio Sci. 39, doi:10.1029/2003RS002994 (2004). [8] Imachi, T., I. Nagano, S. Yagitani, M. Tsutsui and H. Matsumoto: Eﬀective lengths of the dipole antennas aboard Geotail spacecraft. In: Proc. 2000 Int. Sympos. Antennas and Propagat, (ISAP2000), IEICE of Japan, Tokyo 819– 822 (2000). [9] James, H.G.: Whistler-mode hiss at low and medium frequencies in the daysidecusp ionosphere, J. Geophys. Res. 78, 4578–4599 (1973). [10] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites, 2, Whistler-Mode pulses, Radio Sci. 13, 543–558 (1978). [11] James, H.G.: Wave propagation experiments at medium frequencies between two ionospheric satellites, 3, Z mode pulses, J. Geophys. Res. 84, 499–506 (1979). [12] James, H.G.: Electrostatic resonance-cone waves emitted by a dipole in the ionosphere, IEEE Trans. Antennas Propagat. 48, 1340–1348 (2000). [13] James, H.G.: Electromagnetic whistler-mode radiation from a dipole in the ionosphere, Radio Sci. 38, 1009, doi:10.1029/2002RS002609 (2003). [14] James, H.G.: Slow Z-mode radiation from sounder-accelerated electrons, J. Atmos. Solar-Terr. Phys. 66, 1755–1765 (2004).

210

H.G. James

[15] James, H.G. and W. Calvert: Interference fringes detected by OEDIPUS C, Radio Sci. 33, 617–629 (1998). [16] Jordan, E.C.: Electromagnetic Waves and Radiating Systems, Prentice-Hall, Englewood Cliﬀs NJ (1950). [17] Jørgensen, T.S.: Interpretation of auroral hiss measured on POGO-2 and at Byrd Station in terms of incoherent Cerenkov radiation, J. Geophys. Res. 73, 1055–1069 (1968). [18] Kuehl, H.H.: Electromagnetic radiation from an electric dipole in a cold anisotropic plasma, Phys. Fluids 5, 1095–1103 (1962). [19] LaBelle, J. and R.A. Treumann: Auroral radio emissions, 1. Hisses, roars and bursts, Space Sci. Rev. 101, 295–440 (2002). [20] Lim, T.L. and T. Laaspere: An evaluation of the intensity of Cerenkov radiation from auroral electrons with energies down to 100 eV, J, Geophys. Res. 77, 4145– 4157 (1972). [21] Maeda, K.: A calculation of auroral hiss with improved models for geoplasma and magnetic ﬁeld, Planet. Space. Sci. 23:843–865 (1975). [22] Mareev, E.A. and Yu.V. Chugunov: Excitation of plasma resonance in a magnetoactive plasma by external source, 1, A source in a homogeneous plasma, Radiophys. Quantum Electron. 30, 713–718 (1987). [23] Mosier, S.R. and D.A. Gurnett: Observed correlations betwen auroral and VLF emissions, J. Geophys. Res. 77, 1137–1145 (1972). [24] Paschmann, G., S. Haaland and R.A. Treumann: Auroral plasma physics, Space Sci. Rev. 103, No. 1–4 (2002). [25] Sonwalkar, V.S.: The inﬂuence of plasma density irregularities on whistler mode wave propagation. In: Proceedings Volume for the Ringberg Workshop, edited by J. LaBelle and R. A. Treumann, Springer Lecture Notes in Physics, Springer New York-Heidelberg (2005). [26] Sonwalkar, V.S. and U.S. Inan: Measurements of Siple transmitter signals on the DE 1 satellite: wave normal direction and antenna eﬀective length, J. Geophys. Res. 91, 154–164 (1986). [27] Sonwalkar, V.S., D.L. Carpenter, T.F. Bell, M. Spasojevic, U.S. Inan, J. Li, X. Chen, A. Venkatasubramanian, J. Harikumar, R.F. Benson, W.W.L. Taylor and B.W. Reinisch: Diagnostics of magnetospheric electron density and irregularities at altitudes > k⊥ ), or at the upper-hybrid resonance perpendicular to the background magnetic ﬁeld (k⊥ >> k|| ), or at the oblique high-frequency resonances with fpe < f < fuh . When these excitations are followed by conversion to one of the electromagnetic modes (X or O), the resulting mode conversion radiation can propagate long distances to distant space-borne or ground-based receivers. Rocket Observations Indeed, almost every suitably instrumented experiment penetrating the auroral region records plasma waves in the band between the electron plasma frequency and the upper hybrid frequency [1, 15, 58] as reviewed by LaBelle

214

P.H. Yoon et al.

[34]. Early and recent experiments indicated the bursty nature of these auroral Langmuir waves [see, e.g., Ergun, 9]. Recent experiments include detailed comparisons with electron distribution function measurements. For example, McFadden et al. [43] report on waves near fpe and show evidence that these originate from a Landau resonance with the simultaneously measured electron beams. Similarly, Samara et al. [53] show evidence that waves near fuh result from cyclotron resonance with the measured electrons in which electrons transfer energy into the waves. Ergun et al. [10] and Kletzing et al. [31] use wave-particle correlators to measure directly the interaction giving rise to the auroral Langmuir waves. Recent experiments also show the frequency structure of these high frequency waves and their conversion to electromagnetic modes. Beghin et al. [2] show high frequency waves near the plasma frequency observed in the auroral zone under a variety of conditions using wave receivers on the Aureol-3 satellite. For underdense conditions (fpe < fce ), the wave structure is predominantly at the plasma frequency and below, while for overdense conditions (fce < fpe ) the waves are at and above the plasma frequency. The waves below fpe in the underdense case are attributed to mode conversion of Langmuir waves into whistler waves. McAdams et al. [40] observed similar phenomena during an auroral rocket ﬂight, PHAZE-II in 1997, which penetrated both underdense and overdense plasma conditions. Their wave receiver had unprecedented resolution achieved by full sampling of the waveforms. With this resolution, McAdams et al. [40] observed that in the underdense case, the waves just below the plasma frequency form multiple constant-frequency “bands”. These bands are of longduration, up to tens of seconds (tens of km distance). Sometimes individual band structures are punctuated by an intense burst of emission at the plasma frequency, when the rocket crosses the location where the band frequency matches the plasma frequency. McAdams et al. [40] attribute these bands to mode conversion to whistler mode, as suggested earlier by Beghin et al. [2]. The banded structure is a natural consequence of the bursty intermittent nature of the causative Langmuir waves. In the overdense case, McAdams et al. [41] report completely diﬀerent and more complex wave structure. As reported by Beghin et al. [2] the waves occur at and above fpe in this case. However, the high resolution shows that they consist of multiple discrete features, with bandwidths less than a few hundred Hz, separated by the order of one kilohertz. Sometimes up to four or ﬁve multiplets are observed, but more often doublets. Figure 8.1a shows a spectrogram of the high frequency electric ﬁelds observed during a 3 s time interval of the PHAZE-II experiment. The distinct, rapidly varying wave cutoﬀ corresponds to the plasma frequency as argued by McAdams et al. [41]. Just above the plasma frequency, structured waves with amplitudes of ∼0.7 mV/m occur. Where their frequency nearly matches fpe , their amplitude maximizes, and they are clearly associated and appear trapped in electron density depletions. Further away from these electron density depletions, where their frequency

Frequency (kHz)

8 Mode Conversion Radiation 2200

(a)

2165 2130 2095 2060 140.9

Frequency (kHz)

PHAZE-II

215

141.4

141.8

142.2

142.6

1740

143.0

143.4

RACE (b)

1705 1670 1635 1600 760.7

761.1

761.5

761.9

762.3

762.7

763.2

Seconds After Launch

Fig. 8.1. (a) High frequency electric ﬁelds observed during the PHAZE-II experiment show multiple discrete frequency features, with bandwidths less than a few hundred Hz, separated by the order of one kHz. (b) High frequency electric ﬁelds observed by RACE show similar structures but with diﬀerent frequency-time variations (Reprinted with permission of the American Geophysical Union)

exceeds the plasma frequency by ten percent or so, they appear as descending tones with short durations of order 100 ms. Hundreds of descending tones occur during a 10 s interval. By the time the rocket travelled 10 s (or about 10 km) away from the region with the electron beam and the electron density irregularities, the waves died out to an undetectable level. Since 1997, several further auroral rocket experiments have included fullwaveform measurements of high frequency plasma waves. Figure 8.1b shows a spectrogram of the high frequency electric ﬁelds observed during 10 s interval on one of these experiments, RACE, launched in 2002. The Langmuir waves in the overdense auroral ionospheric plasma show many features in common with those previously observed by McAdams et al. [41]. They are associated with the combination of auroral electron beam and electron density structure evidenced, as before, by variations in the wave cutoﬀ associated with the plasma frequency. Again the waves are most intense where their frequency nearly matches fpe , and there is evidence for trapping of the waves in density depletions in this region. Though not shown in Fig. 8.1b, observations from many kilometers away from this source region, where the wave frequency exceeds the local plasma frequency by ten percent or so, the wave intensity is lower, and again by the time the rocket travels about 10 km from the source region, the waves are undetectable [Samara et al., 52]. The detailed wave

216

P.H. Yoon et al.

structure consists of discrete features separated by about 1–2 kHz. The discrete features often occur in pairs. However, unlike the previous case, features with decreasing frequency are not observed exclusively. Several examples of features with increasing frequency are shown in Fig. 8.1b. In addition, some examples last far more than 100 ms. In one case not shown in Fig. 8.1b, a discrete feature is intermittent but continuously detectable for many seconds. McAdams et al. [41] labelled their observed features “chirps” based on the consistent decreasing frequency and short 100 ms lifetime. However, the recent observations show that this name is overly restrictive, since a far broader variety of frequency variations occurs. As mentioned above, auroral HF waves are detected throughout the band between the plasma and upper hybrid frequencies, and recent experiments include full waveform measurements of the waves near the upper-hybrid frequency, complementing the observations near fpe cited above. Figure 8.2, from Samara et al. [53], shows a spectrogram of waves at and just below fuh in an active aurora penetrated by the HIBAR rocket, launched in 2003. Two bursts of highly structured waves occur where the upper hybrid frequency matches the electron cyclotron harmonic (fuh ≈ 2fce ). The waves lie at and just below fuh , which is indicated by the weak wave cutoﬀ and conﬁrmed by a second wave cutoﬀ at the plasma frequency (not shown). In the ﬁrst wave burst, of duration 0.5 s and total amplitude up to 10 mV/m, the waves occur in 2–4 visible bands with bandwidth of about 1–5 kHz separated by about 8 kHz. Each of these bands is composed of several sub-bands with separation of order 1–2 kHz. The entire band structure decreases in frequency with increasing time and altitude, at a rate of 21 kHz/s, approximately ten times too high to be explained by the decrease of the electron cyclotron harmonic frequency with increasing altitude. The second wave burst is of shorter duration. In the second burst, the emission is also banded, but the separations of the bands are of order 2–4 kHz. It is not clear whether the band separation in the second event should be compared to the larger separation of the principal bands in

Fig. 8.2. A spectrogram highlighting waves just below fuh in an active aurora penetrated by the HIBAR rocket [from 53]. The HIBAR observations suggest trapping of the upper hybrid waves in electron density enhancements, as discussed in Sect. 8.3 (Reprinted with permission of the American Geophysical Union)

8 Mode Conversion Radiation

217

the ﬁrst event, or whether it should be identiﬁed with the sub-band structure in the ﬁrst event, which is about the same size. The electron density observations from HIBAR suggest that these upper hybrid waves are trapped in electron density enhancements: the second wave burst clearly corresponds to an enhancement, whereas the ﬁrst wave burst corresponds to a “shoulder” in the electron density proﬁle, a signature which may result from the one-dimensional sampling of a density enhancement by the rocket trajectory. Earlier lower-resolution rocket observations of auroral upper hybrid waves near fuh also suggested this trapping, supported by careful consideration of the wave dispersion relation by Carlson et al. [6]. Ground-Based Observations Strong evidence suggests that the structured waves excited near the upper hybrid frequency under the matching conditions fuh ≈ 2fce or fuh ≈ 3fce have converted to O mode EM radiation and been detected remotely with satellites overﬂying the auroral zone [e.g., James et al., 24] and with groundbased observatories at northern and southern auroral zone sites [e.g., Kellogg and Monson, 29, 30] and [Weatherwax et al., 59, 61]. Using a topside ionospheric sounder in a passive conﬁguration, James et al. [24] observe LO radiation emerging from sources in the topside ionosphere, and by ray-tracing the observed LO radiation from the satellite to its source, into the topside ionospheric density proﬁle measured with periodic active soundings, they conclude that the radiation is generated near the upper hybrid frequency at the matching conditions. Similarly, Hughes and LaBelle [22] ray-trace the observed radiation from a ground-based direction-ﬁnding antenna array into the bottom-side ionospheric density proﬁle measured with an incoherent scatter radar, and they likewise conclude that the radiation is generated near the upper hybrid frequency at the matching condition. Theoretical work shows that Z-modes at the fuh = N fce matching condition can be excited in auroral conditions [see, Kaufman, 28] and [Yoon et al., 64]. Furthermore, based on such density structures observed at times of auroral roar emissions, Yoon et al. [65, 66] show that for a narrow range of frequencies and initial wave phase angles, trapped Z mode can be converted to O mode via the “Ellis” radio window. Weatherwax et al. [62] also present a detailed maser instability analysis of upper-hybrid, Z, and X mode wave excited along an auroral ﬁeld line, discussing the generation of topside and bottomside radiation within the context of the observations of James et al. [24], Benson and Wong [3], and Hughes and LaBelle [8]. In the case of the ground-level observations, called auroral roar, deployment of high-bandwidth receivers with large data rates reveals that the mode conversion radiation is composed of complex patterns of discrete frequency features as shown by LaBelle et al. [35] and Shepherd et al. [55]. Figure 8.3 shows an example of this ﬁne structure, recorded April 21, 2003, at South Pole Station. The most intense of these emissions have amplitudes up to 500 µV/m.

218

P.H. Yoon et al.

Fig. 8.3. 2595–2685 kHz spectrogram recorded at South Pole Station at 2130 UT on April 21, 2003, showing auroral roar with complex ﬁne structure

As known from previous studies [35, 55] auroral roar consists of multiple ﬁne structures, with width as narrow as 0) can satisfy this criterion. The reverse is true in the overdense case (ωp0 > Ω). Below, we restrict our analysis to the underdense case. Yoon and LaBelle [68] treat the overdense case. The discrete frequency spectrum associated with eigenmodes trapped within the density structure can be derived on the basis of the continuity of the eikonal solutions of (8.14) across the cutoﬀ points, x0 [V (x)]1/2 dx = (n + 1/2) π , (8.15) −x0

where n = 0, 1, 2, . . . Note that the cutoﬀs x ± x0 are real and positive if the condition 2 2 + 3k 2 vT2 /2 < ω 2 < (1 + δ)(ωp0 + 3k 2 vT2 /2) (8.16) ωp0 is satisﬁed. Inserting the speciﬁc expression for V (x) deﬁned in (8.14) to the matching condition (8.15) and carrying out closed-form analytical integration over x, we arrive at 1/2 2 2 v 3k ω p0 T 1+ [K(η) − E(η)] = (2n + 1) π , 4δ 1/2 kL 2 ω ωp0 2 (1 + δ)(ωp0 + 3k 2 vT2 /2) − ω 2 η= 2 1/2 δ 1/2 ωp0 + 3k 2 vT2 /2

!1/2 .

(8.17)

In the above K(η) and E(η) are complete elliptic integrals. Equation (8.17) constitutes a transcendental dispersion equation which supports a discrete

8 Mode Conversion Radiation

227

frequency spectrum as its solution. The simplest approximation is to replace the full elliptic integrals with their small-argument series approximation, K(η) − E(η) ≈ πη 2 /4, which is valid if the normalized cutoﬀs ±x0 /L are suﬃciently small, x20 /L2 1. Adopting such a simpliﬁcation, we obtain an analytical expression for the eigenvalues, ωp0 ω= 2

3k 2 vT2 1+ 2 2ωp0

1/2

$

4(1 + δ) + a2n

!1/2

− an

% ,

(8.18)

(2n + 1) δ 1/2 . kL Applying the above approximate solution for the discrete spectrum requires that the solution ω 2 be conﬁned to the range speciﬁed by (8.16). For typical ionospheric conditions, the condition (8.16) is easily satisﬁed. In the numerical computation of the discrete wave spectrum (8.16), the width of the density structure plays a key role. Taking an example from the auroral ionosphere, if the plasma frequency is fp0 = ωp0 /2π ∼ 2 MHz, and the background electron thermal energy is T ∼ 0.1 eV, then we calculate an =

α ≡ Lωp0 /vT ≈ 105 L[km] . L = 100m implies α ≈ 104 , L =1 km implies α ∼ 105 , and L = 1 m implies α = 102 . The characteristic parallel wave number associated with Langmuir waves is determined by the condition for maximum wave growth rate. For V0 /vT b ≈ 10 and T ≈ Tb , we have kvT /ωp0 ≈ 0.1. In Figure 8.5, we plot the discrete eigenvalue f versus n, for fp0 = 2 MHz, kvT /ωp0 = 0.1, density enhancement factor δ = 0.05, and for two diﬀerent structure widths, L = 100 m and L = 500 m. Note that the wider structure leads to smaller frequency spacing between individual discrete modes. The L = 500 m 2064.8

2064

2064.6

2063

2064.4

f [kHz]

f [kHz]

L = 100 m 2065

2062 2061 2060

2064.2 2064

0

2

4

6 n

8

10

2063.8

0

2

4

6

8

10

n

Fig. 8.5. Discrete Langmuir wave spectrum f [kHz] versus the mode number n, for two diﬀerent density structure widths, L, corresponding to 100 m and 500 m

228

P.H. Yoon et al.

predicted eigenmode spacings are of order 0.5 kHz and 0.1 kHz, respectively. Yoon and LaBelle [68] ﬁnd similar eigenmode spacings in the overdense case, which is appropriate to the observations of McAdams and LaBelle [41]. These spacings are comparable to those observed by McAdams and LaBelle [41] and calculated by McAdams et al. [42]. 8.3.3 Discrete Upper-Hybrid Waves in Density Structures The frequency spacing associated with auroral roar ﬁne structure was ﬁrst addressed by Shepherd et al. [55], who proposed an explanation based upon geometrical considerations in which the excitation of multiplet discrete modes resulted from standing waves within planar ﬁeld-aligned density cavities with vertically converging walls. This approach was similar to an idea originally proposed by Calvert [5] to account for AKR ﬁne structure. This explanation faces several diﬃculties, however, which Shepherd et al. [55] discuss. The principal problem is that it requires the wave-vector to vanish at the reﬂection points, which means that the wave cannot continuously grow as it stands in the cavity. Combined with the relatively low growth rate, it is diﬃcult to envision how suﬃcient gain can be achieved to allow the discrete eigenmodes to grow. Yoon et al. [67] present a model which corrects the primary diﬃculty of the Shepherd et al. [55] approach by considering circulating eigenmodes in a cylindrical geometry. This geometry is similar to that suggested by Carlson et al. [6] to explain the enhancement of upper hybrid waves in an artiﬁcially enhanced density structure, although they did not consider eigenmode structure. As discussed above, recent observations in the auroral ionosphere suggest that waves near the upper hybrid frequency, polarized with wave-number primarily perpendicular rather than parallel to the background magnetic ﬁeld, exhibit ordered frequency spectra with peaks spaced by ∼ 10 kHz, and within those, peaks spaced by ∼1 kHz [Samara et al., 53]. These observations qualitatively conﬁrm the Yoon et al. [67] model of upper hybrid wave eigenmodes in a cylindrical density enhancement. To model these waves, consider the upper-hybrid wave equation (8.12) with the right-hand side ignored, d4 φ(x) ω 2 − ωp2 (x) − Ω 2 d2 φ vT2 ωp2 (x) ω 2 − ≈0. 2Ω 2 Ω 2 ω 2 − Ω 2 dx4 ω2 − Ω 2 dx2

(8.19)

Applying the density model (8.13) to the above equation, we obtain the following equations in dimensionless variables:

8 Mode Conversion Radiation

d2 φ 2 + kuh (X) φ = 0, dX 2 Q=

2 kuh (X) = Q

2 w2 − wn2 , ρ2 α2 w2

wn2 = 1 + α2 ,

X02 =

229

X02 − X 2 , 1 + δ + X2

wx2 − w2 , w2 − wn2

wx2 = 1 + (1 + δ) α2 ,

(8.20)

where

2 ωp0 vT ω x , ρ= , α2 = 2 , w = . (8.21) L LΩ Ω Ω 2 If w2 < 1+α2 , then kuh (X) > 0 everywhere, implying that only continuous 2 2 (X) < 0, and eigenmodes exist. If w > (1 + δ) α2 , on the other hand, then kuh only damped modes exist. Thus, the possibility of trapped eigenmodes exists only in the intermediate range wn2 < w2 < wx2 . One can show that discrete modes cannot exist if δ < 0 (i.e., density depletion), since in this case the internal region (X 2 < X02 ) is characterized by damped solutions while the outer region supports oscillatory solutions. Therefore, the situation is similar to that of discrete Langmuir waves, except that the discrete modes occur for the case of density enhancements rather than density depletions. Asserting the matching condition (8.15) across the cutoﬀ points results in an equation for the eigenfrequencies similar to (8.17),

1 π (2 δ)1/2 (w2 − 1)1/2 [K(ζ) − E(ζ)] = n + , ρα w 2 2

X=

ζ=

[1 + (1 + δ)α2 − w2 ]1/2 , δ 1/2 (w2 − 1)1/2

(8.22)

which can be approximately solved for w2 ,

1/2 1 a2n a2n 2 2 2 2 − an wx (wx − 1) + wx − , w = 1 − a2n 2 4 1/2 δ ρα . an = (2n + 1) 2

(8.23)

In Fig. 8.6 we plot the solution of (8.23) for α = 0.7746, δ = 4, ρ = 10−2 /L, and for two choices for the size of the density structures: L = 100m and L = 500 m. In addition to the above normalized input parameters, we take the electron cyclotron frequency to be 1 MHz, as is approximately the case in the auroral ionosphere. The resulting eigenmode spacings are of order 0.2 kHz for L = 100 m and 0.1 kHz for L = 500 m; both cases fall within the range of frequency spacings observed in ﬁne structures of auroral roar emissions [see Shepherd et al., 55]. These predictions are narrower than the principal band spacing, but not much narrower than the substructure spacing, observed in rocket data of structured upper hybrid waves [Samara et al., 53].

230

P.H. Yoon et al. L = 500 m 2000

1999.5

1999.9 f [kHz]

f [kHz]

L = 100 m 2000

1999 1998.5

1998

1999.8 1999.7 1999.6

0

2

4

6 n

8

10

0

2

4

6

8

10

n

Fig. 8.6. Discrete upper-hybrid wave spectrum f [kHz] versus the mode number n, for two diﬀerent density structure widths, L, corresponding to 100 m and 500 m

8.4 Conclusions As reviewed above, the terrestrial ionosphere and magnetosphere include numerous examples of mode-conversion radiation, which occurs when nonthermal particle distributions excite electrostatic waves which convert to EM waves. In the theory section above, we have presented linear analysis of the normal modes of high-frequency plasma waves in inhomogeneous plasma, where the inhomogeneity, either a density enhancement or depletion, can be described by a Lorentzian form. This generalized treatment contains both the Langmuir wave eigenmodes proposed by McAdams et al. [42] and the upper hybrid wave eigenmodes proposed by Yoon et al. [67]. The frequency spacings derived from these models, for inhomogeneities of scale size 100–1000 m, match to the observed spacings of frequency ﬁne structure observed in auroral Langmuir wave “chirps” [McAdams and LaBelle, 41] and auroral roar ﬁne structure [35, 55]; upper hybrid wave frequency ﬁne structure recently observed directly with a rocket experiment shows somewhat wider frequency spacings [Samara et al., 53]. The theoretical treatment outlined above is independent but does not exclude the possibility that other mechanisms may act to produce wave structure, such as for example the discretization upon wave conversion in inhomogeneous plasma discussed by Willes and Cairns [63]. Perhaps the most signiﬁcant aspect of these studies lies in the potential use of observations of mode conversion radiation to remotely sense the plasma environment of the source region, gaining information about electron densities, magnetic ﬁelds, and depth and spatial scale of irregularities. The solar system is replete with examples of the mode-conversion radiation exhibiting ﬁne frequency structures, ranging from nonthermal emissions of planetary ionospheres and magnetospheres to some types of radio emissions originating in the solar wind. To remotely sense terrestrial ionospheric and magnetospheric plasma using mode conversion radiation has intriguing implications

8 Mode Conversion Radiation

231

for interpretation of the emissions originating in more remote plasmas, where direct measurements are not available and parameters are less well-known.

Acknowledgments This research was supported by the following National Science Foundation grants: ATM-0223764 to the University of Maryland; OPP-0338105, OPP0341470 and ATM-0243645 to Siena College; and OPP-0090545 and ATM0243595 to Dartmouth College. Dartmouth College was further supported by NASA grant NNG04WC27G. We thank Doug Menietti for the Cluster-II data depicted in Fig. 8.4 and Shengyi Ye for the South Pole data shown in Fig. 8.3.

References [1] Bauer, S.J., and R.G. Stone: Satellite observations of radio noise in the magnetosphere, Nature 128, 1145, 1968. [2] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and HF auroral hiss generated at low altitudes, J. Geophys. Res. 94, 1359, 1989. [3] Benson, R.F., and H.K. Wong: Low-altitude ISIS 1 observations of auroral radio emissions and their signiﬁcance to the cyclotron maser instability, J. Geophys. Res. 92, 1218, 1987. [4] Cairns, I. H. and J.D. Menietti: Radiation near 2fp and intensiﬁed emissions near fp in the dayside and nightside auroral region and polar cap, J. Geophys. Res. 102, 4787, 1997 [5] Calvert, W.: A feedback model for the source of auroral kilometric radiation, J. Geophys. Res. 87, 8199, 1982. [6] Carlson, C.W. et al.: unpublished manuscript, 1987. [7] Carpenter, D.L., and R.R. Anderson: An ISEE/whistler model of equatorial electron density in the magnetosphere, J. Geophys. Res. 97, 1097, 1992. [8] Chiu, Y.T., and M. Schulz: Self-consistent particle and parallel electrostatic ﬁeld distributions in the magnetospheric – ionospheric auroral region, J. Geophys. Res. 83, 629, 1978. [9] Ergun, R.E., C.W. Carlson, J.P. McFadden, J.H. Clemmons, and M.H. Boehm: Evidence of transverse Langmuir modulational instability in a space plasma, Geophys. Res. Lett. 18, 1177, 1991a. [10] Ergun, R.E., C.W. Carlson, J.P. McFadden, J.H. Clemmons, and M.H. Boehm: Langmuir wave growth and electron bunching: Results from a wave-particle correlator, J. Geophys. Res. 96, 225, 1991b. [11] Filbert, P.C. and P.J. Kellogg: Electrostatic noise at the plasma frequency beyond the earth’s bow shock, J. Geophys. Res. 84, 1369, 1979. [12] Gough, M.P.: Nonthermal continuum emissions associated with electron injections: Remote plasmapause sounding, Planet. Space Sci. 30, 657, 1982. [13] Green, J.L., B.R. Sandel, S.F. Fung, D.L. Gallagher, and B.W. Reinisch: On the origin of kilometric continuum, J. Geophys. Res. 107, 1105,10.1029/2001JA000193, 2002.

232

P.H. Yoon et al.

[14] Green, J.L., Scott Boardsen, Shing F. Fung, H. Matsumoto, K. Hashimoto, R.R. Anderson, B.R. Sandel, and B.W. Reinisch: Association of kilometric continuum with plasmapsheric structures, J. Geophys. Res. 109, A03203, doi:10.1029/2003JA010093, 2004. [15] Gregory, P.C.: Radio emissions from auroral electron, Nature 221, 350, 1969. [16] Gurnett, D.A. and R.R. Shaw: Electromagnetic radiation trapped in the magnetosphere above the plasma frequency, J. Geophys. Res. 78, 8136, 1973. [17] Gurnett, D.A.: The Earth as a radio source: The nonthermal continuum, J. Geophys. Res. 80, 2751, 1975. [18] Gurnett, D.A., F.L. Scarf, W.S. Kurth, R.R. Shaw, and R.L. Poynter: Determination of Jupiter’s Electron Density Proﬁle from Plasma Wave Observations, J. Geophys. Res. 86, 8199-8212, 1981. [19] Hashimoto, K., W. Calvert, and H. Matsumoto: Kilometric continuum detected by Geotail, J. Geophys. Res. 104, 28645, 1999. [20] Horne, R.B.: Path-integrated growth of electrostatic waves: The generation of terrestrial myriametric radiation, J. Geophys. Res. 94, 8895, 1989. [21] Horne, R.B.: Narrowband structure and amplitude of terrestrial myriametric radiation, J. Geophys. Res. 95, 3925, 1990. [22] Hughes, J.M. and J. LaBelle: Plasma conditions in auroral roar source regions inferred from radio and radar observations, J. Geophys. Res. 106, 21157, 2001. [23] Issautier, K., N. Meyer-Vernet, M. Moncuquet, S. Hoang, and D.J. McComas: Quasi-thermal noise in a drifting plasma: theory and application to solar wind diagnostic on Ulysses, J. Geophys. Res. 104, 6691, 1999. [24] James, H.G., E.L. Hagg, and L.P. Strange: Narrowband radio noise in the topside ionosphere, AGARD Conf. Proc., AGARD-CP-138, 24, 1974. [25] Jones, D.: Source of terrestrial nonthermal radiation, Nature 260, 385, 1976. [26] Jones, D.: Latitudinal beaming of planetary radio emissions, Nature 288, 225, 1980. [27] Kasaba, Y., H. Matsumoto, K. Hashimoto, R.R. Anderson, J.-L. Bougeret, M.L. Kaiser, X.Y. Wu, and I. Nagano: Remote sensing of the plasmapause during substorms: Geotail observation of nonthermal continuum enhancement, J. Geophys. Res. 103, 20389, 1998. [28] Kaufmann, R. L.: Electrostatic wave growth: Secondary peaks in measured auroral electron distribution function, J. Geophys. Res. 85, 1713, 1980. [29] Kellogg, P.J. and S.J. Monson: Radio emissions from aurora, Geophys. Res. Lett. 6, 297, 1979. [30] Kellogg, P.J. and S.J. Monson: Further studies of auroral roar, Radio Sci. 19, 551, 1984. [31] Kletzing, C.A., S.R. Bounds, J. LaBelle and M. Samara: Observation of the reactive component of Langmuir wave phase-bunched electrons, Geophys. Res. Lett., in press, 2005. [32] Kurth, W.S., D.A. Gurnett, and R.R. Anderson: Escaping nonthermal continuum radiation, J. Geophys. Res. 86, 5519, 1981. [33] Kurth, W.S.: Detailed observations of the source of terrestrial narrowband electromagnetic radiation, Geophys. Res. Lett. 9, 1341, 1982. [34] LaBelle, J.: Radio Noise of Auroral Origin: 1968–1988, J. Atmos. Terr. Phys. 51, 197, 1989. [35] LaBelle, J., M.L. Trimpi, R. Brittain, and A.T. Weatherwax: Fine structure of auroral roar emissions, J. Geophys. Res. 100, 21953, 1995.

8 Mode Conversion Radiation

233

[36] LaBelle, J., S.G. Shepherd, and M.L. Trimpi: Observations of auroral medium frequency bursts, J. Geophys. Res. 102, 22221, 1997. [37] LaBelle, J. and R.A. Treumann: Auroral Radio Emissions, 1. Hisses, Roars, and Bursts, Space Sci. Rev. 99, 295, 2002. [38] Lund, E.J., J. LaBelle, and R.A. Treumann: On quasi-thermal ﬂuctuations near the plasma frequency in the outer plasmasphere: A case study, J. Geophys. Res. 99, 23651, 1994. [39] Lund, E.J., R.A. Treumann, and J. LaBelle: Quasi-thermal ﬂuctuations in a beam-plasma system, Phys. Plasmas 3, 1234, 1996. [40] McAdams, K.L., J. LaBelle, M.L. Trimpi, P.M. Kintner, and R.A. Arnoldy: Rocket observations of banded structure in waves near the Langmuir frequency in the auroral ionosphere, J. Geophys. Res. 104, 28109, 1999. [41] McAdams, K.L. and J. LaBelle: Narrowband structure in HF waves above the plasma frequency in the auroral ionosphere, Geophys. Res. Lett., 26, 1825, 1999. [42] McAdams, K.L., R.E. Ergun, and J. LaBelle: HF Chirps: Eigenmode trapping in density deletions, Geophys. Res. Lett. 27, 321, 2000. [43] McFadden, J.P., C.W. Carlson, and M.H. Boehm: High-frequency waves generated by auroral electrons, J. Geophys. Res. 91, 12079, 1986. [44] Melrose, D.B.: A theory for the nonthermal radio continua in the terrestrial and Jovian magnetospheres, J. Geophys. Res. 86, 30, 1981. [45] Menietti, J.D., R.R. Anderson, J.S. Pickett, and D.A. Gurnett: Near-source and remote observations of kilometric continuum radiation from multispacecraft observations, J. Geophys. Res. 108, 1393, doi:10.1029/2003JA009826, 2003. [46] Menietti, J.D., O. Santolik, J.S. Pickett, and D.A. Gurnett: High resolution observations of continuum radiation, Planet. Space Sci., submitted 2004. [47] Meyer-Vernet, N., and C. Perche: Tool kit for antennae and thermal noise near the plasma frequency, J. Geophys. Res. 94, 2405, 1989. [48] Meyer-Vernet, N.: On the thermal noise in an anisotropic plasma, J. Geophys. Res. 21, 397, 1994. [49] Osherovich, V. and J. Fainberg: Dependence of frequency of nonlinear cold plasma cylindrical oscillations, Phys. Plasmas 11, 2314, 2004. [50] Pottelette, R. and R.A. Treumann: Auroral acceleration and radiation, this volume, 2005. [51] R¨ onnmark, K.: Emission of myriametric radiation by coalescence of upper hybrid waves with low frequency waves, Anal. Geophys. 1, 187, 1983. [52] Samara, M., J. LaBelle, C.A. Kletzing, and S.R. Bounds: Rocket Measurements of Polarization of Auroral HF Waves, EOS Trans. Am. Geophys. Union, Fall Meeting, 2002. [53] Samara, M., J. LaBelle, C.A. Kletzing, and S.R. Bounds: Rocket observations of structured upper hybrid waves at fuh = 2fce , Geophys. Res. Lett. 31, L22804, 10.1029/2004GL021043, 2004. [54] Samara, M., J. LaBelle, I.H. Cairns and R.A. Treumann, Statistics of Auroral Langmuir Waves, J. Geophys. Res., submitted, 2005. [55] Shepherd, S.J., J. LaBelle, and M.L. Trimpi: Further investigation of auroral roar ﬁne structure, J. Geophys. Res. 103, 2219, 1998a. [56] Shepherd, S.J., J. LaBelle, and M.L. Trimpi: The polarization of auroral radio emissions, Geophys. Res. Lett. 24, 3161, 1998b.

234

P.H. Yoon et al.

[57] Steinberg, J.-L., S. Hoang, and M.F. Thomsen: Observations of the Earth’s continuum radiation in the distant magnetotail with ISEE-3, J. Geophys. Res. 95, 20781, 1990. [58] Walsh, D., F.T. Haddock, and H.F. Schulte: Cosmic radio intensities at 1.225 and 2.0 Mc measured up to an altitude of 1700 km, in: Space Res., 4, edited by P. Muller, pp. 935–959, North Holland Publishing Company, Amsterdam, 1964. [59] Weatherwax, A.T., J. LaBelle, M.L. Trimpi, and R. Brittain: Ground-based observations of radio emissions near 2fce and 3fce in the auroral zone, Geophys. Res. Lett. 20, 1447, 1993. [60] Weatherwax, A.T., J. LaBelle, and M.L. Trimpi: A new type of auroral radio emission observed at medium frequencies (∼1350–3700 kHz) using groundbased receivers, Geophys. Res. Lett. 21, 2753, 1994. [61] Weatherwax, A. T., J. LaBelle, M. L. Trimpi, R. A. Treumann, J. Minow, and C. Deehr: Statistical and case studies of radio emissions observed near 2fce and 3fce in the auroral zone, J. Geophys. Res. 100, 7745, 1995. [62] Weatherwax, A.T., P.H. Yoon, and J. LaBelle: Interpreting observations of MF/HF radio emissions: Unstable wave modes and possibilities to passively diagnose ionospheric densities, J. Geophys. Res. 107, 1213, doi:10.1029/2001JA000315, 2002. [63] Willes, A.J. and I.H. Cairns: Banded frequency structure from linear mode conversion in inhomogeneous plasmas, Phys. Plasmas 10, 4072, 2003. [64] Yoon, Peter H., A.T. Weatherwax, T.J. Rosenberg, and J. LaBelle, Lower ionospheric cyclotron maser theory: A possible source of 2fce and 3fce auroral radio emissions, J. Geophys. Res. 101, 27,015, 1996. [65] Yoon, P.H., A.T. Weatherwax, and T.J. Rosenberg: On the generation of auroral radio emissions at harmonics of the lower ionospheric electron cyclotron frequency: X, O, and Z mode maser calculations, J. Geophys. Res. 103, 4071, 1998a. [66] Yoon, P.H., A.T. Weatherwax, T.J. Rosenberg, J. LaBelle, and S.G. Shepherd: Propagation of medium frequency (1–4 MHz)auroral radio waves to the ground via the Z-mode radio window, J. Geophys. Res. 103, 29267, 1998b. [67] Yoon, P.H., A.T. Weatherwax, and J. LaBelle: Discrete electrostatic eigenmodes associated with ionospheric density structure: Generation of auroral roar ﬁne frequency structure, J. Geophys. Res. 105, 27580, 2000. [68] Yoon, P.H. and J. LaBelle, Discrete Langmuir waves in density structure, J. Geophys. Res. 110, A11308, doi:10.1029/2005JA011186, 2005.

9 Theoretical Studies of Plasma Wave Coupling: A New Approach D.-H. Lee1 , K. Kim2 , E.-H. Kim1 , and K.-S. Kim1 1

2

Department of Astronomy and Space Science, Kyung Hee University, Yongin, Kyunggi 449-701, Korea, [email protected] Department of Molecular Science and Technology, Ajou University, Suwon, Kyunggi 443-749, Korea

Abstract. New numerical and analytical methods are applied to wave coupling in inhomogeneous plasma. It is found that the X-mode feeds energy into the upper hybrid resonance at plasma inhomogeneities oriented perpendicular to the ambient magnetic ﬁeld. The results are consistent with previous studies using other methods. When a ﬁnite pressure is introduced, the upper hybrid waves are no longer stationary and start propagating. They can form cavity modes and emit a small fraction of O (or X) waves. Limitations of the present study are the neglect of collisions and plasma pressure eﬀects which might limit the growth of the upper hybrid waves; furthermore, this study concentrates on the case for which the density gradient is perpendicular to the magnetic ﬁeld, a condition that is valid near the equator.

Key words: Mode coupling, propagation in inhomogeneous plasma, resonant absorption of X mode waves, upper hybrid resonance, cavity modes

9.1 Introduction Plasma waves become often coupled owing to inhomogeneity in space. A certain mode is reﬂected at the cutoﬀ region, and changes the polarization at the crossover region. At resonances, one mode can be converted into the other resonant mode where wave energy is irreversibly transferred into the resonances. Since mode conversion is often associated with singular solutions, the subject of plasma wave coupling has diﬃculties in both analytical and numerical aspects. For instance, analytical solutions can provide only asymptotic approximations near the resonances or approximate global solutions by adopting the WKB method [4, 5, 16]. When wave equations are strongly coupled, it becomes diﬃcult to treat the coupled equations in an exact manner. Even numerical studies based on eigenmode analysis also meet similar diﬃculties in such coupled systems. Strictly speaking, no pure eigenmodes D.-H. Lee et al.: Theoretical Studies on Plasma Wave Coupling: A New Approach, Lect. Notes Phys. 687, 235–249 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

236

D.-H. Lee et al.

exist owing to the singular behavior in wave coupling, which should limit the application of the eigenmode analysis. To avoid singularities, damping is often introduced and expressed using a complex frequency ω = ωr + iγ. For ﬁnite γ, it is expected that all wave modes will decay via this damping decrement. It should be noted that γ represents an average value resulting from a Fourier transform over inﬁnite time. However, we have interests in ﬁnite time histories in reality and the damping rates among diﬀerent modes can be diﬀerent for a ﬁnite time period. Thus, it is still an important issue to further investigate the process of plasma wave coupling. In this study, we introduce new numerical and analytical techniques, which compensate for such shortcomings mentioned above. One technique is a time-dependent numerical model, which allows multi-ﬂuid components in an arbitrarily inhomogeneous 3-D plasma. This time-dependent model can be very useful in the coupled wave problem [Lee and Lysak 12] since the mode conversion is often associated with singular solutions, which provide only asymptotic approximations. The other technique is an analytical tool, which is called the invariant imbedding method (IIM) [3]. Using the invariant imbedding equations, we are able to calculate the reﬂection and transmission coeﬃcients, and the wave amplitudes for the propagation of arbitrary number of coupled waves in arbitrarily-inhomogeneous stratiﬁed media [10]. In this study, we will show the application of these numerical and analytical techniques to one of the plasma wave coupling problems: the coupling of ordinary (O), extraordinary (X) waves, and upper hybrid resonances (UHR). Linear mode conversion of O and X waves into upper hybrid (UH) waves is investigated numerically by adopting a time-dependent numerical model, and analytically by adopting IIM, respectively.

9.2 Numerical Model We study the wave coupling at the UHR region by using a 3-D multi-ﬂuid numerical model [9, 11]. Our approach diﬀers from previous studies in the sense that plasma waves are studied with time histories of electric and magnetic ﬁeld components. In a cold plasma, the linearized electron waves can be obtained by Maxwell equations, Ohm’s law and the equation of motion. For simplicity, we consider only the motion of electron ﬂuids and assume the cold plasma approximation in this work. ∇×E=−

∂B ∂t

1 ∂E c2 ∂t J = −en0 (x)ve

∇ × B = µo J +

(9.1) (9.2) (9.3)

9 Plasma Wave Coupling

237

Fig. 9.1. (a) The upper hybrid and plasma frequency proﬁles assumed in the model. The solid line represents the local upper hybrid frequency, ωuh (x). The dashed line represents the local plasma frequency, ωp (x). (b) The applied impulse at x = 0, which is assumed on the Ez (or Ey ) component

∂ve = −e(E + ve × B0 ) (9.4) ∂t where E, B, J, v and n represent electric and magnetic ﬁelds, current density, velocity and number density, respectively. In order to solve (9.1)–(9.4) as an initial-valued problem, the ﬁnite diﬀerence method is used in time and space. In our box model, a uniform magnetic ﬁeld B0 is assumed to be parallel to the z-axis and the density gradient is introduced along the x-axis. Details of this numerical model are referred to in Kim and Lee [9] and Kim et al. [11]. The size of box used in this calculation is 102 × 102 × 102 km3 . The proﬁle of plasma frequency (ωp ) and UH frequency (ωuh ) assumed in our model is given in Fig. 9.1a. Frequencies are normalized to the electron cyclotron frequency, ωce (= 6 × 103 rad/sec). Length is normalized to the radial distance, L = 100 km. The impulsive input is assumed on Ez and Ey for O wave and X wave initial inputs, respectively, at x = 0. Fundamental harmonic wave numbers are assumed in both y and z directions. The EM impulse used in the simulations is given by Fig. 9.1b. Time histories of electric and magnetic ﬁeld components at a line of grid points along the x direction are recorded. The boundaries are assumed to become perfect reﬂectors after the impulsive stimulus ends. Thus the total wave energy inside the box model remains constant in time, which enables us to easily examine the energy transfer among diﬀerent wave modes. We start the simulation with the impulsive input on the Ez (or Ey ) component to represent the O wave (or X wave) impulse. me

9.3 Numerical Results The power spectra of electric and magnetic ﬁeld components are obtained through the FFT at each grid point. Figure 9.2 shows the spectra, where the

238

D.-H. Lee et al.

Fig. 9.2. The power spectra of perturbation components E and B when the impulse of O waves is assumed

amplitudes are represented by the degree of brightness after they are scaled logarithmically. In Fig. 9.2, in the spectrum of Ex , one continuous band appears, which corresponds to the local electrostatic (ES) upper hybrid waves in Fig. 9.1a. The Ez component can eﬀectively represent the O wave since the wave vector is almost perpendicular to B0 by assuming relatively small ky and kz . The electromagnetic (EM) waves globally appear in all E and B components since they freely propagate inside the box when an impulsive input excites EM waves. The EM modes show a few cavity modes in Fig. 9.2 owing to the perfect reﬂecting boundary conditions. This feature is well conﬁrmed in the spectra of Bx , By and Bz which are purely EM wave components. To examine the coupling between ES and EM modes at the UHR, we select one point (x1 ) at a certain UH frequency marked by X in Fig. 9.2. Figure 9.3 shows time histories of the electric and magnetic ﬁelds at this resonant point which are obtained by applying the inverse Fourier transform at the given

9 Plasma Wave Coupling

239

Fig. 9.3. The time histories of perturbation components E and B (case of O wave incidence). The electric ﬁelds are normalized to an arbitrary value EA and the magnetic ﬁelds are normalized to BA = EA /c, respectively

frequency. Figure 9.3 indicates that the ES wave grows in time, while the other EM wave damped. Therefore it is found that the EM waves excited by the O wave impulse are mode-converted into UH waves. We now consider an EM impulse of X wave by assuming the impulse above on the Ey component. Figure 9.4 shows the power spectra of E and B. In the spectrum of Ex , one continuous band appears again, which corresponds to the local UH frequency. The spectral feature in Fig. 9.4 is similar to that of Fig. 9.2 except that relatively large power is found in both Ex and Ey compared to Ez , while Fig. 9.2 shows relatively large power in Ez . This diﬀerence arises because the Ez input would produce relatively large O wave power, while the Ey would produce more X wave power inside the box. Figure 9.5 shows the time histories at the resonant point (x2 ) marked by X in Fig. 9.4. Our results show the growth of the ES wave Ex and the decay

240

D.-H. Lee et al.

Fig. 9.4. The power spectra of perturbation components E and B (case of X wave incidence)

of the other EM wave (Bx , By , Bz , Ey and Ez ) at this UH frequency, which are pretty similar to the results of the O wave impulse.

9.4 Invariant Embedding Method In recent work of Kim et al. [10], a new invariant embedding theory was presented for studying the propagation of coupled waves in inhomogeneous stratiﬁed media. We consider N coupled waves propagating in a stratiﬁed medium, where physical parameters depend on only one coordinate. We take this coordinate as the z axis and assume the medium of thickness L lies in 0 ≤ z ≤ L. We also assume that all waves propagate in the xz plane. The x component of the wave vector, q, is then a constant. In a variety of problems, the wave equation of N coupled waves has the form ! d2 ψ dE −1 dψ E (z) + E(z)K 2 M(z) − q 2 I ψ = 0 , − 2 dz dz dz

(9.5)

9 Plasma Wave Coupling

241

Fig. 9.5. The time histories of perturbation components E and B (case of X wave incidence)

where ψ = (ψ1 , · · · , ψN )T is a vector wave function and E and M are N × N matrix functions. We assume that the waves are incident from the vacuum region where z > L and transmitted to another vacuum region where z < 0. I is a unit matrix and K is a diagonal matrix such that Kij = ki δij , where ki is the magnitude of the vacuum wave vector for the i-th wave. By assigning E(z) and M(z) suitably, (9.5) is able to describe various kinds of waves in a number of stratiﬁed media. A wide variety of mode conversion phenomena observed in space and laboratory plasmas can also be studied using this equation [4, 7, 13, 16]. We generalize (9.5) slightly, by replacing the vector wave function ψ by an N × N matrix wave function Ψ , the j-th column vector (Ψ1j , · · · , ΨN j )T of which represents the wave function when the incident wave consists only of the j-th wave. We are interested in the reﬂection and transmission coeﬃcient matrices r = r(L) and t = t(L). Let us introduce a matrix ! +z T exp i z+ dz E(z )P ,! z > z g(z, z ) = ˜ (9.6) z T exp −i z dz E(z )P , z < z

242

D.-H. Lee et al.

where T and T˜ are the time-ordering and anti-time-ordering operators, respectively. When applied to a product of matrices which are functions of z, T (T˜ ) arranges the matrices in the order of increasing (decreasing) z. For instance, T [E(z2 )E(z1 )] = E(z1 )E(z2 ), if z2 > z1 . The matrix P is a diagonal matrix satisfying Pij = pi δij and pi is the negative z component of the vacuum wave vector for the i-th wave. It is easy to prove that g(z, z ) satisﬁes the equations ∂ g(z, z ) = i sgn(z − z ) E(z)P g(z, z ), ∂z ∂ g(z, z ) = −i sgn(z − z ) g(z, z )E(z )P . ∂z

(9.7)

Using (9.6) and (9.7), the wave equation (9.5) is transformed to an integral equation i L Ψ (z, L) = g(z, L) − dz g(z, z ) 2 0 ! × E(z )P − P M(z ) − q 2 P −1 M(z ) + q 2 P −1 E −1 (z ) Ψ (z , L) . (9.8) We take a derivative of this equation with respect to L and obtain ∂Ψ (z, L) = iΨ (z, L)α(L) , ∂L

(9.9)

where 1 α(L) = E(L)P − Ψ (L, L) 2 ! × E(L)P − P M(L) − q 2 P −1 M(L) + q 2 P −1 E −1 (L) .

(9.10)

Taking now the derivative of Ψ (L, L) with respect to L, we obtain dΨ (L, L) = iE(L)P [r(L) − I] + iΨ (L, L)α(L) . dL

(9.11)

Since Ψ (L, L) = I + r(L), we ﬁnd the invariant embedding equation satisﬁed by r(L): dr i = i [r(L)E(L)P + E(L)P r(L)] − [r(L) + I] dL 2 ! × E(L)P − P M(L) − q 2 P −1 M(L) + q 2 P −1 E −1 (L) [r(L) + I] . (9.12) Similarly by setting z = 0 in (9.9), we ﬁnd the invariant embedding equation for t(L): dt i = it(L)E(L)P − t(L) dL 2 ! × E(L)P − P M(L) − q 2 P −1 M(L) + q 2 P −1 E −1 (L) [r(L) + I] . (9.13)

9 Plasma Wave Coupling

243

These equations are supplemented with the initial conditions, r(0) = 0 and t(0) = I. We solve the coupled diﬀerential equations (9.12) and (9.13) numerically using the initial conditions and obtain the reﬂection and transmission coeﬃcient matrices r and t as functions of L. The invariant embedding method can also be used in calculating the ﬁeld amplitude Ψ (z) inside the medium. Rewriting (9.9), we get i ∂Ψ (z, l) = iΨ (z, l)E(l)P − Ψ (z, l) ∂l 2 ! × E(l)P − P M(l) − q 2 P −1 M(l) + q 2 P −1 E −1 (l) [r(l) + I] . (9.14) For a given z (0 < z < L), the ﬁeld amplitude Ψ (z, L) is obtained by integrating this equation from l = z to l = L using the initial condition Ψ (z, z) = I + r(z).

9.5 Application of IEM to the Mode-Conversion of O and X Waves Equations (9.12), (9.13) and (9.14) are the starting point in our exact analysis of a variety of wave coupling and mode conversion phenomena. In the rest of this paper, we demonstrate the utility of our invariant embedding equations by applying them to the high frequency wave propagation and mode conversion in cold, magnetized plasmas. We assume that the plasma density varies only in the z direction and the external magnetic ﬁeld B0 (= (B0 sin θ, 0, B0 cos θ)) is directed parallel to the xz plane and makes an angle θ with the z axis. The cold plasma dielectric tensor, , for high frequency waves in the present geometry is written as 1 + (3 − 1 ) sin2 θ i2 cos θ (3 − 1 ) sin θ cos θ , (9.15) 1 i2 sin θ = −i2 cos θ 2 (3 − 1 ) sin θ cos θ −i2 sin θ 1 + (3 − 1 ) cos θ where 1 = 1 −

2 =

ωp2 (ω + iν) 2

ω (ω + iν) − ωc2 ωp2 ωc 2

ω (ω + iν) − ωc2

3 = 1 −

ωp2 . ω (ω + iν)

,

,

(9.16)

The constant ν is the phenomenological collision frequency. The spatial inhomogeneity of plasmas enters through the z dependence of the electron number density n.

244

D.-H. Lee et al.

For monochromatic waves of frequency ω, the wave equation satisﬁed by the electric ﬁeld in cold magnetized plasmas has the form ∇2 E +

ω2 ·E=0. c2

(9.17)

In this paper, we restrict our interest to the case where plane waves propagate parallel to the z axis. In this situation, we can eliminate Ez from (9.17) and obtain two coupled wave equations satisﬁed by Ex = Ex (z) and Ey = Ey (z), which turn out to have precisely the same form as (9.5) with q = 0 and

ω Ex ψ= , K = I, E = I, Ey c 1 M= 1 + (3 − 1 ) cos2 θ

1 3 i2 3 cos θ ! × . (9.18) −i2 3 cos θ 1 3 − (3 − 1 ) 1 + 22 sin2 θ We use (9.12), (9.13) and (9.18) in calculating the reﬂection and transmission coeﬃcients. In our notation, r11 (r21 ) is the reﬂection coeﬃcient when the incident wave is Ex (that is, linearly polarized in the x direction) and the reﬂected wave is Ex (Ey ). Similarly, r22 (r12 ) is the reﬂection coeﬃcient when the incident wave is Ey (that is, linearly polarized in the y direction) and the reﬂected wave is Ey (Ex ). Similar deﬁnitions are applied to the transmission coeﬃcients. The reﬂectances and transmittances are deﬁned by Rij = |rij |2 and Tij = |tij |2 . When the dielectric permittivities of the incident region and the transmitted region are the same, we can calculate the wave absorption by Aj ≡ 1 − R1j − R2j − T1j − T2j (j = 1, 2). If a mode conversion occurs, this quantity is nonzero even in the limit where the damping constant ν goes to zero. We will call Aj the mode conversion coeﬃcient. For a speciﬁc calculation, we assume that the electron density proﬁle is given by z 2 4 n(z) = 10 9 1 − + 1 m−3 (9.19) L where the plasma thickness L is equal to 105 m. The electromagnetic wave is incident from the vacuum region where z > L and transmitted to the vacuum region where z < 0. We also assume that the electron cyclotron frequency is equal to ωc = 6 × 103 rad/s. In Fig. 9.6, we plot the reﬂectances R11 , R12 (= R21 ), R22 , the transmittances T11 , T12 , T21 , T22 and the mode conversion coeﬃcients A1 , A2 as functions of the normalized frequency ωL/c, when a wave is incident perpendicularly on the stratiﬁed plasma. The external magnetic ﬁeld is directed at θ = 45◦ from the z axis. The resonance associated with the extraordinary wave component occurs for ωe1 ≤ ω ≤ ωe2 where ωe1 and ωe2 are the minimum and maximum values of

9 Plasma Wave Coupling

ω

245

ω

1

ω Fig. 9.6. Reﬂectances, transmittances and the mode conversion coeﬃcients when the external magnetic ﬁeld is directed at 45◦ from the z axis and the wave is incident parallel to the z axis. The mode conversion coeﬃcients are nonzero for ωe1 ≤ ω ≤ ωe2 , where ωe1 and ωe2 are the minimum and maximum values of ωe deﬁned by (9.20) and for ωo1 ≤ ω ≤ ωo2 , where ωo1 and ωo2 are the minimum and maximum values of ωo deﬁned by (9.21)

246

D.-H. Lee et al.

( ωe ≡

1 12 ωp2 + ωc2 4ωp2 ωc2 cos2 θ 2 . 1+ 1− 2 2 ωp2 + ωc2

(9.20)

Both A1 and A2 show a broad and sizable peak in this frequency range. The resonance associated with the ordinary wave component occurs for ωo1 ≤ ω ≤ ωo2 where ωo1 and ωo2 are the minimum and maximum values of ( ωo ≡

ωp2

+ 2

ωc2

12 12 2 2 2 4ωp ωc cos θ . 1− 1− 2 ωp2 + ωc2

(9.21)

A1 and A2 show a tiny peak in this frequency range. In Fig. 9.6, it is evident that ES waves absorb EM wave energy via the resonant absorption at (9.20) and (9.21). These ES waves become the UHR when θ ≈ π/2 and the plasma oscillations θ ≈ 0, respectively. Thus A1 (A2 ) denotes the mode conversion coeﬃcient from linearly polarized EM waves of Ex (Ey ) to the ES resonant modes. It should be noted that these absorptions are not aﬀected by the boundaries between plasma and vacuum since both A1 and A2 disappear when inhomogeneity is excluded and uniform plasmas are assumed in the same region. In Fig. 9.7, we plot the dependence of the mode conversion coeﬃcients on the angle between the external magnetic ﬁeld and the z axis. Both for A1 and A2 , the ordinary wave absorption, the magnitude of which is quite small, is largest when θ = 0 and decreases monotonically as θ increases. The frequency ranges where this absorption occurs agree quite well with ωo1 and ωo2 given by (9.21). The extraordinary wave absorption for A2 is zero when θ = 0 and increases monotonically as θ increases. The frequency ranges where this absorption occurs are designated by square dots, the positions of which agree precisely with ωe1 and ωe2 given by (9.20). For A1 , the extraordinary wave absorption is zero for θ = 90◦ , as well as for θ = 0. In Fig. 9.8, we plot the mode conversion coeﬃcients for two diﬀerent normalized damping coeﬃcients νL/c = 0.01 and 0.0001 when θ = 45◦ .

9.6 Discussion and Summary Our results of both numerical and analytical methods presented above show that both O and X waves are likely to give energy into the UHR when the inhomogeneity lies perpendicular to the ambient magnetic ﬁeld. It is suggested that the EM electron waves are mode-converted to the ES electron waves via the resonant absorption at the UH frequency. Our simulation results are found to be also consistent with previous studies such as investigation of the Budden problem by White and Chen [19], Grebogi et al. [6], Lin and Lin [14], Antani et al. [1, 2], and Ueda et al. [18].

9 Plasma Wave Coupling

θ=0

θ=0

θ=15

θ=15

θ=30

θ=30

θ=45

θ=45

θ=60

θ=60

θ=75

θ=75

θ=90

θ=90

ω

247

ω

Fig. 9.7. Dependence of the mode conversion coeﬃcients on the angle between the external magnetic ﬁeld and the z axis. Square dots represent the positions of ωe1 and ωe2

However, there are several limitations of our calculations. We neglect the eﬀects of collisions and plasma pressure which might limit the growth of UHR amplitude [White and Chen, 19]. In addition, when the ﬁnite pressure is introduced, the ES upper hybrid waves are able to propagate and no longer stationary. Under certain circumstances such as the density striations these ES waves can form the cavity modes and emit a small fraction of O (or X) waves into space [8, 15, 17, 20]. In this study the density gradient is assumed to be perpendicular to the background magnetic ﬁeld. This condition is valid probably only at the equatorial region and it should be extended to the case of arbitrary angles with respect to the Earth’s magnetic ﬁeld in the high-latitude region. Recent

248

D.-H. Lee et al.

ν

ν

ν

ν

ω

ω

Fig. 9.8. Dependence of the mode conversion coeﬃcients on the wave damping parameter ν

analytical methods such as the invariant embedding method of Kim et al. [10] as well as our time-dependent 3-D simulations in Kim et al. [11] enable us to quantitatively investigate such general wave coupling problems. This subject will be left as future work.

Acknowledgments This work was supported by the Korea Science and Engineering Foundation grant R14-2002-043-01000-0 and in part by R14-2002-062-01000-0.

References [1] Antani, S.N., N.N. Rao, and D.J. Kaup: Geophys. Res. Lett. 18, 2285 (1991). [2] Antani, S.N., D.J. Kaup, and N.N. Rao: J. Geophys. Res. 101, 27,035 (1996). [3] Bellman, R. and G.M. Wing: An Introduction to Invariant Imbedding (Wiley, New York, 1976). [4] Budden, K.G.: The Propagation of Radio Waves (Cambridge, Cambridge, 1985). [5] Ginzburg, V.L.: Propagation of electromagnetic waves in plasmas (Pergamon Press, New York, 1964). [6] Grebogi, C., C.S. Liu, and V.K. Tripathi: Phys. Rev. Lett. 39, 338 (1977). [7] Hinkel-Lipsker, D.E., B.D. Fried, and G.J. Morales: Phys. Fluids B 4, 559 (1992). [8] Hughes, J.M. and J. LaBelle: Geophys. Res. Lett, 28, 123 (2001).

9 Plasma Wave Coupling

249

[9] Kim, E.-H. and D.-H. Lee: Geophys. Res. Lett. 30, 2240 (2003). [10] Kim, K., D.-H. Lee, and H. Lim: Europhys. Lett. 69, to appear (2005). [11] Kim, K.-S., D.-H. Lee, E.-H. Kim, and K. Kim: Geophys. Res. Lett., submitted (2005) [12] Lee, D.-H. and R.L. Lysak: J. Geophys. Res. 94, 17,097 (1989). [13] Lee, D.-H., M.K. Hudson, K. Kim, R.L. Lysak, and Y. Song: J. Geophys. Res. 107, 1307 (2002). [14] Lin, A.T. and C.C. Lin: Phys. Fluids 27, 2208 (1984). [15] Shepherd, S.G., J. LaBelle, and M.L. Trimpi: Geophys. Res. Lett. 24, 3161 (1997). [16] Swanson, D.G.: Theory of Mode Conversion and Tunneling in Inhomogeneous Plasmas (Wiley, New York, 1998). [17] Weatherwax, A.T., J. LaBelle, M.L. Trimpi, R.A. Treumann, J. Minow, and C. Deehr: J. Geophys. Res. 100, 7745 (1995). [18] Ueda, H.O., Y. Omura, and H. Matsumoto: Ann. Geophysicae 16, 1251 (1998). [19] White, R.B. and F.F. Chen: Plasma Phys. 16, 565 (1974). [20] Yoon, P.H., A.T., Weatherwax, and T.J. Rosenberg: J. Geophys. Res. 103, 4071 (1998).

10 Plasma Waves Near Reconnection Sites A. Vaivads1 , Yu. Khotyaintsev1 , M. Andr´e1 , and R.A. Treumann2,3 1

2

3

Swedish Institute of Space Physics, Uppsala, Sweden [email protected], [email protected], [email protected] Geophysics Section, The University of Munich, Munich, Germany [email protected] Department of Physics and Astronomy, Dartmouth College, Hanover, NH, USA

Abstract. Reconnection sites are known to be regions of strong wave activity covering a broad range of frequencies from below the ion gyrofrequency to above the electron plasma frequency. Here we explore the observations near the reconnection sites of high frequency waves, frequencies well above the ion gyrofrequency. We concentrate on in situ satellite observations, particularly on recent observations by the Cluster spacecraft and, where possible, compare the observations with numerical simulations, laboratory experiments and theoretical predictions. Several wave modes are found near the reconnection sites: lower hybrid drift waves, whistlers, electron cyclotron waves, Langmuir/upper hybrid waves, and solitary wave structures. We discuss the role of these waves in the reconnection onset and supporting the reconnection, in anomalous resistivity and diﬀusion, as well as a possibility for using these waves as a tool for remote sensing of reconnection sites.

Key words: Reconnection, Hall region, separatrix physics, lower hybrid waves, whistlers, high-frequency waves, electron holes, anomalous resistance, structure of reconnection region, wave signatures

10.1 Background Collisionless magnetic reconnection in space plasma has the two important properties of converting the available free magnetic energy into kinetic energy of charged particles in large regions of space and causing signiﬁcant mass and energy transfer across the boundaries that separate the interacting plasmas. The regions where the energy conversion takes place, e.g. the auroral zone, ionosphere, shocks, etc., emit waves and generate turbulence over a wide frequency range. The occurrence of reconnection is not exceptional. To the contrary, reconnection is abundant in collisionless plasmas taking place everywhere where suﬃciently thin current sheets are generated. Understanding the role of waves and turbulence in the energy conversion, energy transA. Vaivads et al.: Plasma Waves Near Reconnection Sites, Lect. Notes Phys. 687, 251–269 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

252

A. Vaivads et al.

port, and structure formation of the reconnection sites is thus an important and challenging task. Astrophysical environments allow studies of reconnection regions only from remote by observing the emitted electromagnetic radiation. For instance, the electron beams which cause solar and interplanetary type III radio bursts at the local plasma frequency and its harmonics are believed to originate from regions near reconnection sites in the solar corona [see, e.g., Cane et al., 9]. Remote studies of the emission provide solely average information about the reconnection sites with the averaging proceeding over large spatial volumes. They therefore suﬀer from severe limitations on the spatial resolution. As a consequence the information obtained about local conditions and micro processes in the reconnection region is very limited and in most cases no information can be extracted at all. The only places where reconnection sites can be studied in detail are in the laboratory and the Earth’s magnetosphere or other accessible environments in our solar system that have been visited by spacecraft, such as the solar wind, some of the other magnetized planets, comets, and the outer heliosphere. Spacecraft observations give a much more detailed picture of the plasma dynamics than any of the laboratory experiments. This is due mainly to the possibility of resolving the particle distribution functions and electromagnetic ﬁelds at small scales, in some cases down to the smallest electron scales. In the laboratory, in addition, it is practically impossible to reproduce the collisionless and dilute conditions at the temperatures prevailing in space and astrophysical plasmas. Observations in space are thus uniquely suited for the understanding of reconnection. However, since reconnection involves many processes at diﬀerent spatial and temporal scales, numerical simulations serve as a superior tool for understanding the environment and physical processes in the vicinity of reconnection sites. The two main regions in the Earth’s magnetosphere where the processes of reconnection have so far been studied are the subsolar magnetopause and the magnetotail current sheet. Under normal conditions in those regions the plasma is overdense, with fpe fce . Reconnection in the magnetotail proceeds in a relatively symmetric way, in the sense that the plasmas to both sides of the tail current sheet have similar properties. Quite an opposite situation is encountered at the magnetopause. Here reconnection is manifestly asymmetric. Another important diﬀerence between magnetopause and tail reconnection is that the typical spatial scales, e.g. the ion inertial length and the ion gyro radius, are usually a factor of ten smaller at the magnetopause than in the tail. This is important for any in situ studies where the instrumental resolution enters as a limiting factor. Altogether, a signiﬁcant number of studies deal with high frequency waves at the magnetopause and in the magnetotail, but only in a few cases have attempts been made to ﬁnd a direct relation between the observed waves and the reconnection process, even though in some cases the existence of such a relationship has been put forth. In the present paper we summarize in situ observations of high frequency waves under conditions when reconnection signatures are well deﬁned as well

10 Plasma Waves Near Reconnection Sites

253

as observations where one merely speculates about a relation of the observed waves to possible reconnection going on at a distance from the location of the observations. Figure 10.1 shows a sketch of the reconnection site. Two oppositely directed magnetic ﬁelds in the inﬂow regions merge in the diﬀusion region forming an X-line. The magnetic ﬁeld lines connected to the X-line are the separatrices. We call the regions close to the separatrices separatrix regions. Plasma containing reconnected magnetic ﬁelds is ﬂowing away from the X-line in the outﬂow regions, where it escapes in the form of jets. This sketch draws a rather simpliﬁed two-dimensional picture of the reconnection process. In reality, the reconnection site has a considerably more complicated structure, consisting of multiple X-lines and exhibiting a complex and three-dimensional conﬁguration. However, in many cases the simple 2D picture may serve as a lucid and suﬃcient approximation to a reconnection site. A counterexample is the complex structure arising from spontaneous antiparallel reconnection where patchy X-lines form within a narrow current sheet. On the other hand, when a small guide ﬁeld is added, well ordered X-lines develop, and the 2D description can become suﬃciently accurate to serve as an approximate description of the reconnection process [see, e.g., Scholer et al., 33]. Another general property of a single reconnection site is its pronounced inhomogeneity rendering almost all homogeneous plasma models of scales of the order of the spatial extension of the reconnection site invalid. Density and temperature gradients as well as non-Maxwellian particle distribution functions in the vicinity of a reconnection site result in the generation of various plasma wave modes. All of them contribute to the processes of particle acceleration and energy redistribution, generation of transport coeﬃcients, and the wanted diﬀusion of magnetic ﬁeld and plasma near the X-line which is necessary in order to maintain the merging of the oppositely directed magnetic ﬁelds. Space observations allow for the direct in situ observation of the wavegeneration and wave-particle interactions in these merging processes at the reconnection sites. 10.1.1 Observations of Diﬀerent Wave Modes Some of the high frequency wave modes which are usually observed in the vicinity of a reconnection site are sketched in Fig. 10.1. The locations where these modes are observed are rather speculative because there is very limited knowledge so far on the relative locations of the diﬀerent modes. Emissions with the strongest electric ﬁelds are ordinarily detected near or below the lower hybrid frequency fLH , indicating the presence of lower hybrid drift waves (LHD). These emissions have both strong electric and magnetic ﬁeld components. They seem to assume their highest amplitudes just near the steep density gradients like the ones found in the separatrix region. In general the separatrix region seems to be the location of very strong wave emissions of several diﬀerent wave modes in a wide frequency range.

254

A. Vaivads et al.

Fig. 10.1. Sketch of the reconnection site. Separatrices with density gradients are marked. A the top diﬀerent kinds of wave spectra are sketched that are commonly observed near reconnection sites. Some common places to observe those waves are marked in diﬀerent gray shadowing. Typical electron distribution functions in the vicinity of the separatrix are indicated as well

Strong electric ﬁelds are found around the electron plasma frequency fpe . These emissions are usually believed to be Langmuir (L) waves or, if oblique, upper hybrid (UH) waves. Often, electrostatic solitary waves (ESW) can also be associated with reconnection. Whistler emissions (W) are identiﬁed from narrow spectral peaks in the frequency range fLH < f < fce between the lower hybrid frequency and the local electron gyro-frequency fce . Despite the applicability of such waves to astrophysical environments, there have been very few observations of radio emissions from reconnection sites at frequencies around and above the electron plasma frequency even though one expects their presence in the separatrix region where they should be generated by the fast electron beams escaping from the X line. This is probably due to the weakness of the electromagnetic signals in comparison with the electrostatic waves. In the rest of this article we summarize the in situ observations of the above mentioned wave modes and discuss their possible generation mechanisms and relations to reconnection. Before doing so we also summarize some of the results obtained from numerical simulations dealing with high frequency waves or being relevant to the discussion in this paper.

10 Plasma Waves Near Reconnection Sites

255

10.2 Numerical Simulations The generation of high frequency waves involves the dynamics of electron and electron kinetic eﬀects which in most cases are very important. Therefore, fullparticle and Vlasov simulations are best qualiﬁed for this kind of study. Such simulations are also computationally expensive. So far relatively few wave modes have been addressed. An additional diﬃculty is that in some cases, for example in the simulation of lower hybrid drift waves, it is absolutely essential to simulate the phenomena in all three spatial dimensions. The available numerical simulations indicate that the separatrix regions are the most dynamically active regions [see, e.g., Cattell et al., 10]. Intense electron beams are generated in the reconnection process along the separatrices by parallel electric ﬁelds which are distributed along the separatrices as shown by Pritchett [30] and Hoshino et al. [17]. In addition, electron conics and shell distributions can form due to diverging magnetic ﬂux tubes close to the reconnection site. High-frequency wave modes that have been studied in detail by numerical simulations are the following: • Lower hybrid drift waves (LHD). It has been suggested that these waves play a crucial role in the narrowing of the current sheets with the subsequent onset of reconnection [see, e.g., 11, 33, 35]. LHD waves tend to be electrostatic (δE/δB > VA ). Their largest amplitudes are found at the edges of the current sheet. They interact eﬃciently with both electrons and ions and can cause a signiﬁcant anomalous resistivity and corresponding anomalous diﬀusion. It has been realized that these waves possess a signiﬁcant magnetic component in the center of the current sheet that also contributes to anomalous resistivity [Silin et al., 35]. • Solitary waves (SW) and double layers (DL). Solitary wave and double layer generation due to electron beams have been studied in great detail [e.g., Omura et al., 29]. Electron beams form mainly close to the separatrices as shown by Hoshino et al. [17] and Pritchett and Coroniti [31] and particularly under guide ﬁeld conditions. In this case strong double layers can be generated at the reconnection site [Drake et al., 13]. Solitary structures and lower hybrid waves may couple as well, as has been found in the same simulations. High frequency waves that have been studied in numerical simulations much less frequently. The main types investigated are: • Whistlers. Numerical simulations show that the Hall term in the Generalized Ohm’s law is important for the onset of fast reconnection [Birn et al., 8]. The Hall term also introduces whistlers into the system, and it has been speculated that the region close to the reconnection site is some kind of a standing whistler. Observations indicate that electromagnetic modes in the whistler frequency range are observed also at distances far from the reconnection site, but this type of emission has received very little attention in numerical studies.

256

A. Vaivads et al.

• Waves at the plasma frequency. While observations indicate the presence of strong Langmuir/upper hybrid waves in the separatrix regions, these waves have not been studied in any of the reconnection simulations. • Electron cyclotron waves. Similarly, observations indicate the presence of electron cyclotron waves and speculate about their relation to reconnection, but in the numerical simulations they have not been studied yet. • Radio emissions. Free-space modes are usually generated at the local plasma frequency or above. They can freely propagate out into space. They are of primary importance in the astrophysical application of reconnection. There exists a large theoretical eﬀort in studying these waves but the amount of numerical simulations is very limited and has mainly been done for astrophysical plasma conditions.

10.3 Lower Hybrid Drift Waves 10.3.1 Observations Large wave-electric ﬁelds at the magnetopause and in the magnetotail are usually observed at frequencies near the local lower-hybrid frequency fLH [3, ˜ ≥ 1, where 10]. The strongest peak-to-peak amplitudes are of the order δ E ˜ δ E is the normalized root mean square electric ﬁeld. At the magnetopause the waves are strongest on the magnetospheric side where the inﬂow Alfv´enic speed is high. Often, these waves are located in those regions (narrow sheets with spatial scale less than an ion gyroradius) where the DC electric ﬁeld reaches high values, about half of the peak-to-peak amplitude of the waves. At this time there is no statistical analysis available of the occurrence rates of these waves, but it seems that the highest-amplitude lower-hybrid waves are located in the regions of steepest density gradients. Numerical simulations and analytical calculations suggest that the observed waves are lower-hybrid drift (LHD) waves even though the modiﬁed two-stream instability could generate waves with similar properties. An example of LHD wave observations is shown in Fig. 10.2. Particular characteristics of LHD waves are their short wavelengths, kρe ∼ 1, a broadband spectrum extending from frequencies well below to well above fLH , perpendicular wave numbers k⊥ k , a phase velocity of the order of the ion thermal velocity, and a coherence length of the order of one wavelength. The observations also suggest that the wave potential can be close to the electron thermal energy [Bale et al., 6]. Based on spectral and interferometric results, some satellite observations support the LHD-interpretation [Vaivads et al., 38]. However, more detailed studies are required in order to conﬁrm the lower-hybrid drift nature of the observed waves. Observations also show that these waves have a signiﬁcant magnetic component with c δE/δB vA and can have a preferential direction of propagation along the ambient magnetic ﬁeld [Vaivads et al., 38].

10 Plasma Waves Near Reconnection Sites

257

Fig. 10.2. Example of LHD wave observations near the separatrices far from the reconnection X-line. The ﬁgure is adopted from Vaivads et al. [38] and Andr´e et al. [4]. (a), (b): Large-scale density (obtained from the satellite potential) and magnetic ﬁeld observations. The rest of the panels are zoomed-in on the time interval marked bright when the strongest electric ﬁelds were observed. This time is identiﬁed as a possible separatrix region [Andr´e et al., 4]. (c): Electron spectrogram showing electron beams along B. (d): Electric ﬁeld normal and tangential to the magnetopause. (e): Plasma density. (f): Magnetic ﬁeld spectra, and (g): Electric ﬁeld spectra (Reprinted with permission of American Geophysical Union)

10.3.2 Generation Mechanisms In a simpliﬁed picture, the driving force for the LHD waves is a density gradient with relative ﬂow between electrons and ions due to their diﬀerent diamagnetic drifts. In the case of the modiﬁed-two stream instability (MTSI) it is the cross-ﬁeld drift of the electrons with respect to the ions that presents the driving force. Strong DC electric ﬁelds on scales smaller than ion gyroradius are almost always seen in association with LHD waves. From the energetic point of view it has been shown that the observed waves can also be generated

258

A. Vaivads et al.

by the electron beams present at the density gradients [Vaivads et al., 37]. It is well-known that in places like the auroral zone electron beams do indeed generate intense lower-hybrid waves. 10.3.3 Relation to Reconnection Reconnection is one of the mechanisms that is capable of producing strong and narrow density gradients in space. Density gradients are formed along the separatrices which separate the inﬂow from the outﬂow regions. The separatrices contain local density dips which have been observed in numerical simulations [see, e.g., 34] . Numerical simulations and observations also suggest that the separatrices can maintain their steep density gradient structure over distances very far away from the reconnection site (tens of λi ). It has not yet been explained properly why these density dips exist. Such dips can, however, evolve when a static ﬁeld-aligned potential is applied along the separatrix which evacuates part of the plasma locally. The separatrices are also regions where strong electron beams are present [Hoshino et al., 17]. Such beams have a longitudinal pressure anisotropy and therefore are hardly capable of excluding plasma from the separatrix regions. A probable cause is a magnetic-ﬁeld-aligned electric ﬁeld which accelerates the electrons into a beam thereby evaporating the plasma. Nevertheless, it remains unclear which mechanism maintains the required pressure balance. The role of these beams in the generation of waves has not been fully explored. It is not clear, moreover, which other mechanisms besides reconnection could produce such narrow (few λi ) density gradients. There are speculative ideas, like “peeling” or “snow-plowing” due to, e.g., FTEs, but no clear understanding exists as yet of how such a process would work. Thus, while we expect strong LHD waves near the reconnection site along the separatrices it is not yet clear whether all or most of the intense LHD-wave observations are related to ongoing reconnection. Lower hybrid waves can aﬀect reconnection in several important ways: • Through anomalous resistivity: Usually, in simulations and observations the most intense LHD waves generate anomalous collision frequencies for e− ∼ 2πfLH [Silin et al., 35]. electrons of the order of νan • Through electron acceleration: While the phase velocity of lower hybrid waves in the direction perpendicular to the magnetic ﬁeld B is comparable to the ion thermal speed, the phase velocity along the magnetic ﬁeld becomes comparable to the electron thermal velocity thus enabling the LHD waves to resonate with thermal electrons and eﬃciently accelerate these electrons. For the same reason electron beams can be eﬃcient in generating lower hybrid waves by the inverse resonance process. • Through current sheet bifurcation, thinning and reconnection onset: Numerical simulations have shown that LHD waves apparently play a crucial role in the reconnection onset within thin current sheets [11, 33].

10 Plasma Waves Near Reconnection Sites

259

They evolve due to the steep plasma gradients at the current sheet boundary and in fact tend to broaden the current sheet. When propagating into the sheet they contribute to anomalous resistivity, heating and the diﬀusivity necessary for reconnection. This issue is controversial and has not been settled.

10.4 Solitary Waves and Langmuir/Upper Hybrid Waves 10.4.1 Observations Other quasi-stationary structures containing large electric ﬁelds that have been observed close to the reconnection sites are electrostatic solitary waves (ESW), broad-band electrostatic noise (BEN) and Langmuir/upper hybrid waves [e.g., Deng et al., 12]. They all tend to appear approximately in the same region and have similar amplitudes, which are usually several or many times smaller than the amplitudes of the LHD waves. Part of BEN observations are due to ESW passing over the spacecraft thus giving rise to a broadband spectrum. Observations show that the strongest emissions are observed along the separatrices [10, 14, 38]. The emissions change their character rapidly but it seems that narrow-band emissions at the Langmuir frequency and broadband emissions do not appear simultaneously [Khotyaintsev et al., 22]. So far only a rough comparison has been possible between the electron distribution functions and the wave characteristics [Deng et al., 12]. In the magnetotail these waves are usually located at the boundary between lobe and plasmasheet. It has been suggested that they are related to the reconnection process [Kojima et al., 24]. In order to distinguish Langmuir waves from upper hybrid waves one must study the polarization of the waves. It has been found that near the electron plasma frequency the wave electric ﬁelds are often polarized at large angles with respect to the ambient magnetic ﬁeld indicating that the observed waves are upper hybrid waves [12, 14, 21, 23] rather than Langmuir waves. Examples of observations are shown in Fig. 10.3. 10.4.2 Generation Mechanisms It is known from numerical simulations, [e.g. Omura et al., 29], that Langmuir modes are usually driven by the weak-beam instability. ESW can be the result of saturation of the nonlinear bump-on-tail instability or the two-stream instability. Upper hybrid waves can also be generated by beams, however they can also be generated by loss cone and shell distributions. Such distributions form preferentially in diverging magnetic ﬁelds either close to the reconnection site or when magnetic ﬂux tubes approach the Earth.

260

A. Vaivads et al.

Fig. 10.3. Example of wave observations near the separatrices of a reconnection X-line. (a, b) Monochromatic waves which can be interpreted as Langmuir waves when E E⊥ and as upper hybrid waves when E⊥ E . (c) Example of a mixture between electrostatic solitary waves and Langmuir waves. (d) Electrostatic solitary waves [ﬁgure adapted from Deng et al., 12]

10.4.3 Relation to Reconnection One of the major questions of reconnection is how parallel electric ﬁelds are distributed near the reconnection site. These ﬁelds are required to create changes in the magnetic ﬁeld-line topology that is associated with reconnection. ESW with a net potential drop can be one source for parallel electric ﬁelds. At the same time ESW and L/UH waves interact eﬃciently with electrons and generate high energy tails on the electron distribution functions. ESW also contribute to the anomalous resistivity in ﬁeld-aligned currents.

10 Plasma Waves Near Reconnection Sites

261

10.5 Whistlers 10.5.1 Observations There are many observations related to whistler emission in regions that are directly related to reconnection, such as the plasma sheet boundary layer [see, e.g., Gurnett et al., 16] and the magnetopause [LaBelle and Treumann, 26] and more recently [Stenberg et al., 36]. Whistler wave modes can be identiﬁed on the basis of the observed frequency of the waves (in between the electron and ion gyro-frequencies), the presence of a strong magnetic component that can reach ∼0.1 nT [Gurnett et al., 16], or the direct measurement of wave polarization [Zhang et al., 39]. Narrow spectral peaks in a wide region of frequencies between the lower hybrid and electron cyclotron frequencies are typical for these waves. However, broadband spectra extending over whistler frequencies are also often observed [LaBelle and Treumann, 26]. Such waves can consist of whistlers or they can be associated with the magnetic component of lower hybrid drift waves. Figure 10.4 gives an example of wave measurements in the high-altitude magnetopause/cusp region showing indications that the reconnection process proceeds in a high-beta plasma [Khotyaintsev et al., 22]. 10.5.2 Generation Mechanisms In addition to the temperature anisotropy or loss-cone instabilities by which it is conventionally known, whistlers can be excited as a consequence of the perpendicular anisotropy of the electron distribution function, or ion beams [Akimoto et al., 1]. Electron beams [Zhang et al., 39] have also been suggested as possible generators for whistlers under conditions when the loss cone is absent and the plasma is isotropic. This becomes possible since electrons or ions of suﬃciently high speed can undergo resonance with whistlers which not only Landau damps but, under certain circumstances, also excites whistlers. In this spirit it has been shown that whistler waves in the magnetotail are most probably generated by electron beams through the cyclotron resonance and not through the temperature anisotropy or other non-beam instability mechanisms [Zhang et al., 39]. The estimated resonance energy of the electron beam is about 10 keV suggesting reconnection as the most probable source of the beam. The instability mechanism producing broad-band magnetic turbulence is less clear. Laboratory experiments are consistent with the modiﬁed two-stream instability producing the emissions, since the waves propagate in the same direction of the electron ﬂow [Ji et al., 20]. Part of these emissions can be associated with the magnetic component of lower hybrid drift waves. 10.5.3 Relation to Reconnection The most direct evidence that whistlers can play a crucial role in the reconnection process comes from laboratory experiments which show that the reconnection rate correlates with the amplitude of obliquely propagating broad-band

262

A. Vaivads et al.

Fig. 10.4. Example of high frequency wave observations in the high-altitude magnetopause/cusp [adopted from Khotyaintsev et al., 22]. (a, b): The magnetic ﬁeld magnitude and components, (c): convection velocity E × B, (d): plasma beta, (e): spectra of the electric ﬁeld in the 2–80 kHz range, (f): spectra of the magnetic ﬁeld in the 8–4000 Hz range, and electron-cyclotron frequency, (g): polarization of wave magnetic ﬁeld with respect to the ambient magnetic ﬁeld. Whistlers are identiﬁed from right-hand polarization

10 Plasma Waves Near Reconnection Sites

263

whistler waves inside the reconnecting current sheet [cf., Ji et al., 20]. Also, numerical simulations suggest that the magnetic ﬂuctuations in the whistler band can cause a signiﬁcant anomalous resistivity [Silin et al., 35]. Such observations are yet to be conﬁrmed by the observations of reconnection in space. So far, space observations indicate that whistlers are associated with the reconnection processes [Stenberg et al., 36] and most probably are related to electron beams generated during reconnection [Zhang et al., 39], however much more detailed studies are required to conﬁrm that the whistlers are not related secondarily but play an essential role in the process of reconnection itself. An important aspect of whistlers is their ability to propagate over large distances away from the reconnection site without appreciable damping. This property makes whistlers a perfect tool for use in remote sensing of reconnection sites. In addition, whistlers by this particular property may transport information from the reconnection site to other places in the plasma.

10.6 Electron Cyclotron Waves 10.6.1 Observations Electron cyclotron waves are commonly observed in the inner regions of the magnetosphere, i.e. the polar cusp, auroral zone, and plasmasphere. The observations from the outer magnetosphere are not as abundant. Electron cyclotron waves have been observed in association with ﬂux transfer events [2, 25] and the emerging of energetic plasma in the magnetotail [Gurnett et al., 16]. Both electrostatic and electromagnetic cyclotron waves have been observed at the magnetopause (Anderson et al. 1982). Observations in the cusp and close to the magnetopause indicate that electron cyclotron waves tend to be generated on open ﬁeld lines [Menietti et al., 28]. In addition, observations show that there can be a close correspondence between observations of electron cyclotron waves and solitary waves [Menietti et al., 28]. 10.6.2 Generation Mechanisms The simplest way to excite electron cyclotron waves is again by transverse temperature anisotropies or loss cones which excite diﬀuse electron cyclotron waves between the harmonic bands. This has been recognized early. Purely transverse electron cyclotron waves are Bernstein modes which normally are damped and represent merely resonances. However, again, when other kinds of distribution functions are present, the resonance can be inverted and waves can be excited instead of being damped. This is, in particular, the case when electron beams pass the plasma and has been suggested as a possible generation mechanism based on particle observations and close association of electron cyclotron waves with solitary waves [Menietti et al., 28]. Other possible sources of instability as for instance loss-cones, temperature anisotropies,

264

A. Vaivads et al.

and horse-shoe distributions exist as well. The diﬀerent excitation mechanisms have been discussed in LaBelle and Treumann [27]. In case of electron beams, it is crucial that in addition to the beam a cold electron population is present in order for the waves to become unstable. Unfortunately, however, direct measurements of such a component in relation to reconnection are in most cases missing [Menietti et al., 28]. 10.6.3 Relation to Reconnection It has been suggested that electron beams generating electron cyclotron waves originate in the reconnection region. The presence of these waves in association with ﬂux transfer events as well as mainly on open ﬁeld lines indicates that the reconnection process is important in creating unstable electron distribution functions. However, observations of electron cyclotron waves close to the reconnection site are so far missing. It has been suggested that the steep magnetic ﬁeld gradients encountered near the reconnection site should preclude the generation of electron cyclotron waves or inhibit them from developing signiﬁcant amplitudes.

10.7 Free Space Radiation 10.7.1 Observations Electromagnetic radio emissions above the electron plasma frequency can propagate freely throughout the plasma and thus be detected at large distance from the source. This makes the observation of radiation a perfect tool for remote diagnostics of reconnection whenever reconnection sites emit such radiation. Observation of radiation emitted from reconnection is thus of high importance in particular in the astrophysical application, e.g. in the interpretation of the radiation emitted from the solar corona [see, eg., the compilation given in Aschwanden, 5]. In fact, most of the solar radio emission at meter wavelengths is believed to be generated in some coherent emission process that is somehow related to ongoing reconnection in the solar corona [Bastian et al., 7]. These emissions can be classiﬁed into diﬀerent classes of which the most important for our purposes are those which are emitted close to the local plasma frequency fpe and become free space modes as they propagate away from the generation region. Coherent radio emissions from other astronomical objects, such as stellar ﬂares and brown dwarfs, are believed to be generated in similar ways [G¨ udel, 15]. The impossibility of performing in situ measurements in the coronal reconnection regions raises the importance of observations at the accessible magnetospheric reconnection sites. Unfortunately, so far there are no in situ studies of electromagnetic wave generation close to reconnection sites in the magnetosphere. Until now, most attention has been paid to the electromagnetic emission generation near the bow-shock where these waves

10 Plasma Waves Near Reconnection Sites

265

are strongest. However, since reconnection sites are sources of fast particles which are injected into the environment one expects that they are also sources of radio wave emission. 10.7.2 Generation Mechanisms The generation mechanisms of possible free space modes in reconnection are not entirely clear. Several possibilities have been suggested. For example, transverse free space modes (T ) can be generated by mode conversion of Langmuir (L-waves) at steep density gradients, by mode coupling of Langmuir waves with ion sound (S) or other Langmuir (L ) waves according to the relations L + L → T, L + S → T or through direct electron gyro-resonance emission. Of these mechanisms, the latter is the least probable in reconnection as it depends on two conditions. First, the plasma has to be relativistic or at least weakly relativistic. Second, and even more crucial, the plasma has to be underdense with fpe < fce . Close to the reconnection site the involved plasma is, however, overdense as stated in the introduction. In this case cyclotron damping of the free space modes inhibits their excitation. Thus, it seems improbable that reconnection sites would radiate by the gyro-resonance mechanism. In situ measurements close to the reconnection sites are required to distinguish between the remaining possible generation mechanisms. Finally, the large number of energetic electrons generated in reconnection might be another source of nonthermal synchrotron radiation under conditions when the electron energies reached are high. 10.7.3 Relation to Reconnection The localization of electromagnetic radiation generation with respect to the reconnection site has not been investigated. Numerical simulations have studied how electron beams generated in the reconnection process can generate Langmuir waves which, in their turn, can mode convert to the emission of electromagnetic radiation [see, e.g. Sakai et al., 32]. This is the most probable radiation mechanism since electron beams are involved. In addition this mechanism is non-thermal. Depending on the available number of electron beam electrons the emission coming from one single reconnection site might still be below the detection threshold in the magnetosphere or at the magnetopause. For astrophysical applications like the sun and stars the ejected electron beams in type III radiation generating plasma waves are dense and intense enough to provide observable intensities. In remote astrophysical objects, however, synchrotron emission is more important as a nonthermal emission mechanism [Jaroschek et al., 18, 19]. Though it is very weak, the large numbers of particles injected from the reconnection site into a large volume and distributed there increase the emission measure in proportion to the involved volume, making such radiation a good candidate for observation even though it will not provide information about the microscopic scale of the involved astrophysical reconnection sites.

266

A. Vaivads et al.

10.8 Summary and Outlook We have summarized the in situ observations of high frequency waves, at frequencies near the lower hybrid frequency and up to the plasma frequency, generated near the reconnection sites in the Earth magnetosphere. There are many observational studies dealing with the most intense waves, such as lower hybrid drift waves, solitary waves, and Langmuir waves. Some of the observations suggest that these waves are most intense along the separatrices emanating from the reconnection sites. However detailed studies of the wave locations are still missing. Electron beams generated in the reconnection process seem to be a major free energy source that can generate diﬀerent waves, but also density gradients or diﬀerent kinds of distribution functions (e.g., loss-cone or horse-shoe) are important. Electromagnetic waves such as whistlers and radio emissions are important for remote diagnostic possibilities of the reconnection sites. In the case of radio emissions there are direct astrophysical applications. However, in situ studies of both these modes in association with reconnection are very limited. We identify several topics of high importance for further study in the near future: • Wave location. How are diﬀerent wave modes located with respect to the inner structure of the current sheet and the separatrices, and how does this location depend on the reconnection parameters, such as density gradients, velocity shear, plasma beta, temporal evolution? Could some of the wave modes be used to determine the distance to the reconnection X-line? Possible candidates are the intense solitary waves and Langmuir/upper hybrid waves. • Wave-particle interaction. Which wave modes are most important for particle acceleration, heating, and formation of energetic tails on the electron distribution functions? Are electron beams generated in the wave particle interaction or are they generated in prompt electron acceleration in reconnection-generated electric ﬁelds? • Anomalous resistivity. There exist ﬁrst estimates of the anomalous resistivity and anomalous diﬀusion for lower hybrid drift turbulence in connection with reconnection. These results should be conﬁrmed for diﬀerent reconnection conditions. Moreover, the electromagnetic part of the anomalous resistivity needs to be studied more closely both theoretically and experimentally. • Radio emissions. Where and by which mechanism is free space radiation generated near the reconnection sites? How could it be used to remotely sense the reconnection site properties such as the stationarity of reconnection, extension of the reconnection line, etc.?

10 Plasma Waves Near Reconnection Sites

267

References [1] Akimoto, K., S.P. Gary, and N. Omidi: Electron/ion whistler instabilities and magnetic noise bursts, J. Geophys. Res. 92, 11209, 1987. [2] Anderson, R.R., T.E. Eastman, C.C. Harvey, M.M. Hoppe, B.T. Tsurutani, and J. Etcheto: Plasma waves near the magnetopause, J. Geophys. Res. 87, 2087, 1982. [3] Andr´e, M., R. Behlke, J.-E. Wahlund, A. Vaivads, A.-I. Eriksson, A. Tjulin, T.D. Carozzi, C. Cully, G. Gustafsson, D. Sundkvist, Y. Khotyaintsev, N. Cornilleau-Wehrlin, L. Rezeau, M. Maksimovic, E. Lucek, A. Balogh, M. Dunlop, P.-A. Lindqvist, F. Mozer, A. Pedersen, and A. Fazakerley: Multispacecraft observations of broadband waves near the lower hybrid frequency at the earthward edge of the magnetopause, Ann. Geophysicæ 19, 1471, 2001. [4] Andr´e, M., A. Vaivads, S.C. Buchert, A.N. Fazakerley, and A. Lahiﬀ: Thin electron-scale layers at the magnetopause, Geophys. Res. Lett. 31, 3803, 2004. [5] Aschwanden, M.J.: Physics of the Solar Corona: An Introduction, Springer, 2004. [6] Bale, S.D., F.S. Mozer, and T. Phan: Observation of lower hybrid drift instability in the diﬀusion region at a reconnecting magnetopause, Geophys. Res. Lett. 29, 33, 2002. [7] Bastian, T.S., A.O. Benz, and D.E. Gary: Radio Emission from Solar Flares, Ann. Rev. Astron. Astrophys. 36, 131, 1998. [8] Birn, J., J.F. Drake, M.A. Shay, B.N. Rogers, R.E. Denton, M. Hesse, M. Kuznetsova, Z.W. Ma, A. Bhattacharjee, A. Otto, and P.L. Pritchett: Geospace Environmental Modeling (GEM) magnetic reconnection challenge, J. Geophys. Res. 106, 3715, 2001. [9] Cane, H.V., W.C. Erickson, and N.P. Prestage: Solar ﬂares, type III radio bursts, coronal mass ejections, and energetic particles, J. Geophys. Res. 107, 14, 2002. [10] Cattell, C., J. Dombeck, J. Wygant, J.F. Drake, M. Swisdak, M.L. Goldstein, W. Keith, A. Fazakerley, M. Andr´e, E. Lucek, and A. Balogh: Cluster observations of electron holes in association with magnetotail reconnection and comparison to simulations, J. Geophys. Res. 110, 1211, 2005. [11] Daughton, W., G. Lapenta, and P. Ricci: Nonlinear Evolution of the LowerHybrid Drift Instability in a Current Sheet, Phys. Rev. Lett. 93, 105004, 2004. [12] Deng, X.H., H. Matsumoto, H. Kojima, T. Mukai, R.R. Anderson, W. Baumjohann, and R. Nakamura: Geotail encounter with reconnection diﬀusion region in the Earth’s magnetotail: Evidence of multiple X lines collisionless reconnection?, J. Geophys. Res. 109, 5206, 2004. [13] Drake, J.F., M. Swisdak, C. Cattell, M.A. Shay, B.N. Rogers, and A. Zeiler: Formation of electron holes and particle energization during magnetic r, Science 299, 873, 2003. [14] Farrell, W.M., M.D. Desch, M.L. Kaiser, and K. Goetz: The dominance of electron plasma waves near a reconnection X-line region, Geophys. Res. Lett. 29, 8, 2002. [15] G¨ udel, M.: Stellar Radio Astronomy: Probing Stellar Atmospheres from Protostars to Giants, Ann. Rev. Astron. Astrophys. 40, 217, 2002. [16] Gurnett, D.A., L.A. Frank, and R.P. Lepping: Plasma waves at the distant magnetotail, J. Geophys. Res. 81, 6059, 1976.

268

A. Vaivads et al.

[17] Hoshino, M., T. Mukai, T. Terasawa, and I. Shinohara: Suprathermal electron acceleration in magnetic reconnection, J. Geophys. Res. 106, 25979, 2001. [18] Jaroschek, C.H., R.A. Treumann, H. Lesch, and M. Scholer: Fast reconnection in relativistic pair plasmas: Analysis of particle acceleration in self-consistent full-particle simulations, Phys. Plasmas 11, 1151, 2004a. [19] Jaroschek, C.H., H. Lesch, and R.A. Treumann, Relativistic kinetic reconnection as the possible source mechanism for high variability and ﬂat spectra in extragalactic radio sources, Astrophys. J. 605, L9, 2004b. [20] Ji, H., S. Terry, M. Yamada, R. Kulsrud, A. Kuritsyn, and Y. Ren: Electromagnetic Fluctuations during Fast Reconnection in a Laboratory Plasma, Phys. Rev. Lett. 92, 115001, 2004. [21] Kellogg, P.J. and S.D. Bale: Nearly monochromatic waves in the distant tail of the Earth, J. Geophys. Res. 109, 4223, 2004. [22] Khotyaintsev, Y., A. Vaivads, Y. Ogawa, B. Popielawska, M. Andr´e, S. Buchert, P. D´ecr´eau, B. Lavraud, and H.R`eme: Cluster observations of high-frequency waves in the exterior cusp, Ann. Geophysicæ 22, 2403, 2004. [23] Kojima, H., H. Furuya, H. Usui, and H. Matsumoto: Modulated electron plasma waves observed in the tail lobe: Geotail waveform observations, Geophys. Res. Lett. 24, 3049, 1997. [24] Kojima, H., K. Ohtsuka, H. Matsumoto, Y. Omura, R.R. Anderson, Y. Saito, T. Mukai, S. Kokubun, and T. Yamamoto: Plasma waves in slow-mode shocks observed by Geotail Spacecraft, Adv. Space Res. 24, 51, 1999. [25] LaBelle, J., R.A. Treumann, G. Haerendel, O.H. Bauer, and G. Paschmann: AMPTE IKRM obsevations of waves assosicated with ﬂux transfer events in the magnetosphere J. Geophys. Res., 92, 5827, 1987. [26] LaBelle, J. and R.A. Treumann: Plasma waves at the dayside magnetopause, Space Sci. Rev., 47, 175, 1988. [27] LaBelle, J. and R.A. Treumann: Auroral radio emissions, 1. Hisses, roars, and bursts, Space Sci. Rev. 101, 295, 2002. [28] Menietti, J.D., J.S. Pickett, G.B. Hospodarsky, J.D. Scudder, and D.A. Gurnett: Polar observations of plasma waves in and near the dayside magnetopause/magnetosheath, Planet. Space Sci. 52, 1321, 2004. [29] Omura, Y., H. Matsumoto, T. Miyake, and H. Kojima: Electron beam instabilities as generation mechanism of electrostatic solitary waves in the magnetotail, J. Geophys. Res. 101, 2685, 1996. [30] Pritchett, P.L.: Collisionless magnetic reconnection in a three-dimensional open system, J. Geophys. Res., 106, 25961, 2001. [31] Pritchett, P.L. and F.V. Coroniti: Three-dimensional collisionless magnetic reconnection in the presence of a guide ﬁeld, J. Geophys. Res. 109, 1220, 2004. [32] Sakai, J.I., T. Kitamoto, and S. Saito: Simulation of Solar Type III Radio Bursts from a Magnetic Reconnection Region, Astrophys. J. Lett. 622, L157, 2005. [33] Scholer, M., I. Sidorenko, C.H. Jaroschek, R.A. Treumann, and A. Zeiler: Onset of collisionless magnetic reconnection in thin current sheets: Three-dimensional particle simulations, Phys. Plasmas 10, 3521, 2003. [34] Shay, M.A., J.F. Drake, B.N. Rogers, and R.E. Denton: Alfv´enic collisionless magnetic reconnection and the Hall term, J. Geophys. Res. 106, 3759, 2001. [35] Silin, I., J. B¨ uchner, and A. Vaivads: Anomalous resistivity due to nonlinear lower-hybrid drift waves, Phys. Plasmas 12, submitted, 2005.

10 Plasma Waves Near Reconnection Sites

269

[36] Stenberg, G., T. Oscarsson, M. Andr´e, M. Backrud, Y. Khotyaintsev, A. Vaivads, F. Sahraoui, N. Cornilleau-Wehrlin, A. Fazakerley, R. Lundin, and P. D´ecr´eau: Electron-scale structures indicating patchy reconnection at the magnetopause? J. Geophys. Res. 110, submitted, 2005. [37] Vaivads, A., M. Andr´e, S.C. Buchert, J.-E. Wahlund, A.N. Fazakerley, and N. Cornilleau-Wehrlin: Cluster observations of lower hybrid turbulence within thin layers at the magnetopause, Geophys. Res. Lett. 31, 3804, 2004. [38] Vaivads, A., Y. Khotyaintsev, M. Andr´e, A. Retin` o, S.C. Buchert, B.N. Rogers, P. D´ecr´eau, G. Paschmann, and T.D. Phan: Structure of the Magnetic Reconnection Diﬀusion Region from Four-Spacecraft Observations, Phys. Rev. Lett. 93, 105001, 2004. [39] Zhang, Y., H. Matsumoto, and H. Kojima: Whistler mode waves in the magnetotail, J. Geophys. Res. 104, 28633, 1999.

Part III

High-Frequency Analysis Techniques and Wave Instrumentation

11 Tests of Time Evolutions in Deterministic Models, by Random Sampling of Space Plasma Phenomena H.L. P´ecseli1,3 and J. Trulsen2,3 1

2

3

University of Oslo, Institute of Physics, Box 1048 Blindern, 0316 Oslo, Norway [email protected] University of Oslo, Institute of Theoretical Astrophysics, Box 1029 Blindern, 0315 Oslo, Norway [email protected] Centre for Advanced Study, Drammensveien 78, 0271 Oslo, Norway

Abstract. We discuss general ideas, which can be used for estimating models for coherent time-evolutions by random sampling of data. They turn out to be particularly useful for interpreting data from instrumented spacecraft. These “new methods” are applied to examples of localized bursts of lower-hybrid waves and correlated density depletions observed on the FREJA satellite. In particular, lower-hybrid wave collapse is investigated. The statistical arguments are based on three distinct elements. Two are purely geometric, where the chord length distribution is determined for given cavity scales, together with the probability of encountering those scales. The third part of the argument is based on the actual time variation of scales predicted by the collapse model. The cavities are assumed to be uniformly distributed along the spacecraft trajectory, and it is assumed that they are encountered with equal probability at any time during the dynamical evolution. Cylindrical and ellipsoidal cavity models are discussed. It turns out that the collapsing cavity model can safely be ruled out on the basis of disagreement of data with the predicted cavity lengths and evolution time scales. Application to Langmuir wave collapse is suggested in order to check its reality and relevance.

Key words: Wave packet dynamics, lower-hybrid wave collapse, Freja observations, nonlinear dynamics, random data sampling

11.1 Introduction When studying data from rocket or satellite observations, one might often encounter situations where the results invite an interpretation in terms of analytical models for the space-time evolution of some physical phenomena, nonlinear plasma waves for instance. Very often the spacecraft velocity is H.L. P´ ecseli and J. Trulsen: Tests of Time Evolutions in Deterministic Models, by Random Sampling of Space Plasma Phenomena, Lect. Notes Phys. 687, 273–297 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

274

H.L. P´ecseli and J. Trulsen

so large that the observations have to be interpreted as “snap-shots” of the phenomena, and a direct comparison between analytical results for the time evolution and the observations is not feasible. The problem seems almost a dead-lock, by the implication that time evolutions can under no circumstances be observed under such conditions. This situation easily stimulates speculations leading to unsubstantiated claims, in the sense that the interpretation of data has been extended beyond what can be justiﬁed by the observations alone. In the present communication we will describe a general method which allows tests of models for deterministic time evolutions, provided the database covers suﬃciently many observations. The basic idea is that one can predict the probability density of observable quantities analytically, based on the deterministic model together with some plausible assumptions, and that these distributions can then be tested experimentally [1, 2]. The results cannot prove a theory to be correct (logically, this might not be possible anyhow), but can certainly be useful in disproving certain models. We give an outline of the statistical arguments, and illustrate the ideas by an analysis of data for lower-hybrid waves as observed by the Freja satellite. For this case an interpretation in terms of lower-hybrid wave collapse might be tempting and has indeed been suggested. Our analysis indicates that this interpretation is in error. Later on we will present more detailed examples, but as an introduction it might here suﬃce to consider a simple (over-)idealized case, where we assume that spherical voids are forming in a plasma. Such a void is assumed to be embedded in the plasma with density n0 , and have a density depletion of ∆n. Any void forms spontaneously, and its radius has a deterministic time variation as, say,

t − t0 t − t0 R(t) = r0 1− , (11.1) T T for t0 < t < t0 + T and R = 0 otherwise. There is a well deﬁned initial time, t0 , for the formation of individual voids, but for the ensemble of voids, these times are random, and mutually independent. The time evolution of the voids is thus completely deterministic, but the formation time (and consequently the collapse times as well) are statistically distributed. When such an ensemble of voids is sampled by a spacecraft, the sampling process is itself associated with statistically distributed quantities. At a certain time, t∗ , the width of the void is 2R(t∗ ), but if it is determined by analyzing the data obtained by a satellite moving along a straight line orbit, we ﬁnd a chord-length , which is in general shorter. We can argue that a chord length distribution can be determined analytically, with the assumption that we encounter a cavity at any position in its cross section with equal probability: any other assumption will imply that the formation of the void is correlated with the presence of the satellite! It is then a simple matter to obtain the probability density for the observed chord-lengths, .

11 Tests of Time Evolutions in Deterministic Models

275

We ﬁrst consider the impact parameter b as measured from the center of the void. We note that with the given assumptions, the probability for having an impact parameter in the interval {b, b + db}, is obtained as the ratio of the two areas 2πbdb and πR2 (t∗ ), assuming that the void does not change appreciably during the passage of the detector. We have P (b)db being the probability of ﬁnding ' an impact parameter in the interval {b, b + db}, and also the relation b = R2 (t∗ ) − 2 /4. Using P (b)db = P ()d we ﬁnd P (|R) =

, 2R2 (t∗ )

(11.2)

for 0 < ≤ 2R(t∗ ) and P (|R) = 0 otherwise. Since the maximum value of R is Rm = r0 /4, we have the maximum value of being m = r0 /2. Implicit in the argument is, as said, the assumption that the void develops slowly, i.e. it does not change appreciably during the transit time of the satellite. The result (11.2) is conditional, as indicated, in the sense that R is assumed given at the relevant time. The information concerning the density in the void is trivial in the present model, since we assumed ∆n = const. In addition to , also the time of interception, t∗ , is statistically distributed with respect to the formation time t0 . Again, we can safely argue that the satellite intercepts the time-evolving void at a time randomly distributed in the interval {0, T }, i.e. P (t∗ )dt∗ = dt∗ /T for 0 < t∗ < T . For the simple example in (11.1), the time variation is monotonic and symmetric in the two intervals {0, T /2} and {T /2, T }, so we need only be concerned with one of them, say the ﬁrst one. The variation of R is restricted to the interval {0, r0 /4}. We want to determine the statistical distribution of radii in the voids, R, which is a consequence of the random distribution of detection times in the relation (11.1). We have P (R)dR = dt∗ /T . With (11.1) we ﬁnd dR/dt∗ = r0 /T − 2r0 t∗ /T 2 , giving P (R) =

2 1 ' . r0 1 − 4R/r0

(11.3)

For R > r0 /4, we have P (R) = 0. Each realization of the plasma has “blobs” randomly distributed in all different stages of their time-evolution, and we encounter them with a probability depending on their cross section, σ = πR2 (t∗ ) , which gives P () =

1 C

(11.4)

r0 /4

σ(R)P (|R)P (R)dR ,

/2

(11.5)

276

H.L. P´ecseli and J. Trulsen

where C is a normalization constant, introduced because the probability density obtained by introducing σ is not automatically normalized. The expression (11.5) is readily solved for this case to give ( 15 1−2 , (11.6) P () = r0 r0 r0 and P () = 0 for > r0 /2. The result (11.6) is the probability density which can be tested experimentally. An estimate for the probability density (11.5) could be obtained experimentally, and we might then hope to ﬁnd support (if not proof) for the proposed model, including the time variation it implies. The model described here is oversimpliﬁed in many respects, in particular also by assuming all structures to develop identically. As a minimum requirement to make the model at least slightly convincing, we have to relax the assumption of identical r0 ’s, and allow the reference magnitude, r0 , of the “blobs” to be randomly distributed, in general over the interval {0; ∞}. We can assume that this distribution is also random, and assign a probability density P (r0 ). The result for the probability density (11.5) or (11.6) of is then conditional, P (|r0 ) and we ﬁnd ∞ P (|r0 )P (r0 )dr0 , (11.7) P () = 2

for a given P (r0 ). Unfortunately we may not know P (r0 ) a priori. The problem can, however, be “inverted”, and we might see whether it is possible to propose a physically realistic P (r0 ), which gives agreement with the observed probability densities. If not, we have a contradiction, and have to reject the model, and may be a little wiser, at least in this respect. If reasonable agreement is found, we might proceed with the model, by making it more detailed and explore its limitations. Finally, before entering into discussions of more realistic cases and their associated models, we might draw attention to a “practical” problem. Even if the phenomena we consider are persistent, we might be in need of data. Thus, we would like to make a data-basis consisting of statistically independent observations, and might be tempted to retain only one observation from each satellite pass, for instance. This seems a safe and sound idea, but it leaves many observations redundant, and the estimate of the probability densities may become uncertain. In addition, the plasma parameters are likely to change from one pass to the next, and having to model statistically distributed plasma as well may become troublesome. It is preferable to use as many data as possible from consistent plasma parameters. If the observations are abundant in some orbits (fortunately, they are, sometimes), we might like to know to what extent, if at all, these observations form a set of independent data? This question is not easily answered, but we can test at least one hypothesis against the observations. By randomly distributed events in space, we mean that the position of one is statistically independent of the position

11 Tests of Time Evolutions in Deterministic Models

277

of all the others. The probability of ﬁnding a structure in a short interval dx measured along the satellite orbit is taken to be µdx, with µ being the density of structures along this line (the dimension of µ is length−1 ). This assumption, we know, leads to a Poisson distribution of the number, N , of structures in an interval L, i.e. P (N, L) = (µL)N exp(−µL)/N !. This distribution can be tested, usually by simply checking ( N 2 − N 2 )/ N = 1, as valid for the Poisson distribution. More extensive tests can also be carried out [3]. In the following Sect. 11.2 we extend the model discussions, by considering the consequences of having more general density variations than the simple “inverted top hat” density depletion model used for illustrations in this Introduction. Then, in Sect. 11.3, we consider one particular problem, namely collapse of lower hybrid waves, and apply some of the ideas developed here.

11.2 Model Discussions Assume that we have an analytical model for the time evolution of a physical phenomenon, predicting a quantity Ψ (r, t), where Ψ might denote the space-time varying plasma density, or the potential or something else. As an example, we might have Ψ representing the space time evolution of a sound wave pulse, phase space vortex, a shock, or similar. Often, we can assume that the phenomenon is associated with one or more characteristic lengths. We can take the width at half-amplitude, for instance. Let such a length scale be Lp (t), which is in general a function of time. We predict Lp (t), but measure Lp (t = t0 ), where t0 is the time we encounter the structure. We do not know what time this is in relation to the beginning of the time evolution of the phenomenon, and this is basically one of the roots of our problem. Another is that we predict a characteristic length as the width of a wave pulse, but the trajectory of the spacecraft need not cross the structure at its maximum diameter. If we later cross a similar object in precisely the same stage of its development we will measure a diﬀerent width, simply because we are likely to cross the object along a diﬀerent trajectory in diﬀerent trials. The problem thus has two basic statistically distributed quantities, one associated with the temporal, and another with the spatial variable. There is a further statistical distribution associated with the fact that diﬀerent events are not identical, but originate from statistically distributed initial conditions. 11.2.1 Spatial Sampling with One Probe Available Let us ﬁrst discuss the distribution in a characteristic quantity as a length scale, due to the random distribution of sampling trajectories. In general, we may not have any a priori knowledge about the actual shape of the structures, i.e. n = n(x, y) in this case. A model can be proposed, however, and subsequently be tested against the data. This will be feasible in particular when the detecting spacecraft is equipped with two or more spatially separated probes.

278

H.L. P´ecseli and J. Trulsen

As a suﬃciently general generic form, useful as an illustration, we consider a density pulse in a simple two dimensional model n(x, y) ≡ n0 − n & = n0 − ∆n e− 2 (x 1

2

+y 2 )m /R2m

.

(11.8)

This model contains an inverted top-hat model as the limiting case m → ∞, and a rotationally symmetric Gaussian model corresponding to m = 1. We might let ∆n have either sign, corresponding to density “humps” or depletions. We can use the same exponent on both x and y if a symmetry condition is satisﬁed, i.e. in the absence of a preferred direction. The chord length, , corresponding to the 1/e-width of the density depletion, as obtained along the satellite orbit, is easily obtained from (11.8). For m > 1 an increase in the impact parameter implies a decrease in measured chord length. For m < 1 the opposite impact parameter variation is obtained; a somewhat counter-intuitive result. The depth of cavities as detected along the sector determined by the satellite trajectory can also be obtained from (11.8), giving an expression varying with impact parameter y. A straight forward parametric representation of the normalized depth – chord length relation can be obtained as 0 1 2m (11.9) {& n/∆n, /R} = e− 2 ξ , 2 (2 + ξ 2m )1/m − ξ 2 , with ξ ≡ y/R being the normalized impact ' parameter. In particular, for m → 2m ∞, we have ξ → 0, so that /R = 2 1 − ξ 2 . The expression (11.9) gives the peak cavity depth and chord length for a given normalized impact parameter y for cavities deﬁned by (11.8), assuming ∆n, R and m to be known. When examining an actual record, the impact parameter is in general varying from cavity to cavity, with y being statistically distributed. The only acceptable assumption concerning its probability density is a uniform distribution, P (|y|) = 1/LM for 0 < |y| < LM and zero otherwise. Here LM is an assumed maximum impact parameter from the satellite to a cavity. Eventually we let LM → ∞. With this assumption we can in principle obtain the probability densities P () and P (& n). For instance, using P (& n)d& n= P (|y|)d|y| we readily ﬁnd P

n & ∆n

=

−1+1/(2m) n & R ∆n 2 ln , mLM n & ∆n

(11.10)

&/∆n ≤ 1. The probability density (11.10) is normalized for e− 2 (LM /R) < n for all ﬁnite LM , but the expression may for some values of m give trivial results in the limit of LM → ∞. The corresponding results for P () are not so easily obtained in a closed form, but can be determined for selected values of m. 1

2m

11 Tests of Time Evolutions in Deterministic Models

279

1) For m = 1/2, we ﬁnd, using P ()d = P (y)dy P () =

, 8RLM

(11.11)

' for 4 < /R < 4 1 + LM /R, see the second half of the expression (11.9). 2) For m = 1 we have √ (11.12) P () = δ − 2R 2 . 3) For m = 2 we have (y/R)2 = [32 − (/R)4 ]/[8(/R)2 ] giving P () =

1 √ LM 2 2

R

2

32 + (/R)4 ' , 32 − (/R)4

(11.13)

' with 2 (2 + (LM /R)4 )1/2 − (LM /R)2 < /R < (32)1/4 . Models with other values of m can be analyzed similarly. It is evident that the results depend sensitively on the model, here the exponent m, and a statistical analysis for determining this and other parameter values is worthwhile. All these foregoing results are conditional in the sense that they assume all cavities to be identical, i.e. given ∆n and R all cavities are of the same constant form (11.8). Given a priori knowledge, or at least a qualiﬁed guess, of the probability density of the characteristic parameters, we can then use Bayes’ theorem to obtain unconditioned distributions. 4) For the case where m → ∞ the analysis becomes more lengthy. We consider two cases, a cylindrical and an ellipsoidal model. Cylindrical Model As an illustration assume ﬁrst that the cavity has a cylindrical shape with a circular cross section of radius L⊥ (t), while L is for the moment considered inﬁnite. The observations are then essentially restricted to a plane. As a simple case, we here assume that we have an “inverted top-hat” density depletion, i.e. n = n0 outside the cavities, and n = n0 − ∆n inside. The geometrical cross-section for the cavity is thus σ = 2 L⊥ with dimension “length” in this planar approximation. The probability for actually encountering a cavity in an interval of length dx along the spacecraft trajectory is proportional to µ(L⊥ ) 2L⊥ dL⊥ dx, where, again, µ(L⊥ )dL⊥ is the + ∞density of cavities with perpendicular radii within L⊥ , L⊥ +dL⊥ . We have 0 µ(L⊥ )dL⊥ = µc , where µc is the cavity density irrespective of diameter. For the present model, the dimensions of µc are length−2 , i.e. the dimensions of µ(L⊥ ) are length−3 . Assuming the cavities to appear and disappear at random, the dynamics can be assumed to be time-stationary in a statistical sense, with the density of cavities µ(L⊥ ) being constant, even if L⊥ for the individual cavity varies with time. In a given realization there are many diﬀerent scale sizes present

280

H.L. P´ecseli and J. Trulsen

at the same time. The probability for encountering one particular value for L⊥ depends on the relative density of cavities with that particular diameter. Assuming that the spacecraft has encountered a cavity it is evident that the probability of its radius being L⊥ is proportional to the density of cavities with that particular radius as well as to the corresponding geometrical crosssection. We can, more generally, derive the probability density for the radii of observed cavities P(L⊥ ), as illustrated later on in (11.25). Let the angle between the spacecraft trajectory and the magnetic ﬁeld be θ. A straight-line trajectory intercepts cavities of radius L⊥ along chords at a constant y-value in the ellipse x2 y2 = 1, + L2⊥ / sin2 θ L2⊥ all |y|-values in the interval {0, L⊥ } being equally probable. With the foregoing assumption, the distribution of chord lengths, , is readily obtained as P (|L⊥ ) =

sin2 θ 1 0 2L⊥ 4L2 − 2 sin2 θ

(11.14)

⊥

for 0 < < 2L⊥ / sin θ for a given ﬁxed L⊥ . The angle θ is considered as a constant. It was also here assumed that the satellite speed is so large that the cavity does not change appreciably during its passage. Evidently (11.14) predicts that there is a large probability of ﬁnding ∼ 2L⊥ / sin θ. For distributed cavity radius L⊥ we have ∞ P () = P (|L⊥ )P(L⊥ )dL⊥ (11.15) 0

= sin2 θ

L⊥max

sin θ/2

1 P(L⊥ ) 0 dL⊥ 2L⊥ 4L2 − 2 sin2 θ ⊥

with the actual probability density P(L⊥ ) for the cavity radii to be inserted. Ellipsoidal Model The foregoing result for P (|L⊥ ), the probability density of observed chord lengths, was derived for a cylindrical form of the cavity. For a spherical cavity we have the simple expression (11.2) already derived in the introduction. More generally a rotationally symmetric ellipsoidal model (i.e. a “cigar”), with half axes L⊥ perpendicular to B and L parallel to B, can be used for the density cavities. We can generalize (11.8) to n = n0 − ∆n e− 2 [(x 1

2

+y 2 +(z/L )2 L2⊥ )m /L2m ⊥ ]

,

(11.16)

Again, the case m = 1 is relatively simple, so we consider the simple “inverted-top-hat” model, with m → ∞.

11 Tests of Time Evolutions in Deterministic Models

281

Because of the rotational symmetry of the problem with respect to the magnetic ﬁeld, (11.16) is presumably an adequate model. The centers of the ellipsoids are assumed to be randomly and uniformly distributed in three dimensions. The cross-section of a given cavity for a spacecraft moving at an angle θ to the major axis is 0 σ(L , L⊥ ) = πL⊥ L2⊥ cos2 θ + L2 sin2 θ. For θ = 0, i.e. for satellite propagations along the major axis of the ellipsoid, which will generally be the magnetic ﬁeld direction, we ﬁnd σ = πL2⊥ , as expected. The probability density P(L⊥ , L ) of encountering a certain cavity speciﬁed by (L , L⊥ ), is proportional to σ(L , L⊥ ) and to the appropriate density of cavities, µ(L , L⊥ ), as discussed before. We have µ(L , L⊥ )dL dL⊥ = +∞+∞ µc P (L⊥ , L )dL dL⊥ , with 0 0 µ(L , L⊥ )dL dL⊥ = µc being the density of cavity centers, irrespective of cavity widths, as before. The probability density P(L⊥ , L ) can not oﬀ-hand be equated to P(L⊥ )P(L ) since the time evolutions of L and L⊥ are in general not statistically independent. As an illustrative model, which will be useful later on with (11.24), we assume L = βL2⊥ , where the constant β is determined by initial conditions. Assuming the probability density P (L⊥ ) to be known (an example will be analytically determined in (11.25), for a speciﬁc model) as well as the cross section obtained previously, we obtain P(L⊥ , L ) = δ(L − βL2⊥ ) P (L⊥ )σ(L , L⊥ ) −1 L⊥max Lmax 2 × δ(L − βL⊥ )P (L⊥ )σ(L , L⊥ ) , 0

0

where Lmax = βL2⊥max . The normalizing factor in the angular parentheses can in many cases be calculated analytically [1], but might become a rather lengthy expression. We now consider the conditional probability density for observed chord lengths for a given cavity speciﬁed by (L , L⊥ ). After some calculations we obtain L2 P ( | L⊥ , L ) = 2 2 , (11.17) 2L L⊥ for 0 < < 2L L⊥ /L with L2 = L2⊥ cos2 θ + L2 sin2 θ. For propagation across the major axis of the ellipsoid, i.e. θ = 90◦ , we have P ( | L⊥ , L ) = /(2L2⊥ ) as a particular case of (11.17), independent of L . This result is identical to the one obtained for a sphere with radius L⊥ , and can be understood by simple geometrical arguments. For distributed values of L⊥ and L we ﬁnd P () as in (11.14) 2 cos θ sin2 θ + (11.18) P(L⊥ , L )dL⊥ dL , P () = 2 L2 L2⊥

282

H.L. P´ecseli and J. Trulsen

with the integration restricted to that part of the L⊥ ,L -plane where 2L2 L2⊥ > 2 (L2⊥ cos2 θ + L2 sin2 θ) and 0 < L⊥ < L⊥max , 0 < L < Lmax . The parenthesis in (11.16) originates from P ( | L⊥ , L ). The maximum values Lmax and L⊥max , are also here assumed to be given. In the actual situation they will diﬀer from case to case, and the corresponding probability density must be included in the analysis at a later stage. 11.2.2 Spatial Sampling with Two Probes Available The models can be tested to a greater accuracy in cases where the spacecraft is equipped with two or more probes for measuring ﬂuctuations in the plasma density. We can then have two cross sections for the same structure, and be able to determine, or at least estimate, parameters in (11.8). Assume that the two probes are separated by a distance ∆s in the direction perpendicular to the velocity vector of the satellite. Referring to the model (11.8) we let again ξ be the normalized impact parameter for probe ' 1, so the normalized chord length detected by this probe is 1 /R = 2 (2 + ξ 2m )1/m − ξ 2 . The normalized chord length detected'by probe 2 crossing the same cavity will be one of the two values 2± /R = 2 (2 + (ξ ± ∆)2m )1/m − (ξ ± ∆)2 with equal probability, since the satellite can impact the structure on either side with equal probability. We introduced the normalized probe separation ∆ = ∆s /R. If it is possible to eliminate ξ from the expressions for 1 and 2± , we can obtain a distribution for the two measurable chord-lengths. If, in particular, we have m = 1, we ﬁnd 1 = 2± irrespective of ξ, emphasizing the special properties of structures with Gaussian shape. For m = 1/2 we have ξ = (1 /4R)2 − 1 to give 0 2 2 2± = 2 (2 + (1 /4R)2 − 1 ± ∆) − ((1 /4R)2 − 1 ± ∆) ' = 4 1 + (1 /4R)2 − 1 ± ∆ . (11.19) For the limiting case m → ∞ we have 1 0

2 2± = 2 1 − 1 − 21 /(4R2 ) ± ∆ ,

(11.20)

although this latter model is complicated by the possibility that one probe crosses the structure, while the other one does not, i.e. ξ < 1 but ξ + ∆ > 1, for instance. If we take an expression like (11.19) or (11.20) to be a hypothesis, it can be directly tested against available data. 11.2.3 Temporal Sampling Independent observations happen at diﬀerent times of the evolution of the structures. If we, for instance, were to analyze whistler wave-packets as they

11 Tests of Time Evolutions in Deterministic Models

283

are often excited in the ionosphere by lightning strokes, we might detect the individual whistler wave-packets at diﬀerent times after the lightning, this time diﬀerence being statistically distributed. Since the arrival of the spacecraft is safely assumed to be independent of the generation mechanism, we can assume the time of observation to be uniformly distributed in a (large) time interval {0, T }, where we might eventually let T → ∞. We have P (t) = 1/T , which gives an immense simpliﬁcation of the analysis, as already illustrated in the Introduction. This assumption is consistent with our other basic assumptions, namely that spatial structures are traversed at positions which are uniformly distributed in the plane perpendicular to the spacecraft orbit.

11.3 Nonlinear Lower-Hybrid Wave Models A number of spacecraft observations demonstrated the existence of small density depletions in the Earth’s ionosphere [4, 5, 6], see Fig. 11.1. The structures were associated with a localized enhanced wave activity, where the waves were identiﬁed as electrostatic lower-hybrid waves [7]. The observations had a close resemblance to what could be expected from an intermediate stage of wave-collapse, and the observations evidently received considerable attention. However, it is evident that a simple visual inspection is not suﬃcient for identifying a collapse phenomenon, and a more detailed analysis had to be performed [1, 3, 8]. As the ﬁrst part of such an analysis it is necessary to ﬁnd a model for the (deterministic) time evolution of observable quantities. A set of nonlinear model equations for the lower-hybrid wave dynamics has been proposed [9, 10, 11]. They are based on a set of nonlinear equations for the wave potential φ and the slowly varying density perturbation n −2i

1 ωLH

2 2 ωpe ∂ 2 M ωLH ∇ φ − Λ2 ∇4 φ + φ ∂t m c2 Ωce

M ∂2 n M ωLH ˆ + φ = −i ∇φ × ∇ ·z m ∂z 2 m Ωce n0

with Λ2 =

2 ωpe Te +2 2 3 2 2 + Ω2 ωLH M Ωce m ωpe ce

Ti

(11.21)

,

2 2 2 2 2 where ωLH = Ωci + ωpi /(1 + ωpe /Ωce ). The time evolution of n is, in one limiting case of the model, given by 2

2 ε0 ωpe ∂ n 2 2 ˆ. − C ∇ = i ∇2 (∇φ∗ × ∇φ) · z (11.22) s ∂t2 n0 4n0 M Ωce ωLH

A driving term in (11.21) is not included as the waves are frequently observed in regions where such a mechanism cannot be identiﬁed. The magnetic ﬁeld is

284

H.L. P´ecseli and J. Trulsen

Fig. 11.1. Example of lower hybrid wave cavity detected by instruments on the FREJA satellite. The lower hybrid wave electric ﬁeld is shown in the upper frame, giving the medium frequency band of the detecting probe circuits. The signal has a bandwidth of 0–16 kHz and a sampling rate of 32 · 103 samples/s. The bottom frame shows the low frequency relative plasma density variations in percent as obtained from two Langmuir probes with an 11.0 m separation. The middle frame shows a wavelet transform of the upper frame [1]. The spin-period of the satellite is approximately 6 s, so its spin phase can be considered constant during a given data sequence

11 Tests of Time Evolutions in Deterministic Models

285

ignored in (11.22) for the slow plasma variation because it is anticipated that the appropriate time scale is shorter than the ion gyro-periods, approximately 2.5 ms for H+ and 35–40 ms for O+ for the present conditions. The assumption has to be veriﬁed a posteriori, but in case the assumption is violated, the equations are readily modiﬁed. For time-stationary conditions this equation is reduced to 2 ε0 ωpe n ˆ. =i (∇φ × ∇φ∗ ) · z (11.23) n0 4n0 Te Ωce ωLH Contrary to the case of Langmuir oscillations, the lower-hybrid waves can be localized within density wells, n < 0, as well as density “humps”, n > 0 [11]. This apparent symmetry is broken by small terms left out in the analysis. The properties of the model equations (11.21)–(11.22) have been studied in great detail [11]. Most important is the observation of collapsing solutions. For suﬃciently large wave intensities the nonlinearities form wave-ﬁlled plasma cavities that collapse into singularities within a ﬁnite time with time variations of their perpendicular and parallel diameters given by L⊥ ∼ (tc − t)1/2 and L ∼ (tc − t) ,

(11.24)

where tc is the collapse time and t0 an arbitrarily chosen initial time for the process, t0 < t < tc . At the same time the electric ﬁeld amplitude at the center of the collapsing cavity increases without bounds. The result in (11.24) is based on a sub-sonic model ' for the collapse. The wave cavity is strongly Bﬁeld aligned with L > L⊥ M/m. There is no natural propagation velocity associated with the cavity. We introduce the initial maximum length scales as L⊥ = L⊥max and L = Lmax at t = t0 . Ultimately, for very large ﬁeld amplitudes and small cavity sizes, the model equations (11.21)–(11.22) become unphysical also because they lack a proper dissipative mechanism. With linear damping being small or negligible, we expect transit time damping to be the dominating mechanism [12, 13]. Here, particles from the surrounding plasma pass through the collapsing cavity and interact with the intensiﬁed electric ﬁelds there. A particle passing through this region in less than one period of the oscillating electric ﬁeld will on average gain energy, giving rise to a corresponding damping of the electric ﬁeld. For parameters Ωe > ωpe discussed here, the collapse will be damped by electrons with small pitch angles passing through the collapse region in the direction parallel to the magnetic ﬁeld. Signiﬁcant damping and eventual arrest of the collapse process are expected when the parallel length of the density depletion becomes comparable to the distance travelled by a thermal electron during one period of the lower-hybrid wave. For the present parameters we ﬁnd a corresponding minimum length scale of Lmin ∼ 500 m. Electrons gain energy (on average) at a single pass when propagating along the magnetic ﬁeld. Fast, light, ions moving with a large Larmor radius from the surrounding plasma into the cavity can also contribute to the damping by being accelerated in that fraction of their gyroperiod they spend inside the cavity. This contribution to

286

H.L. P´ecseli and J. Trulsen

the wave damping is somewhat more complicated to describe analytically, and it is eﬀective only for perpendicular cavity widths smaller than or comparable to the ion Larmor radius, L⊥ ≤ ρL . Due to their gyration, ions interact many times with an intensiﬁed ﬁeld in a long ﬁeld aligned cavity. Since ωLH > Ωci we can assume that the wave phase is randomly varying at each ion encounter with the cavity. This multiple interaction gives a multiplication factor of the order of L /(v τi ), where v is the B-parallel ion velocity and τi = 2π/Ωci . Thus, even for cases where ions (on average) obtain only a small energy increment at a single pass through a cavity, the net result can be signiﬁcant. This mechanism clearly favors transverse ion energization where v is small. In agreement with the analytical results based on (11.21) and (11.22), we would expect lower-hybrid wave ﬁelds to build up by some mechanism, e.g. a linear instability. When a suﬃciently large amplitude has been reached, a modulational instability sets in, breaking the wavetrain up into wave cavities which collapse and ultimately dissipate the wave energy at small scales. In the case where the waves are generated at short wavelengths the model outlined here may have to be modiﬁed by inclusion of parametric decay processes. Two basically diﬀerent scenarios can be envisaged; one where the lowerhybrid waves are continuously maintained, essentially rendering the cavity formation a statistically time stationary process. Alternatively, we can imagine a situation where lower-hybrid waves are excited in a large region of space as a burst, which eventually breaks up into cavitons and ﬁnally dissipates. It is not possible to discriminate between these two scenarios on the basis of the available data. However, large amplitude lower-hybrid waves are frequently observed at medium frequencies, and it is likely that they constitute a background wave component for extended periods of time.

11.4 Probability Densities for Observables In this section we discuss the observations with the a priori assumption that their explanation is to be sought in a collapse of lower-hybrid waves, without reference to the wave generation mechanism. As it is not possible to study the full space-time evolution of individual events in detail, we attempt to derive probability densities for observable quantities within such a model. These can then be compared with those obtained from the data. In the early investigations of Freja observations, when data were sparse, a simple “inverted-top-hat” density depletion was used to model the density cavities [1]. Later, more detailed, studies [8] demonstrated that a more simpler Gaussian depletion model was the most accurate. In particular, we discuss here the probability density for the cavity width determined from the data, as this is one of the quantities which is most easily and accurately obtained. This width coincides with the width of the localized wave packets, so these need not be analyzed separately in this context. The

11 Tests of Time Evolutions in Deterministic Models

287

measurements of the electric ﬁeld intensities are uncertain and give underestimates, and a statistical analysis of the lower-hybrid wave amplitudes is not really meaningful. At early stages a cavity is presumably large and irregular and will hardly be distinguished from the background of low frequency density ﬂuctuations. Assume that after reaching a signiﬁcant amplitude, a cavity can be eﬀectively recognized in a time interval {t0 , tc }, and that the time variation here can be described by (11.24) with tc being the collapse time. Evidently, we can assume that the spacecraft intercepts a cavity at times uniformly distributed in {t0 , tc }, i.e. the probability density for the time of interception is P (t) = 1/(tc − t0 ) for t0 < t < tc . The distribution of length scales is obtained from P (L)dL = P (t)dt. By use of (11.24) the results are P (L⊥ ) =

2 L⊥ L2⊥max

and

P (L ) =

1 , Lmax

(11.25)

for 0 < L⊥ < L⊥max and 0 < L < Lmax , respectively. For a uniform random temporal distribution of collapse times tc , the density of cavities having a transverse scale in the interval {L⊥ , L⊥ + dL⊥ } is given by µ(L⊥ )dL⊥ = µc P (L⊥ )dL⊥ , where µc is the spatial density of collapse centers. It was here implicitly assumed that all cavities start out with essentially the same value for Lmax . Later we relax this condition. As the satellite moves (essentially in the direction perpendicular to B) we expect predominantly large scales to be detected, observations of the actual collapse being statistically improbable. Although the collapse itself is thus unlikely to be detected (its “cross-section” is too small), the entire time evolution preceding it will in principle be reﬂected in the probability distribution of length scales. The cavity spends a comparatively large time in a state with large diameters, where it also has the largest cross-section for observation. For the simple cylindrical Gaussian model, with m = 1 in (11.8), we have L⊥max √ √ (11.26) δ( − 2L⊥ 2)P (L⊥ )dL⊥ = P (|L⊥max ) = 2 2L⊥max 2 0 √ using (11.25). For > 2L⊥max 2, we have P (|L⊥max ) = 0. If we have a statistical distribution of L⊥max for the ensemble of cavities, we evidently have P (L⊥max ) ∞ P () = dL⊥max . σ(L⊥max ) C /2 L2⊥max The integral, and the normalizing quantity C, can readily be solved for realistic choices of P (L⊥max ), using that σ(L⊥max ) ∼ 2 L⊥max for the present cylindrical model. Within an “inverted top hat” cylindrical density depletion model, m → ∞ in (11.8), and the time evolution for L⊥ given by (11.24) and (11.25) as appropriate for a lower hybrid collapse, the probability density of observed chord lengths becomes

H.L. P´ecseli and J. Trulsen 1.5

1.5

a

P2Lmax

P2Lmax sinΘ

288

1 0.5 0

0.5 1 sinΘ2Lmax

b

1 0.5 0

1.5

0.5 1 2Lmax

1.5

Fig. 11.2. Probability densities for chord lengths, , in the case of randomly distributed time-stationary cylindrical cavities in (a), and ellipsoidal ones in (b). Solid lines show cases where all cavities are identical, the dashed line illustrates the eﬀect of distributed diameters, according to a model distribution. The results are obtained by explicitly using the properties of an “inverted top-hat” density depletion, m → ∞ in (11.8)

3 sin2 θ P (|L⊥max ) = 4 L2⊥max

1

1−

sin θ 2L⊥max

2 (11.27)

for sin θ < 2L⊥max and P (|L⊥max ) = 0 otherwise. This result is shown in Fig. 11.2a. In case there are reasons to expect that cavities have a significant distribution in scale sizes, L⊥max , the averaging over the appropriate probability density is easily included in (11.27). Still with an “inverted top hat” density depletion model and the time evolution for L⊥ given by (11.24), we consider a three dimensional case with randomly distributed ellipsoids. For simplicity we consider the case θ ∼ 90◦ , i.e. the satellite propagates essentially perpendicular to B. Inserting the previously obtained P(L⊥ , L ) we obtain

2 1− P () = 2 . (11.28) L⊥max 2L⊥max This result is shown in Fig. 11.2b with a full line. This result, as well as that in Fig. 11.2a, is applicable for the case where the spread in the values of L⊥max is small. The dashed line on the ﬁgure indicates a corresponding result for a large spread, where for the sake of argument, we assumed a probability density of the form

4 64 L⊥max √ P (L⊥max ) = exp(−2(L⊥max /L0 )2 ) , L0 L0 3 2π where L0 is a typical scale length for the conditions.√(For the form of P (L⊥max ) used here, we actually ﬁnd L⊥max = 8/(3 2π)L0 ≈ 1.06L0 .) We have no a priori physical arguments for choosing the probability density given here, but ﬁnd it suﬃciently general to accommodate realistic problems. Now, (11.27) as well as (11.28) assigns a ﬁnite probability for measuring

11 Tests of Time Evolutions in Deterministic Models

289

very small chord lengths, corresponding to cases when the satellite trajectory crosses almost at the boundary of the cavity. In reality such cases are unlikely to be properly recognized in the background noise level and these contributions will be under represented in the experimental estimate for the probability density. It is interesting that (11.27) as well as (11.28) gives a ﬂat maximum for P () for in the range ∼ (1 − 2)L⊥max , see Fig. 11.2. Also, we ﬁnd it comforting that the cylindrical model and the ellipsoidal one in Figs. 11.2 a) and b) are similar, this means that the result is relatively robust and does not depend on ﬁne details in the models. In the foregoing discussion, the length scale L⊥max entered simply as a parameter. It can be related to other basic parameters for the wave-ﬁeld. Thus, the typical length-scale for lower-hybrid cavities formed by the modulational instability can be estimated by balancing the nonlinear term with the dispersive term [11], using (11.23) in (11.21). The estimate for the resulting maximum, B-transverse length scale is given by L⊥max ∼ Λ

2 4n0 Te m Ωce 2 ε0 | ∇φ |20 M ωpe

1/2 ,

(11.29)

where | ∇φ |20 is the square of the electric ﬁeld in the center of the cavity at the initial stage. In case the high-frequency ﬁeld is turbulent with a broad wavenumber range, we expect that the scale-length obtained here have to be shorter than the correlation length of the ﬂuctuations. The B-parallel scale length becomes ( L2⊥max M . (11.30) Lmax ∼ Λ m Also a characteristic time-scale for the collapse process can be estimated by balancing the time derivative with the nonlinear term in (11.21) and using (11.23), giving 2 −1 m Ωce 4n0 Te tLH ∼ ωLH . (11.31) 2 ε | ∇φ |2 M ωpe 0 0 The time scale obtained in this way may, of course, be a somewhat crude estimate. A more accurate expression can be obtained from the growth rate γLH for the modulational instability which can be obtained with some algebra. −1 , at least for order of magnitude estimates. A It was found [11] that tLH ∼ γLH characteristic contraction rate for the cavity evolution based on the analytical estimates is given by the ratio L⊥max /tLH . This contraction rate is below the ion-sound speed consistent with sub-sonic collapse. Assuming that we have determined a characteristic shape for the structures, for instance by estimating an optimum value for m in (11.8), we can attempt to obtain a distribution of the peak absolute value (with positive or negative sign) of the density detected by a satellite pass of a structure. For instance for m = 1 we ﬁnd distributions illustrated in Fig. 11.3 by using (11.10).

10 8 6 4 2

Pn ΣLM R

H.L. P´ecseli and J. Trulsen Pn nLM R

290

a 0 0

0.5

n n

1

5 4 3 2 1 0

b 0

0.1

0.2

n Σ

Fig. 11.3. The conditional probability density for normalized density depletions, n ˜ /∆n is shown in (a), obtained for the preferred model with m = 1 in (11.10), with a given ∆n. Assuming a Rayleigh distribution for the peak densities ∆n we obtain the result shown in (b)

Assuming that we deal with density depletions we have in Fig. 11.3a the distribution of detected density minima, assuming that all cavities are identical, i.e. ∆n is the same for all. The two singularities in P (& n/∆n) are easily understood by simple geometrical arguments. In Fig. 11.3b we illustrate the eﬀect of Rayleigh distributed values for ∆n.

11.5 Observations As already mentioned, our emphasis will be on the statistical properties of the observed density depletions associated with lower hybrid wave cavitation, as detected by the Freja satellite. First we want to investigate to what extent we might assume these cavities to be randomly distributed along the spacecraft orbit. The experimentally obtained distribution of the number of cavities in intervals of given length are shown in Fig. 11.4, together with the probability density for the relative distance between cavities. For the present case we estimate the density of cavities to be approximately µ = 2.5 × 10−3 m−1 . The distributions are very reproducible, also when data are combined from orbits obtained with more than a year time separation. Very large distances are under-represented in the data because of the ﬁnite sample duration (usually 0.75 s). The dots show an exponential ﬁt, exp(−x/ζ), where ζ corresponds to approximately 200 m. The exponential ﬁt indicates that the probability of ﬁnding a cavity in a small interval dx is proportional to dx itself, with a constant of proportionality given by the density, µ, of cavities along the spacecraft trajectory. In particular, we should like to point out that a model with spatially uniformly distributed, statistically independent density depletions results in a Poisson distribution, P (N ) = (µL)N exp(−µL)/N !, for the number of cavities, N , in an interval, L, along the spacecraft trajectory, as discussed in Sect. 11.1. In order to give an estimate for the shape of the cavities, we note that very often we have two probes for density measurements active on Freja. It

25

25

20

20 Occurrence (%)

Occurrence (%)

11 Tests of Time Evolutions in Deterministic Models

15 10

15 10 5

5 0

291

0 0

5

10 Number of cavities

15

0

500

1000

1500

2000

2500

3000

Distance between cavities (m.)

Fig. 11.4. Left: Experimental estimate for the probability density for the number of cavities in a segment of duration 0.375 s (corresponding to a distance of 2330 m along the spacecraft trajectory). Right: distribution of distance between cavities. A slight data gap appears for cavity separations smaller than a typical cavity width. The small ﬁlled circles indicate the appropriate results for a Poisson distribution in both ﬁgures

Fig. 11.5. Scatter plot of the two corresponding chord lengths ( 1 , 2 ) as detected by the two Langmuir probes crossing a cavity at two separated positions. The ﬁgure contains observations of 130 cavities. All points are essentially located on the center line, indicating that the chord-lengths detected by the two probes are independent of the probe separation, giving an indication for the validity of the Gaussian model with m = 1 in (11.8).

is therefore feasible to follow the ideas outlined in Sect. 11.2.2. In Fig. 11.5 we thus show the two dimensional distribution of chord-lengths, as obtained by simultaneous observations of the same cavity by both probes, irrespective of the spin-phase of the satellite, i.e. independent of the component of the probe separation in the direction of the satellite’s velocity vector. We note basically that all observation points are located on the line 1 = 2 , indicating

H.L. P´ecseli and J. Trulsen 25

25

20

20

Occurrence (%)

Occurrence (%)

292

15 10

15 10 5

5

0

0 0

25

50

75

100

Chord lengths (m)

125

150

0

0.05

0.10

0.15

Depth of cavities (relative amplitude dn/n)

Fig. 11.6. Left: Distribution of chord lengths (not to be confused with cavity widths if m = 1 in (11.8)) obtained from the data. Right: Distribution of cavity depths as detected along the spacecraft orbit

an excellent ﬁt to the model assuming m = 1 in (11.10). We can thus argue that the depletions have a Gaussian shape to a good approximation [8]. The distribution of cavity widths obtained on the basis of data from one selected orbit is shown in Fig. 11.6. Here chord lengths are identiﬁed as the separation between the two points of maximum curvature at the baseline of the signal. The distribution shown in Fig. 11.6 is very robust; it is well reproduced for data obtained with a time separation of more than a year. We can also obtain the distribution of the maximum depth of the density depletion along the probe trajectories, with results shown in Fig. 11.6. We note a convincing similarity with the results shown in Fig. 11.3b as obtained for the Gaussian model-shape, which is supported by Fig. 11.5, provided that we include a wide statistical distribution of the peak values of the density depletions over the ensemble of cavity realizations. In Fig. 11.7, we show a scatter-plot of the distribution of density depletions and chord lengths, as detected along the spacecraft orbit. As already demonstrated, we have solid evidence for the Gaussian model, with m = 1 in (11.8) to be representative for the density depletions associated with the cavities [8]. Then the abscissa on Fig. 11.7 in eﬀect gives the cavity width, just as Fig. 11.6.

11.6 Discussions By comparing the observed probability densities for chord lengths with those derived on the basis of a simple collapse process we ﬁnd that the shape of the theoretical probability density are not readily made to agree with the observed distributions. The disagreement is most conspicuous in Fig. 11.7, where we readily note that the deepest cavities are associated with the intermediate scales, and not with the most narrow ones, as one would expect by a collapse model. This observation can be supported even more strongly by data from other passes [8]. We note also nontrivial disagreements when

11 Tests of Time Evolutions in Deterministic Models

293

Fig. 11.7. Left: Scatter plot for distribution of density depletions and chord lengths (which for the m = 1 model equals cavity widths), as detected along the spacecraft orbit. The vertical “stripes” in the ﬁgure are due to the temporal sampling of the data. Right: scatter plot of corresponding values for peak observed relative density variations, n ˜ /n0 in percent, and the available electric ﬁeld components obtained from the peak electric ﬁeld detected at medium frequencies by the antennas within the cavities. The electric ﬁelds are measured as the half peak-to-peak value (pp/2), measured at maximum. For a given electric ﬁeld component, we usually have two measurements for the density depletion, one from each probe. We plot both points, shown with small open and ﬁlled circles, respectively

characteristic times, tLH , and scale sizes, L⊥max , from theory are compared with observations [8]. The seeming disagreement with an interpretation based on collapsing lower-hybrid waves can be substantiated also by use of the analytical expressions (11.21) and (11.22). 11.6.1 Length Scales The observational results speak in favor of a situation where the lower-hybrid waves are of a somewhat “bursty” nature, and it is the distance between the bursts which would determine the separation, while the amplitude and spatial extent of the bursts set the value for L⊥max in the individual cavitations. This interpretation is physically quite plausible and is in itself not contradictory to the collapse model. Concerning individual cavities, the data are not able to provide a direct estimate of L , so we are here only concerned with the characteristic value for the perpendicular length scale. From Figs. 11.6 and 11.7 we estimate L⊥ ≈ 60 m, while L⊥max ∼ 150−200 m. A typical relative density depletion is approximately 25%. It should be brought in mind that the analytical estimate (11.29) can be used at any stage of the collapse [10], i.e. the instantaneous value of the cavity !1/2 2 2 width can be estimated as L⊥est ∼ Λ 4n0 Te mΩce /(ε0 | ∇φ |2 M ωpe ) as an order of magnitude, with ∇φ now being the actual value of the electric ﬁeld in the center of the cavity. Taking the value of 40 mV/m from Fig. 11.1, we ﬁnd L⊥est ∼ 2 m. The electric ﬁeld is a lower limit and L⊥est derived from it is then an upper limit for the cavity diameter. This value cannot by any

294

H.L. P´ecseli and J. Trulsen

means be accommodated within the results of Fig. 11.1. The disagreement between estimates from the collapse model and the observations become even more pronounced by taking | ∇φ | ∼ 70–100 mV/m which are, after all, also being observed, see Fig. 11.7. For these cases, the observed chord lengths are similar to those in Fig. 11.6, while the analytically estimated maximum cavity width should be less than 1 m. Such narrow deep cavities are never observed, although this is, with a slight margin, within the instrumental capability of the detecting circuits. Evidently, the arguments in these sections are based on the assumption that the cavity parameters L⊥ and L do not change appreciably during the time it takes the spacecraft to traverse a cavity. The slow time variation in the density signals from the two probes conﬁrm that this requirement is fully satisﬁed. It is, however, not so for an arbitrary collapse scenario predicted from the model equations (11.21) and (11.22). It could be argued that the transit time damping arrests the contraction of cavities by damping out the lower-hybrid waves at a certain small scale. This argument would, however, imply that predominantly the smallest cavities should be void of wave activity, in disagreement with observations. 11.6.2 Time Scales From the observations (Fig. 11.2) a typical size of the observed cavities perpendicular to the magnetic ﬁeld is ∼60 m. The satellite (with velocity ∼ 6 km/s in the B-perpendicular direction) is traversing this distance in approximately 10 ms. Cavities are frequently traversed without indications of any deformation during the time of passage, as judged from the two density probe signals, see for instance Fig. 11.1. In the case of a signiﬁcant cavity contraction during the passage of the satellite, the time record of the density depletions and corresponding wave-envelopes should appear skew, or signiﬁcantly non-symmetric. Signiﬁcantly skew density variations are observed only very rarely, and when they occur their skewness can have both signs. On the basis of the experimental data, we thus argue that the characteristic time for the cavity evolution must be signiﬁcantly larger than 10 ms, a time-scale of 100 ms is probably an underestimate. This means that at least the H + -component is magnetized on relevant time scales. The characteristic time scale, tLH , for a wave-collapse process estimated from the analytical result (11.31) is, on the other hand, comparable to or smaller than 10 ms and therefore too short to be in agreement with the experimental results. The probability densities derived on the basis of a collapse model are giving unfair representation of the smallest scales, i.e. no collapse model is expected to remain valid when the length scales approaching scales where transit time damping becomes important. This argument is not signiﬁcant for our interpretation because such small scales will escape observation due to their small cross section. The smallest scales thus have negligible weight in our results. The essential element of the collapsing time-variation of the cavity

11 Tests of Time Evolutions in Deterministic Models

295

widths, as given by (11.24), is that the time variation of large scales is slow, while small scales change rapidly. Signiﬁcant modiﬁcations of the obtained statistical distributions will only be obtained if these conditions are reversed, thus making the small scales being those most likely to be observed. This type of time evolution can, however, not be accommodated within a description based on wave collapse.

11.7 Conclusions In this paper we discussed general ideas which can be used for estimating models for coherent time-evolutions by random sampling of data. The ideas may turn out to be particularly useful for interpreting data from instrumented spacecrafts, and were here illustrated by discussions of localized bursts of lower-hybrid waves and correlated density depletions observed on the FREJA satellite [5, 6]. Particular attention was given to an explanation in terms of wave-collapse. The statistical arguments are based on three distinct elements. Two are purely geometric, where the chord length distribution is determined for given cavity scales, together with the probability of encountering those scales. The third part of the argument is based on the actual time variation of scales predicted by the collapse model. The statistical assumption is basically that the cavities are uniformly distributed along the spacecraft trajectory and that they are encountered with equal probability at any time during the dynamic evolution. We believe that the cylindrical and ellipsoidal models discussed are suﬃciently general to accommodate actual forms of a collapsing cavity. The time variation given by (11.24) is only an approximation at early times of the evolution of large cavities. This uncertainty cannot be of importance as the large scales seem not to be signiﬁcantly represented in the data, in spite of their large cross-section. Since the measured electric ﬁelds are somewhat uncertain [1, 3, 8], we have not discussed the statistics of this quantity in any greater detail. We concluded that the interpretation in terms of wave collapse in its simplest form can be ruled out on the basis of a pronounced disagreement between the length and time scales predicted by the collapse-model and those observed in the data. The experimentally observed distribution of chord lengths in the cavities was explained best by assuming a large number of B-elongated cavities, uniformly distributed in space, with values of 40–80 m for the diameters in the direction perpendicular to the ambient magnetic ﬁeld [14]. We proposed [1] a mechanism where the cavities start out with a B-perpendicular scale size very close to the one they end up with. It is assumed that the waves give up energy to ions as well as electrons. This seems to be the best candidate for explaining this characteristic length scale, which is then a consequence of the thermal expansion of the plasma, with electrons streaming along B while ions expand across the magnetic ﬁeld lines for transverse structures smaller than or comparable to twice the ion Larmor diameter of ions. It was demonstrated

296

H.L. P´ecseli and J. Trulsen

[1] that based on this model, a probability density for chord-lengths can be derived which agrees well with observations. With minor modiﬁcations, the statistical analysis presented in the present work can be generalized also for studies of the possible evidence of Langmuir wave collapse in rocket or satellite data. In particular, the ellipsoid approximation discussed here contains also the “pancake” model which is relevant for the Langmuir problem in weak magnetic ﬁelds [15]. It is self evident that the analysis summarized in the present communication refers to coherent phenomena. In case the structures of interest are embedded into a turbulent background, a ﬁltering of the data might be advantageous. Several such methods have been discussed in the literature, conditional sampling for instance [16], but also matched ﬁlters might be useful.

Acknowledgments The present work was carried out as a part of the project “Turbulence in Fluids and Plasmas,” conducted at the Centre for Advanced Study (CAS) in Oslo in 2004/05.

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

P´ecseli, H. L. et al.: J. Geophys. Res. 101, 5299, 1996. Kofoed-Hansen, O., H. L. P´ecseli, and J. Trulsen: Phys. Scr. 40, 280, 1989. Kjus, S. H. et al.: J. Geophys. Res. 103, 26633, 1998. Vago, J. L. et al.: J. Geophys. Res. 97, 16935, 1992. Dovner, P. O., A. I. Eriksson, R. Bostr¨ om, and B. Holback: Geophys. Res. Lett. 21, 1827, 1994. Eriksson, A. I. et al.: Geophys. Res. Lett. 21, 1843, 1994. Schuck, P. W., J. W. Bonnell, and P. M. J. Kintner: IEEE Trans. Plasma Sci. 31, 1, 2003. Høymork, S. H. et al.: J. Geophys. Res. 105, 18519, 2000. Musher, S. L. and B. I. Sturman: Pis’ma Zh. Eksp. Teor. Fiz. 22, 537, 1975, english translation in: JETP lett. 22, 265 1975. Sotnikov, V. I., V. D. Shapiro, and V. I. Shevchenko:, Fiz. Plazmy 4, 450, 1978. Shapiro, V. D. et al.: Phys. Fluids B5, 3148, 1993. Robinson, P. A.: Phys. Fluids B3, 545, 1991. Skjæraasen, O. et al.: Phys. Plasmas 6, 1072, 1999. McBride, J. B., E. Ott, J. P. Boris, and J. H. Orens: Phys. Fluids 15, 2367, 1972. Krasnosel’skikh, V. V. and V. I. Sotnikov: Fiz. Plazmy 3, 872, 1977, see also Sov. J. Plasma Phys. 3, 491, 1977. Johnsen, H., H. L. P´ecseli, and J. Trulsen: Phys. Fluids 30, 2239, 1987.

12 Propagation Analysis of Electromagnetic Waves: Application to Auroral Kilometric Radiation O. Santol´ık1 and M. Parrot2 1

2

Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic; also at IAP/CAS, Prague, Czech Republic. [email protected] LPCE/CNRS, Orl´eans, France. [email protected]

Abstract. We give a brief tutorial description of techniques for determination of wave modes and propagation directions in geospace, based on multi-component measurements of the magnetic and electric ﬁeld ﬂuctuations. One class of analysis methods is based on the assumption of the presence of a single plane wave and can be used to determine the direction of the wave vector. If the wave ﬁeld is more complex, containing waves which simultaneously propagate in diﬀerent directions and/or wave modes, the concept of the wave distribution function becomes important. It is based on estimation of a continuous distribution of wave energy with respect to the wave-vector direction. This concept can furthermore be generalized to the distribution of energy in diﬀerent wave modes. As an example of analysis of satellite data, our emphasis is on application of these techniques to the high-frequency waves, for example to Auroral Kilometric Radiation (AKR). We analyze multicomponent data of the MEMO instrument obtained using multiple magnetic and electric antennas onboard the Interball 2 spacecraft. Results of diﬀerent analysis techniques are compared. The intense structured AKR emission is found to propagate predominantly in the R-X mode with wave energy distributed in relatively wide peaks at oblique angles with respect to the terrestrial magnetic ﬁeld. As expected, the AKR sources correspond to multiple active regions on the auroral oval.

Key words: Electromagnetic waves in plasmas, direction ﬁnding, wave distribution function, auroral kilometric radiation

12.1 Introduction Waves in space plasmas can often simultaneously propagate in diﬀerent modes (with diﬀerent wavelengths) at a given frequency. To trace these waves back to their original source regions and to estimate their source mechanisms, recognition of their modes and propagation directions in the anisotropic plasma O. Santol´ık and M. Parrot: Propagation Analysis of Electromagnetic Waves: Application to Auroral Kilometric Radiation, Lect. Notes Phys. 687, 297–312 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

298

O. Santol´ık and M. Parrot

medium is often crucial. Experimental analysis of propagation modes and directions of waves in plasmas is easier and more reliable if we measure several components of the ﬂuctuating magnetic and electric ﬁelds at the same time. Such measurements using multiple antennas on spacecraft have been ﬁrst proposed by Grard [12] and Shawhan [66]. The complete set of three Cartesian components of the vector of magnetic ﬁeld ﬂuctuations and three Cartesian components of the electric ﬁeld ﬂuctuations (and, in some cases, also the density ﬂuctuations) would be the basis of an ideal data set. However, most often some of those measurements are missing for various, predominantly technical, reasons. For example, it is technically diﬃcult to place a long electric antenna in the direction of the spin axis of a spin-stabilized spacecraft. Several analysis methods applicable to the multi-component measurements have been ﬁrst developed for the ground-based geophysical data [e.g. 2, 35, 36, 54, 55]. These methods and other newly developed techniques [e.g. 8, 25, 58, 64, 71, 72] have been later used for analysis of data of spacecraft missions carrying instruments for multi-component measurements of the wave magnetic and electric ﬁelds in various frequency ranges, such as GEOS, Aureol 3, Freja, Polar, Interball 2, Cluster, Double Star, and DEMETER. Although these missions were designed to investigate diﬀerent regions of the geospace, similar analysis methods have been used for their wave measurements [e.g. 29, 30, 31, 32, 33, 45, 46, 47, 48, 56, 57, 58, 59, 60, 61, 62, 73]. Analysis methods which are described in this chapter rely on this heritage. We will concentrate on their application to the high-frequency waves, namely to Auroral Kilometric Radiation (AKR). AKR is a strong emission of radio waves at frequencies from a few tens of kHz up to 600–800 kHz. It was ﬁrst observed more than 30 years ago by Benediktov et al. [5] and Dunckel et al. [9] but still is a subject of active research. The ﬁrst proper interpretation was done by Gurnett [15], who demonstrated the electromagnetic nature of the waves and found the origin of the radiation in the terrestrial auroral zone at altitudes of a few Earth radii (RE ). A widely accepted generation mechanism is based on the relativistic cyclotron maser interaction with a “horse-shoe” or “shell” electron distribution function [34, 51, 53, 74] connected to active auroral regions [15, 20, 39]. The AKR emissions consist of many narrow-band components with varying center frequencies [see, e.g., Gurnett et al., 16]. The bandwidth of these components is typically 1 kHz but it could be as low as 5–10 Hz [see, e.g., Baumback and Calvert, 3]. This ﬁne structure could be explained by a wave ampliﬁcation in a resonator represented by small ﬁeld-aligned structures in the auroral region [6]. Measurements inside the AKR source have shown that the radiation originates in ﬁeld-aligned density depletions with transverse dimensions of the order of 100 km, ﬁlled by hot and tenuous plasmas [7, 10, 11, 19]. Theory considering relativistic eﬀects on the wave dispersion relation [Pritchett, 51, 52] and small scale gradients [see, e.g., Le Qu´eau and Louarn, 26] appears to be necessary to explain all the details of the wave generation and subsequent propagation from the source cavity.

12 Propagation Analysis of Electromagnetic Waves

299

Since AKR propagates at frequencies higher than all the characteristic frequencies of the plasma medium (above the plasma frequency and the electron cyclotron frequency) it can leave the source region in one of the two basic propagation modes: R-X (right-handed/extraordinary) and L-O (lefthanded/ordinary) [67]. The two modes can be recognized using measurements of phase shifts between the diﬀerent components of the electric and/or magnetic ﬁeld ﬂuctuations. Kaiser et al. [22] analyzed direct measurements of the polarization sense made by multiple electric antennae of Voyager 1 and 2. Supposing that the satellites are always in the northern magnetic hemisphere they found R-X mode in more than 80% of cases. Analysis of wave propagation directions was used in many AKR studies to localize the source region or to verify the generation mechanism. Gurnett [15], using Imp 6 and Imp 8 data, ﬁrst estimated the direction of AKR propagation analyzing the spin pattern of a single electric antenna. At a distance of about 30 RE , he found the wave vector within 6◦ from the Earthward direction. Kurth et al. [24] conﬁrmed his results by a detailed study based on large Hawkeye-1 and Imp-6 data sets. To ﬁnd the propagation direction they extended the spin pattern method using the least-squares ﬁt of the modulation envelope of the electric antenna. However, they had to work with 1-hour averages to achieve suﬃcient angular resolution. Similar results were also obtained by Alexander and Kaiser [1] who analyzed the RAE-2 recordings of lunar occultations of the Earth at radio frequencies to locate the emission region in three dimensions. Gurnett et al. [16] proposed to use a phase interferometry method using two-point measurements of AKR waveforms onboard ISEE-1 and ISEE-2 satellites, and Baumback et al. [4] found by this method that the AKR source has a diameter between 10 and 20 km. James [21] applied the spin-pattern method to the data of ISIS-1 sounder receiver. He measured wave normals declined by 90◦ –140◦ from the terrestrial magnetic ﬁeld, and in a ray-tracing study he inferred originally downward propagation and a subsequent reﬂection. Calvert [6] developed a direction ﬁnding method based on DE-1 onboard correlation measurements. He used signals from two orthogonal electric dipoles, and with a model of their phase diﬀerence with respect to the spin phase he obtained the wave-normal direction. This method was also used by Mellott et al. [37] who studied diﬀerent propagation of R-X and L-O mode emissions. Several studies using the Viking data have again used a simple spin-pattern method. For instance, de Feraudy et al. [7] found that the spin-pattern phase abruptly changes by about 80◦ at the boundary of the AKR source region. Morioka et al. [38] used ﬁve-component measurement of the Akebono satellite to calculate the Poynting ﬂux vector in the satellite frame. With onboard narrow-band receivers they studied phase relations between the ﬁeld components in a band of 100 Hz inside an AKR emission. Gurnett et al. [13] and Mutel et al. [39, 40] implemented the very long baseline interferometer (VLBI) technique to determine locations of individual AKR bursts. They used a triangulation method based on diﬀerential delays from cross-correlated wide-band electric ﬁeld waveforms recorded by

300

O. Santol´ık and M. Parrot

the WBD instruments on board the four Cluster spacecraft and conﬁrmed that the AKR bursts are generally located above the auroral zone with a strong preference for the evening sector. Multi-component measurements of the MEMO (Mesures Multicomposantes des Ondes) instrument onboard the Interball 2 spacecraft have been analyzed by Lefeuvre et al. [32] and Parrot et al. [46], showing that the wave normals of the right-hand quasi-circularly polarized (R-X mode) waves have wave vectors inclined by approximately 30◦ from the direction of the terrestrial magnetic ﬁeld. Santol´ık et al. [58] showed for another AKR case measured by the same instrument at an altitude of 3 RE that the analysis of wave propagation indicates a source region located at an altitude of 1.2 RE above the northern auroral zone. Schreiber et al. [65] performed, based on similar Interball-2 measurements of the MEMO and POLRAD [17] instruments, a three-dimensional ray tracing study using the analysis of wave-normal directions at the spacecraft position. Rays traced back toward the sources imply the existence of two diﬀerent large active regions, as seen at the same time by the ultra-violet camera on board the Polar spacecraft. Analysis presented in this chapter is also based on the data set of the MEMO instrument. The chapter is organized as follows. A short tutorial description of the analysis methods is given in Sect. 12.2. Subsection 12.2.1 introduces methods based on the approximation of the single wave-normal direction. Subsection 12.2.2 then brieﬂy describes methods of estimation of the wave distribution function. We show that this method can be naturally extended to estimate also a distribution of energy between diﬀerent wave modes. Section 12.3 follows showing results of analysis of AKR measured by multiple antennas onboard the Interball 2 spacecraft.

12.2 Analysis Methods The multi-component measurements of the wave magnetic and electric ﬁelds allow us to determine, for example, the average Poynting ﬂux. Supposing the presence of a single plane wave, the direction of the wave vector can be determined. For a more complex wave ﬁeld we can estimate a continuous distribution of wave energy with respect to the wave-vector direction (wave distribution function). These results are useful for the localization of sources of observed emissions. 12.2.1 Plane Wave Methods Supposing the presence of a single plane wave at a frequency f with a wave vector k, the magnetic ﬁeld B as a function of time t and position x can be written ! B(t, x) = B0 + B(f, k)exp i(2πft − k · x) , (12.1)

12 Propagation Analysis of Electromagnetic Waves

301

where B0 is the ambient stationary magnetic ﬁeld (terrestrial magnetic ﬁeld in the magnetosphere), means the real part, and B is the “magnetic vector complex spectral amplitude” for a given frequency f and wave vector k. Under the same circumstances, similar expression can be used for the ﬂuctuating electric ﬁeld E using the “electric vector complex spectral amplitude” E, and similarly also for the “current density complex spectral amplitude” J and √ “charge density complex spectral amplitude” . Using the SI units with 1/ ε0 µ0 = c (speed of light), the Maxwell’s equations can be written for the complex spectral amplitudes (12.2) k × B = i µ0 J − 2πf E/c2 , k × E = 2πf B , k · E = /ε0 ,

(12.3) (12.4)

k · B = 0.

(12.5)

Equation (12.3) (Faraday’s law) implies that B is always perpendicular to both wave vector k and E, k·B =0 (12.6) E ·B =0

(12.7)

Note that the ﬁrst of these conditions (12.6) is equivalent also to the fourth Maxwell’s equation (12.5). If we now write it in Cartesian coordinates, where B = (B1 , B2 , B3 ) ,

(12.8)

and multiply (12.6) successively by three Cartesian components of the complex conjugate B∗ we obtain a set of six mutually dependent real equations S11 S12 S13 S12 S22 S23 k1 S13 S23 S33 · k2 = 0 , A·k = (12.9) 0 −S12 −S13 k 3 S12 0 −S23 S13 S23 0 where means the real part, means the imaginary part, and the components of the Hermitian magnetic spectral matrix Sij are obtained from the three Cartesian components of the magnetic vector complex amplitude using the relation i, j = 1 . . . 3 . (12.10) Sij = Bi Bj∗ , Note that the homogeneous set of equations (12.9) can be multiplied by any real coeﬃcient. Consequently, this set cannot be used to determine the modulus of the unknown vector k. It only can determine the direction of this vector. Note also that the set of equations (12.9) naturally contains only two independent real equations corresponding to the single complex equation

302

O. Santol´ık and M. Parrot

(12.6). The importance of this expansion, however, becomes evident when it is used with experimental data. Using the experimentally measured multi-component signals of the wave magnetic ﬁeld ˆ = (B ˆ2 , B ˆ3 ) ˆ1 , B (12.11) B we can use spectral analysis methods (for example, fast Fourier transform or wavelet analysis) to estimate, at a given frequency, the components of ˆ and, subsequently, the the “magnetic vector complex spectral amplitude” B magnetic spectral matrix, Sˆij = Bˆi Bˆj∗ ,

i, j = 1 . . . 3 ,

(12.12)

where means average value. The homogeneous set of six equations (12.9) can be then rewritten, ˆ ·κ ˆ =0, A

(12.13)

ˆ is composed of the superposed imaginary and real parts of where the matrix A the experimental spectral matrix Sˆij (instead of the idealized spectral matrix ˆ is an unknown unit vector deﬁning the estimate of the wave vector Sij ), and κ direction, ˆ k| ˆ . ˆ = k/| κ (12.14) Using Cartesian coordinates connected to the principal axis of symmetry of the plasma medium where the waves propagate (the direction of the ambient stationary magnetic ﬁeld B 0 ), the wave vector direction can be deﬁned by two angles θ and φ, where θ is the deviation from the B 0 direction and φ is an azimuth centered, for instance, to the plane of the local magnetic meridian (see Fig. 12.1). The wave vector direction then reads ˆ = (sin θ cos φ, sin θ sin φ, cos θ) . κ

(12.15)

ˆ thus reduces to two real unknowns θ and The unknown unit vector κ φ. As a consequence, the system (12.13) is over-determined, containing six

Fig. 12.1. Cartesian coordinate system for the determination of the wave vector ˆ = k/|k| direction κ

12 Propagation Analysis of Electromagnetic Waves

303

equations for two unknowns. Generally, these 6 equations are independent. This is diﬀerent compared to the case of the ideal set of equations (12.9). ˆ is composed of experimental data which The reason is that the matrix A can contain natural and/or experimental noise and do not necessarily exactly correspond to an ideal plane wave. Since we only have two unknowns, a subset of any two independent equations picked up from the set (12.9) is suﬃcient to obtain a unique solution for θ and φ. This is the basis of several analysis methods. The method of Means [36] is based on imaginary parts of three cross-spectra and the procedure is equivalent to solving any two of the last three equations in (12.13). The method of Samson and Olson [55] (their equation 11) is equivalent to ﬁnding a unique solution from another subset of equations selected in (12.13). The method of McPherron et al. [35] uses the ﬁrst three equations and ﬁnds a unique solution using the eigenanalysis of the real part of the spectral matrix. However, the choice of the subset of equations is rather arbitrary and for a diﬀerent subset we do not generally obtain the same result from given experimental data. Other methods thus attempt to estimate an “average” solution of the entire set (12.9) using diﬀerent techniques. Samson [54], again using the eigenanalysis, presented methods of decomposition of the entire complex spectral matrix. Santol´ık et al. [64] used a singular value decomposition (SVD) technique to estimate a solution of the entire set of equations (12.13) in the ˆ “least-squares” sense, decomposing the matrix A ˆ = U · W · VT , A

(12.16)

where U is a matrix 6 × 3 with orthonormal columns, W is a diagonal matrix 3 × 3 of three non-negative singular values, and VT is a matrix 3 × 3 with orthonormal rows. Note that the SVD algorithm can often be found in nuˆ is then directly merical libraries [e.g., 50]. The “least-squares estimate” for κ found as the row of VT corresponding to the minimum singular value at the diagonal of W. The results of the plane-wave analysis often allow a straightforward interpretation of results. This was found useful, for example, in the analysis of sub-auroral ELF hiss emissions from the measurements of the Aureol 3 and Freja spacecraft [33, 62]. These diﬀerent methods often provide us also with estimates of the validity of the initial assumption of the presence of a single plane wave. Diﬀerent deﬁnitions of such an estimator (“degree of polarization”, “polarization percentage” or “planarity”) have been introduced, based on diﬀerent descriptions of the coherence of the magnetic components and their conﬁnement to a single polarization plane [49, 54, 64]. Similar techniques can also allow us to estimate the sense of the magnetic polarization with respect to the ambient stationary magnetic ﬁeld B 0 . This has been used, for example, to analyze electromagnetic emissions in the auroral region by Lefeuvre et al. [30, 31], Santol´ık et al. [59], and Santol´ık and Gurnett [56], using the data of the Aureol 3, Interball 2, and Polar spacecraft, respectively.

304

O. Santol´ık and M. Parrot

The above mentioned SVD technique can also be used with both the measured magnetic and electric components. In that case, an “average” solution to an over-determined set of 36 equations derived from equation (12.3) is esˆ , i.e., to distinguish timated. This allows us to determine also the sign of κ between the two antiparallel wave vector directions [for more details, see 64]. This technique also allows us to estimate the validity of the plane-wave assumption, but this time it is deﬁned as a measure of closeness of the observed wave ﬁelds to equation (12.3). This determination of the “electromagnetic planarity” was, for example, used in [57] to estimate the dimension of the source of chorus emissions from the data of the four Cluster spacecraft. 12.2.2 Wave Distribution Function The concept of wave distribution function is necessary when the wave ﬁeld is more complex, for example when waves from multiple distant sources are simultaneously detected. The wave distribution function (WDF) is deﬁned as a continuous distribution of wave energy with respect to the wave-vector direction; see the review by Storey [69]. It was ﬁrst introduced by Storey and Lefeuvre [70]. The theoretical relation of the WDF to the experimentally measurable spectral matrix has been called the WDF direct problem. Supposing a continuous distribution of elementary plane waves at a frequency f having no mutual coherence and a narrow bandwidth ∆f , the relationship between the spectral matrix Sij (f ) and the WDF Gm (f, θ, φ) is given by 4 Sij (f ) = (12.17) amij (f, θ, φ) Gm (f, θ, φ) d2 κ , m

where m represents the diﬀerent simultaneously present wave modes. The integration is carried out over the full solid angle of wave-normal directions κ, and for a given wave mode m the integration kernels amij are calculated from ∗ (f, θ, φ) ξmi (f, θ, φ) ξmj , (12.18) amij (f, θ, φ) = ∆f um (f, θ, φ) where ξmi , ξmj are complex spectral amplitudes of the ith and jth elementary signals of the wave electric or magnetic ﬁelds, and um is the energy density. All these quantities correspond to an elementary plane wave propagating in a mode m with a normal direction deﬁned by θ and φ. f represents the Dopplershifted frequency in the spacecraft frame. The complex spectral amplitudes of the wave electric or magnetic ﬁelds can be calculated by considering the physical properties of the medium. This calculation requires the knowledge of the theoretical solutions to the wave dispersion relation. Characteristics of the particular wave experiment should also be taken into account. The theory of the WDF direct problem for the cold-plasma approximation has been developed by Storey and Lefeuvre [71, 72], and revisited by Storey [68]. This basic theory has been used by Lefeuvre [27], Lefeuvre and

12 Propagation Analysis of Electromagnetic Waves

305

Delannoy [28] and Delannoy and Lefeuvre [8] to develop practical methods for estimation of the WDF from the spacecraft measurements (the WDF inverse problem). Using a slightly diﬀerent deﬁnition of the WDF and abandoning the explicit dependence of the WDF on the wave frequency, Oscarsson and R¨onnmark [42, 43] and Oscarsson [41] introduced the hot plasma theory into the WDF reconstruction techniques. With the wave-vector dependent WDF they also introduced the Doppler eﬀect in a natural way. Santol´ık and Parrot [60] used the hot plasma theory for the frequency-dependent WDF and further investigated the inﬂuence of the Doppler eﬀect in this more complex situation. Santol´ık and Parrot [63] compared diﬀerent techniques for resolution of the WDF inverse problem, mainly based on the minimization of the least-squares type merit function, in the context of the plane wave estimates. The WDF techniques have been used in numerous studies with both ground based and spacecraft data. For instance, Lefeuvre and Helliwell [29], Parrot and Lefeuvre [45], Hayakawa et al. [18], and Storey et al. [73] used the multi-component measurements of the GEOS spacecraft to characterize the WDF of the ELF chorus and hiss emissions on the equatorial region. Based on the data of the Aureol 3, Akebono, and Freja spacecraft, Lefeuvre et al. [33], Kasahara et al. [23] and Santol´ık and Parrot [61, 63], respectively, estimated the WDF of the down-coming ELF hiss in the sub-auroral and auroral regions. Oscarsson et al. [44] compared the diﬀerent reconstruction schemes for the data of the Freja spacecraft. The up-going funnel-shaped auroral hiss has been investigated by Santol´ık and Gurnett [56] using the WDF analysis of measurements of the Polar spacecraft. Parrot et al. [46] estimated the WDF for auroral kilometric radiation observed on board the Interball spacecraft, compensating at the same time for the a priori unknown experimentally induced phase between the electric and magnetic signals. Simultaneous WDF estimation of the Z-mode and the whistler mode in the auroral region has been done by Santol´ık et al. [59], based on the data of the Interball 2 spacecraft.

12.3 Analysis of Auroral Kilometric Radiation The MEMO instrument on board the Interball 2 spacecraft [32] had two basic modes of measurement: burst mode and survey mode. During the survey mode, low-resolution overview spectrograms were recorded. In its burst mode, the instrument measured waveforms of several components of the electric and magnetic ﬁeld ﬂuctuations in three frequency bands from 50 Hz to 200 kHz. In the high-frequency band (30–200 kHz) which is relevant to this study, the device recorded waveforms measured at the same time by three magnetic antennas and one electric antenna. On 28 January 1997, between 1950 and 2125 UT, the Interball 2 spacecraft was located on the night side and moved over the northern auroral region at altitudes of 2.6-3 RE . During this time interval, the MEMO instrument recorded a highly structured emission of auroral kilometric radiation at frequencies

306

O. Santol´ık and M. Parrot

2125 Polar 2100 2030 2000 1952

1952 Interball 2 2125

60o 30o 0o h 14h 16h 18h 20h 22h 0h2

Fig. 12.2. Comparison of AKR observed by the Interball 2 and Polar spacecraft on 28 January 1997 between 1950 and 2125 UT. Left: Time-frequency spectrograms recorded by the MEMO instrument onboard Interball 2 (bottom) and by the SFR analyzer of the PWI instrument onboard Polar. The same time interval and frequency range is used for both spectrograms. Coordinates of the two spacecraft are given on the bottom of the spectrograms. Over-plotted white lines show the local electron cyclotron frequency. Right: Portions of orbits of the two spacecraft in the corresponding time interval and their projections along the magnetic ﬁeld lines onto the Earth’s surface. The arrows pointing out of the Interball-2 orbit show the average wave-vector directions for two burst-mode intervals at a frequency of 80 kHz (see Fig. 12.5)

between 30 kHz and more than 270 kHz, as reported in [46]. At the same time, the Polar spacecraft also moved over the northern auroral zone but at approximately twice the altitude and on the evening side. The PWI instrument on board Polar [for details of the instrument see, Gurnett et al., 14] observed the same AKR emission. Comparison of observations of the Interball 2 and Polar spacecraft is given in Fig. 12.2 on the left. The strong AKR emission between 2030 and 2125 UT is clearly seen on both spectrograms, demonstrating the global character of AKR. Some detailed time-frequency

12 Propagation Analysis of Electromagnetic Waves

307

Fig. 12.3. Spectrograms calculated from one snapshot of burst-mode MEMO waveform data recorded on 28 January 1997 after 2107:15.48 UT. The spectrograms show the electric component in the frequency interval 0–220 kHz with a frequency resolution of 260 Hz (left) and detailed analysis of pronounced structures around 32 kHz (right) with a frequency resolution of 52 Hz. Black rectangle represents data used later for wave propagation analysis (Fig. 12.4). The local electron cyclotron frequency is 23.5 kHz

features are, however, diﬀerent. Some of those diﬀerences may be attributed to diﬀerent sensitivities of the two instruments as a function of frequency, but most of them are probably owing to diﬀerent positions of the two spacecraft. As their local time is diﬀerent, the two spacecraft can observe AKR emitted from diﬀerent portions of the auroral oval. This also concerns slight timing diﬀerences for the onset of intense AKR around 2030 UT. During this orbit, several intervals of burst-mode data were recorded by the MEMO instrument on board the Interball 2 spacecraft. One electric and three magnetic components were always sampled at 533.3 kHz during 0.32 sec. Figure 12.3 shows the power spectrograms calculated from the electric-ﬁeld data measured during one of those burst-mode intervals. We can see that, at short time scales, the emission still has a complex time-frequency structure with many spectral lines observed at slowly drifting frequencies. Detailed analysis of the middle portion of this burst-mode interval has been done using the available data of one electric and three magnetic ﬁeld components. The wave distribution function (WDF), representing the distribution of the wave energy for diﬀerent wave-vector directions and wave-propagation

308

O. Santol´ık and M. Parrot

Fig. 12.4. Detailed analysis of the multicomponent MEMO data in the frequency band of 0.4 kHz around 32.1 kHz, between oﬀsets 0.19 and 0.22 sec (see Fig. 12.3). The wave distribution function (WDF) is plotted in the coordinate system connected to the Earth’s magnetic ﬁeld B0 and to the local magnetic meridian (Fig. 12.1). The polar diagrams represent the energy density as a function of angles θ and φ, on the left for downgoing waves, and on the right for upgoing waves, on the top for the R-X mode and on the bottom for the L-O mode

modes, is shown in Fig. 12.4. We use a narrow frequency band around the maximum of the power-spectral density of the intense spectral line observed around 32 kHz. A cold plasma model is used to calculate theoretical wave spectral densities, necessary for simultaneous WDF estimation in both R-X and L-O propagation modes and for both hemispheres of wave-normal directions (upgoing and downgoing with respect to the direction of the ambient magnetic ﬁeld). The method of “discrete regions” was used to estimate the WDF [63]. It is based on least-squares optimization of the non-negative WDF values in the total number of 824 discrete regions, homogeneously covering

Fig. 12.5. Wave distribution function for the upgoing waves in the R-X mode propagating in the frequency band of 2.2 kHz centered at 80.1 kHz. The same coordinate system as in Fig. 12.4 is used. Two 0.32-s burst-mode snapshots have been used: (a) time interval starting at 2037:23.58 UT; (b) time interval starting at 2107:15.48 UT. Overplotted are plane wave estimates: triangles for the method of McPherron et al. [35], squares for the method of Means [36], diamonds for the SVD method [64]

the 4π solid angle of wave normal direction in the two propagation modes. The results show that nearly all the wave energy is found in the upgoing R-X mode. Within statistical uncertainties of our analysis, we can neither exclude nor conﬁrm the presence of a small fraction of the L-X mode. Figure 12.5 shows a comparison of obtained results of wave propagation analysis for two diﬀerent time intervals separated by 30 minutes. The second of these time intervals is the same as in Fig. 12.3, and both of them were analyzed using diﬀerent methods by Parrot et al. [46]. For the WDF estimation in Fig. 12.5 we use the model of Gaussian peaks [63]. This model describes concentration of wave energy near one or several wave-normal directions. Information about the WDF can then be reduced to a set of the peak directions and accompanied by the respective energy densities and parameters describing the degree of concentration of the wave energy near the respective directions (peak widths). In this case we model the WDF G as a single peak represented by a Gaussian function in 2D space of wave-normal directions (or, equivalently, on the surface of a unit sphere),

2 [1 − cos θ0 cos θ − sin θ0 sin θ cos(φ0 − φ)] Γ exp − , G(θ, φ) = π∆2 ∆2 (12.19) where Γ is the peak energy density, ∆ is the peak width, and θ0 and φ0 deﬁne the central direction. These free parameters are optimized by a nonlinear least-

310

O. Santol´ık and M. Parrot

squares method to obtain the best model (12.17) for the given experimental spectral matrix. The results show that the emission propagates at oblique angles from the Earth’s magnetic ﬁeld, at diﬀerent azimuth angles φ for the two time intervals. The obtained WDF estimates show large peak widths, (∆ = 20◦ for Fig. 12.5a and ∆ = 28◦ for Fig. 12.5b). The plane wave results are localized within the 75% intensity level compared to the peak maximum. The diﬀerences between the diﬀerent plane wave estimates are larger for a larger width of the WDF peak. The corresponding angular sizes of the global sources of observed AKR have been investigated by Schreiber et al. [65] leading to the suggestion that we observe superposed waves from many elementary sources. The positions of these sources roughly corresponded to active auroral regions remotely sensed by the Ultraviolet Imager (UVI) instrument onboard the Polar spacecraft.

12.4 Conclusions We have shown that wave propagation analysis based on multicomponent spacecraft data can be a useful tool for investigation of high-frequency wave phenomena. Several analysis methods have been described, some using the plane wave approach and others based on a general continuous distribution of wave energy for diﬀerent propagation directions. This second approach has been shown to be useful also for recognition of the propagation mode of observed waves. Observations of AKR by the MEMO instrument on board the Interball-2 spacecraft have been analyzed using these methods. We have shown that the observed highly structured AKR emission propagates predominantly in the R-X mode. The propagation analysis has allowed us to trace the waves back to the source regions which, consistently with previously published results, corresponded to large active regions on the auroral oval. These examples demonstrate the value of using the described analysis methods in future in situ investigations of high frequency waves in geospace and in the solar wind.

Acknowledgments We thank D. A. Gurnett of the University of Iowa, PI of the Polar PWI instrument for the SFR data used in Fig. 12.2. We thank J. D. Menietti of the University of Iowa for useful discussions. This work was supported by the ESA PECS contract No. 98025 and by the GACR grant 202/03/0832.

References [1] Alexander, J.K. and M.L. Kaiser: J. Geophys. Res. 81, 5948, 1976. [2] Arthur, C.W., R.L. McPherron, and J.D. Means: Radio Sci. 11, 833, 1976.¨ o

12 Propagation Analysis of Electromagnetic Waves

311

[3] Baumback, M.M. and W. Calvert: Geophys. Res. Lett. 5, 857, 1978. [4] Baumback, M.M., D.A. Gurnett, W. Calvert, et al.: Geophys. Res. Lett. 13, 1105, 1986. [5] Benediktov, E.A., et al.: Kossm. Issled. 3, 614, 1965. [6] Calvert, W.: Geophys. Res. Lett. 12, 381, 1985. [7] de Feraudy, H., et al.: Geophys. Res. Lett. 14, 511, 1987. [8] Delannoy, C. and F. Lefeuvre: Comp. Phys. Comm. 40, 389, 1986. [9] Dunckel, N., et al.: J. Geophys. Res. 75, 1854, 1970. [10] Ergun, R.E., et al.: Astrophys. J. 538, 456, 2000. [11] Ergun, R.E., et al.: Geophys. Res. Lett. 25, 2061, 1998. [12] Grard, R.: Ann. Geophys. 24, 955, 1968. [13] Gurnett, D.A., et al.: Ann. Geophys. 19, 1259, 2001. [14] Gurnett, D.A., et al.: Space Sci. Rev. 71, 597, 1995. [15] Gurnett, D.A.: J. Geophys. Res. 79, 4227, 1974. [16] Gurnett, D.A., et al.: Space Sci. Rev. 23, 103, 1979. [17] Hanasz, J., et al.: J. Geophys. Res. 106, 3859, 2001. [18] Hayakawa, M., M. Parrot, and F. Lefeuvre: J. Geophys. Res. 91, 7989, 1986. [19] Hilgers, A.: Geophys. Res. Lett. 19, 237, 1992. [20] Huﬀ, R.L., et al.: J. Geophys. Res. 93, 11445, 1988. [21] James, H.G.: J. Geophys. Res. 85, 3367, 1980. [22] Kaiser, M.L., et al.: Geophys. Res. Lett. 5, 857, 1978. [23] Kasahara, Y., et al.: J. Geomag. Geoelectr. 47, 509, 1995. [24] Kurth, W.S., M.M. Baumback, and D.A. Gurnett: J. Geophys. Res. 80, 2764, 1975. [25] LaBelle, J. and R.A. Treumann: J. Geophys. Res. 97, 13789, 1992. [26] Le Qu´eau, D. and P. Louarn: Planet. Space Sci. 44, 211, 1996. [27] Lefeuvre, F.: Analyse de champs d’ondes ´ electromagn´etiques al´eatoires observ´ees dans la magn´etosph` ere, ` a partir de la mesure simultan´ ee de leurs six composantes. Doctoral thesis, Univ. of Orl´eans, Orl´eans, France, 1977. [28] Lefeuvre, F. and C. Delannoy: Ann. Telecommun. 34, 204, 1979. [29] Lefeuvre, F. and R.A. Helliwell: J. Geophys. Res. 90, 6419, 1985. [30] Lefeuvre, F., et al.: Ann. Geophys. 4, 457, 1986. [31] Lefeuvre, F., et al.: Ann. Geophys. 5, 251, 1987. [32] Lefeuvre, F., et al.: Ann. Geophys. 16, 1117, 1998. [33] Lefeuvre, F., et al.: J. Geophys. Res. 97, 10601, 1992. [34] Louarn, P., et al.: J. Geophys. Res. 95, 5983, 1990. [35] McPherron, R.L., C.T. Russel, and P.J. Coleman, Jr.: Space Sci. Rev. 13, 411, 1972. [36] Means, J.D.: J. Geophys. Res. 77, 5551, 1972. [37] Mellott, M.M., R.L. Huﬀ, and D.A. Gurnett: Geophys. Res. Lett. 12, 479, 1985. [38] Morioka, A., H. Oya, and A. Kobayashi: J. Geomagn. Geoelectr. 42, 443, 1990. [39] Mutel, R.L., D.A. Gurnett, and I.W. Christopher: Ann. Geophys. 22, 2625, 2004. [40] Mutel, R.L., et al.: J. Geophys. Res. 108, 1398, doi: 10.1029/2003JA010011, 2003. [41] Oscarsson, T.: J. Comput. Phys. 110, 221, 1994. [42] Oscarsson, T. and K.R¨ onnmark: J. Geophys. Res. 94, 2417, 1989. [43] Oscarsson, T. and K. R¨ onnmark: J. Geophys. Res. 95, 21,187, 1990.

312

O. Santol´ık and M. Parrot

[44] Oscarsson, T., G. Sternberg, and O. Santol´ık: Phys. Chem. Earth (C) 26, 229, 2001. [45] Parrot, M. and F. Lefeuvre: Ann. Geophys. 4, 363, 1986. [46] Parrot, M., et al.: J. Geophys. Res. 106, 315, 2001. [47] Parrot, M., et al.: Ann. Geophys. 21, 473, 2003. [48] Parrot, M., et al.: Ann. Geophys. 22, 2597, 2004. [49] Pin¸con, J.L., Y. Marouan, and F. Lefeuvre: Ann. Geophys. 10, 82, 1992. [50] Press, W.H., et al.: Numerical Recipes. Cambridge Univ. Press, New York, 1992 [51] Pritchett, P.L., et al.: J. Geophys. Res. 107, 1437, doi: 10.1029/2002JA009403, 2002. [52] Pritchett, P.L.: J. Geophys. Res. 89, 8957, 1984. [53] Roux, A., et al.: J. Geophys. Res. 98, 11657, 1993. [54] J.C. Samson: Geophys. J. R. Astron. Soc. 34, 403, 1973. [55] Samson, J.C. and J.V. Olson: Geophys. J. R. Astron. Soc. 61, 115, 1980. [56] Santol´ık, O. and D.A. Gurnett: Geophys. Res. Lett. 29, 1481, doi: 10.1029/ 2001GL013666, 2002. [57] Santol´ık, O., et al.: Geophys. Res. Lett. 31, L02801, doi: 10.1029/ 2003GL018757, 2004. [58] Santol´ık, O., et al.: J. Geophys. Res. 106, 13191, 2001. [59] Santol´ık, O., et al.: J. Geophys. Res. 106, 21137, 2001. [60] Santol´ık, O. and M. Parrot: J. Geophys. Res. 101, 10639, 1996. [61] Santol´ık, O. and M. Parrot: J. Geophys. Res. 103, 20469, 1998. [62] Santol´ık, O. and M. Parrot: J. Geophys. Res. 104, 2459, 1999. [63] Santol´ık, O. and M. Parrot: J. Geophys. Res. 105, 18885, 2000. [64] Santol´ık, O., M. Parrot, and F. Lefeuvre: Radio. Sci. 38, 1010, doi: 10.1029/ 2000RS002523, 2003. [65] Schreiber, R., et al.: J. Geophys. Res. 107, 1381, doi: 10.1029\-/\-2001\ -JA009061, 2002. [66] Shawhan, S.D.: Space Sci. Rev. 10, 689, 1970. [67] Stix, T.H.: Waves in Plasmas. Am. Inst. of Phys., New York, 1992. [68] Storey, L.R.O.: Ann. Geophys. 16, 651, 1998. [69] Storey, L.R.O.: The measurement of wave distribution functions. In: M.A. Stuchly, editor, Modern Radio Science 1999, page 249, Oxford University Press, Oxford, 1999. [70] Storey, L.R.O. and F. Lefeuvre: Theory for the interpretation of measurements of a random electromagnetic wave ﬁeld in space. In M.J. Rycroft and R.D. Reasenberg (Eds.), Space Research XIV, page 381, Akademie-Verlag, Berlin, 1974. [71] Storey, L.R.O. and F. Lefeuvre: Geophys. J. R. Astron. Soc. 56, 255, 1979. [72] Storey, L.R.O. and F. Lefeuvre: Geophys. J. R. Astron. Soc. 62, 173, 1980. [73] Storey, L.R.O., et al.: J. Geophys. Res. 96, 19469, 1991. [74] Wu, C.S. and L.C. Lee: Astrophys. J. 230, 621, 1979.

13 Phase Correlation of Electrons and Langmuir Waves C.A. Kletzing1 and L. Muschietti2 1

2

Department of Physics and Astronomy, University of Iowa [email protected] Space Sciences Lab., University of California [email protected]

Abstract. Multiple spacecraft observations have conﬁrmed the ubiquitous nature of Langmuir waves in the presence of auroral electrons. The electrons show variations consistent with bunching at or near the plasma frequency. Linear analysis of the interaction of a ﬁnite Gaussian packet of Langmuir waves shows that there are two components to the perturbation to the electron distribution function, one in-phase (or 180◦ out-of-phase) with respect to the wave electric ﬁeld called the resistive component and one which is 90◦ (or 270◦ ) out-of-phase with respect to the electric ﬁeld. For small wave packets, the resistive perturbation dominates. For longer wave packets, a non-linear analysis is appropriate which suggests that the electrons become trapped and the reactive phase dominates. Rocket observations have measured both components. The UI observations diﬀer from those of the UC Berkeley observations in that a purely reactive phase bunching was observed as compared to a predominantly resistive perturbation. The resistive phase results of the UC Berkeley group were interpreted as arising from a short wave packet. The UI observations of the reactive phase can be explained by either a long, coherent train of Langmuir waves or that the narrower velocity response of the UI detectors made it possible to capture only one side of the reactive component of the perturbed distribution function for a short wave packet in the linear regime. Future wave-particle correlator experiments should be able to resolve these questions by providing more examples with better velocity space coverage.

Key words: Electron bunching, Langmuir waves, resistive and reactive bunching phases, phase correlators

13.1 Introduction The precipitation of auroral electrons provides an example of a beam-plasma interaction which generates Langmuir waves from the free energy in the electrons. The resulting waves play several important roles in the Earth’s auroral ionosphere. First, the waves Landau damp on the thermal electron population C.A. Kletzing and L. Muschietti: Phase Correlation of Electrons and Langmuir Waves, Lect. Notes Phys. 687, 313–337 (2006) c Springer-Verlag Berlin Heidelberg 2006 www.springerlink.com

314

C.A. Kletzing and L. Muschietti

and thereby form a direct conduit for energy exchange between the auroral electron beam and the thermal electrons. Several authors have speculated that Langmuir waves play a signiﬁcant role in establishing the electron temperature in the auroral ionosphere [3, 15]. In addition to heating electrons, the Langmuir/upper hybrid waves radiate away some of their energy into electromagnetic radiation, which can serve for remote sensing of auroral plasma processes from ground level and from satellites. For example, auroral roar is an EM emission observed near 2–3 and 4–4.5 MHz at ground level [14, 30] and from satellites [1, 12]. Understanding these auroral wave emissions is important not only to fully understand terrestrial aurora and related phenomena, but also because they shed light on analogous emission processes elsewhere in the solar system and beyond. For example, the generation of auroral roar is similar to that of terrestrial continuum radiation, which is generated via mode conversion of upper hybrid waves at the plasmapause, and possibly continuum radiations at other planets as well. Solar type III radiation results from mode conversion of Langmuir waves in the solar wind, and recent observations of structured type III emission [Reiner et al., 23, 24] indicate the signiﬁcance of frequency structure in the causative Langmuir waves. High frequency (HF) electric ﬁeld observations in the topside auroral ionosphere, for which the plasma frequency is typically greater than the electron cyclotron frequency, have revealed plasma waves in the range fpe ≤ f ≤ fuh ever since the earliest measurements [2, 29]. Simultaneous wave and electron distribution measurements have shown that the waves are excited by Landau resonance but that temporal variation of the distribution function or wave refraction from vertical density gradients can limit wave amplitudes as shown by McFadden et al. [19]. Many examples of waves near fpe were observed using Aureol/ARCAD 3 satellite wave receivers and the free-energy source was identiﬁed as the electrons [Beghin et al., 3]. Although rocket-borne receivers have detected HF waves at E-region altitudes attributed to generation by secondary electrons [Kelley and Earle, 13], most observations pertain to altitudes from 300 km and up, where the waves have amplitudes ranging from less than 1 mV/m to as large as 1 V/m [4, 6, 8, 17, 18, 28] These waves are highly bursty in time, sometimes lasting as little as 1 ms but sometimes appearing continuous for ∼1s. Recent high-resolution experiments reveal that they can have complex frequency structure [17, 18, 25]. Other investigators have detected plasma waves through particle-particle correlator techniques. Using a rocket-borne electron detector of large geometric factor, ∼5% modulation at frequencies 4.2–5.6 MHz was found during a 7-second interval when the rocket passed the boundary of two oppositely directed Birkeland sheet currents [27]. Strong modulation (∼30%) was observed at 2.65 MHz of 4–5 keV electrons, corresponding to energies at which a positive slope was observed in the perpendicular and parallel velocity distribution functions [Gough and Urban, 11]. Rocket measurements in which 1.4 MHz ﬂuctuations were detected in the 7.5 keV electrons, just below the electron beam energy, were reported to

13 Langmuir Wave-Electron Phase Correlation

315

occur simultaneously with a positive slope in the electron distribution function [Gough et al., 10]. Linear theory explains the existence and parallel polarization of auroral Langmuir waves [see, e.g., Nicholson, 22]. Theory and simulations have shown that Langmuir waves in the auroral plasma most likely do not develop into strong turbulence but that observed non-linear features in amplitudes and modulations of the waves are consistent with nonlinear interactions between the Langmuir waves and ion acoustic waves [Newman et al., 21]. More recently, the diﬀerence between Langmuir wave-electron interactions both at rocket altitudes and at the altitude of the Freja spacecraft have been studied, ﬁnding that standard quasi-linear diﬀusion theory does not hold for large amplitude Langmuir waves at Freja altitudes, but should hold at lower altitudes [Sanbonmatsu et al., 26]. Correlating waves and particles directly probes the physics of their interaction by providing a superior picture of the microphysics compared to statistically associating an unstable feature of the distribution function with the presence of waves. Elementary theory implies that if the electrons and the waves are exchanging energy, the electrons will have an oscillatory component at a velocity equal to the phase velocity of the waves. Identifying the velocity at which the electron distribution function has this oscillatory component determines the wave phase velocity and therefore the wave number. Measuring the phase of this oscillatory component of the electron distribution function relative to the wave electric ﬁeld yields further information. The phase bunching splits into two pieces: 1) the resistive component, in phase with the electric ﬁeld indicating wave-electron energy exchange; and 2) the reactive component, 90◦ out of phase with the electric ﬁeld indicating trapping [Nicholson, 22]. Detailed treatment of phase relationships by Muschietti et al. [20] for Gaussian Langmuir wave packets, show that the linear perturbation in the distribution function may be considered as the sum of a resistive and reactive components. For both components, the perturbation narrows and increases in magnitude as the wave packet length increases. Electron detectors with broad energy resolution can only detect the resistive component because it has only positive polarity and adds over the entire energy range. Detecting the bipolar nature of the reactive component requires narrow energy response (∆v/v ≤ 5 − 6%) detectors. Only three high frequency (MHz) wave-particle correlation experiments have been reported in the literature [7, 9, 16]. Very similar experiments have been tried on FREJA by Boehm et al. [5] and on FAST by Ergun et al. [7] but were limited in phase resolution. The ﬁrst two of these experiments used a correlator that worked by binning individual detected particles according to the phase of the strongest wave detected by a broadband (0.2–5 MHz) wave receiver [Ergun et al., 7, 9]. The most recent experiment used a correlator which had higher phase resolution and detected electrons bunched 90◦ outof-phase with respect to the electric ﬁeld, suggesting non-linear evolution and trapping of the electrons.

316

C.A. Kletzing and L. Muschietti

In what follows, we discuss the theory of linear perturbations to the distribution by a Gaussian wave packet and also present the theory of extended wave packets via a BGK analysis. We then describe in detail two of the three high frequency wave-particle correlator measurements referred to above. We then conclude with a discussion of these results in the context of theory.

13.2 Finite-Size Wave Packet in a Vlasov Plasma Consider a coherent, Langmuir packet propagating in the x direction with a phase ψ = kx − ωt where the frequency ω is close to the plasma frequency ωp . The wave packet is assumed to be localized, which we specify by a form factor η(x, t) which is piecewise continuous, bounded, and vanishing for x → ±∞. Its slow time dependence describes the drift due to the group velocity of the Langmuir wave or the growth (or damping) due to the interaction with the electrons. Explicitly, the wave electric ﬁeld is written as E(x, t) = E0 η(x, t)eiψ + c.c.

(13.1)

where c.c. is the complex conjugate. The electrons are assumed to consist of two populations: a dense background and energetic, streaming particles. In the problem that we treat, the density of the energetic electrons is orders of magnitude smaller than the background density, so that only the latter determines the real part of the dispersion relation. The electrons are described by a homogeneous distribution function F (v) which includes velocities around the phase velocity of the Langmuir wave, vp ≡ ω/k. Under the inﬂuence of the wave ﬁeld, the streaming population develops a time-dependent, inhomogeneous component f (x, v, t) which satisﬁes the Vlasov equation

∂ ∂ d eE ∂ f≡ +v (F + f ) (13.2) f= dt ∂t ∂x m ∂v where −e and m are the electron charge and mass, respectively. Our goal is to ﬁnd explicit expressions for f (x, v, t) and to analyze their phases versus the wave phase ψ in view of correlator applications. In the linear approximation, one neglects the perturbation ∂f /∂v on the right-hand side of (13.2) and formally integrates the equation, eE0 fL (x, v, t) = m

t

−∞

η(x , t )eiψ(x ,t )

∂ F (v ) dt + c.c. ∂v

(13.3)

The integration is carried out along the trajectories x (t ), v (t ) that have x and v for end points at t = t, namely x (t) = x and v (t) = v. The perturbation at large negative times when x → ±∞ is ignored. From (13.3) one sees that the linear perturbation fL oscillates in time with the same

13 Langmuir Wave-Electron Phase Correlation

317

frequency as the electric ﬁeld. One also notices that the perturbed distribution depends upon the structure of the form factor η along the past trajectories of the particles, which leads to a phase shift relative to the present wave phase ψ(x, t). 13.2.1 Linear Perturbation of the Electrons Taking as characteristics the straight trajectories v (t ) = v and x (t ) = x + v(t − t), where v is constant and positive, one rewrites (13.3) as fL (x, v, t) = A0 ei(kx−ωt)

∂F ∂v

0

η(x + vτ, τ + t)ei(kv−ω)τ dτ + c.c. (13.4) −∞

where A ≡ eE0 /m. The integral explicitly relates the distribution fL to the motion of the particles at earlier times, so that the eﬀect of spatial gradients in the electric ﬁeld amplitude will be directly seen in the resulting phase relation. Let us for instance consider the simple proﬁle of a square window with a width L, 0 : if x < 0 η(x) = 1 : if 0 < x < L (13.5) 0 : if x > L From (13.4), one readily obtains iAeiψ ∂F : if 0 < x < L 1 − βx + c.c. (13.6) fL (x, v, t) = ω − kv ∂v β (x−L) − β x : if L < x where β ≡ exp i(ω/v −k). In the square bracket to the right, the exponents reﬂect the presence of the boundaries and modify the phase of the perturbation relative to the wave phase ψ. They also yield a ballistic term β (x−L) −β x downstream of the interaction region. An important point is that these boundary terms keep the perturbed distribution function of resonant electrons bounded. The usual expression for a plane wave [(13.6) with the square bracket equal to unity] shows the perturbation to have a singularity at resonance, where the assumed linear solution, therefore, breaks down. Instead, the right-hand side for 0 < x < L can be expanded for velocities close to the phase velocity vp . The resulting expression is ﬁnite. One can then diﬀerentiate it with respect to v and thus evaluate at v = vp the derivative we neglected in the linearization of Vlasov equation (13.2). Using the notation F ≡ ∂F/∂v, the result can be written as ∂fL ∂lnF 2A 2 |v = F (vp ) − 1)kx cos ψ + (kx) sin ψ (13.7) (vp ∂v p ωvp ∂v The term proportional to sin ψ (thus out of phase with the electric ﬁeld) is seen to grow quadratically with the distance into the packet. Therefore, the packet must have a ﬁnite extent to assure the validity of the linear perturbation,

318

C.A. Kletzing and L. Muschietti

which we write (kL)2 ωvp /(2A). This inequality introduces an important parameter to the problem at hand 2 2eE0 k kL . (13.8) µ≡ m ω This quantity measures the eﬀect of the localized electric ﬁeld on the electrons and can be used as a small expansion parameter. We can think of it as the square of the bounce frequency times the transit duration of a resonant particle. 13.2.2 Case of a Gaussian Packet A more realistic envelope of wavepacket is provided by a Gaussian. In fact, since the dispersion relation of Langmuir waves is quadratic, a Gaussian packet moving at the group velocity describes well a propagating Langmuir packet. Its dispersion time is given by td ≈ (L/λd )2 ωp−1 with λd the Debye length. This time is very long compared to, for example, the transit duration of resonant electrons, tt ≈ kL ωp−1 . For simplicity, we choose here a static form factor η(x) = exp[−x2 /(2L)2 ] ,

(13.9)

which is justiﬁed since for many applications the group velocity u is much smaller than the velocity of resonant electrons, u/vp = 3 (kλd )2 1. After substituting (13.9) into (13.4) and some algebra, one obtains fL (x, v, t) = A η(x)ei(kx−ωt)

∂F L (−i)Z(ξ) + c.c. ∂v v

(13.10)

where Z is the usual plasma dispersion function, yet has here a completely diﬀerent argument: L x ξ = (ω − kv) − i . (13.11) v 2L The real part of the argument is the Doppler-shifted frequency seen by a traversing electron times its transit duration through the wavepacket. It determines the proximity to resonance during the interaction. The imaginary part of the argument describes the position with respect to the center of the packet. Let us evaluate now (13.10) at the center of the packet. We can deﬁne a resonance function that represents the response of the electron distribution to the wave ﬁeld: ξr 2 2 L √ −ξr2 −iL Z(ξr ) = πe − i 2e−ξr ey dy (13.12) R≡ v v 0 The real part of R, associated with the Gaussian e−ξr , represents the resistive contribution (in phase with the electric ﬁeld). The imaginary part, associated 2

13 Langmuir Wave-Electron Phase Correlation

319

Resistive component

150. kL=80

100.

50.

kL=30 kL=15

0. -0.1

0.0

0.1

Reactive component

100. kL=80

50.

kL=30 kL=15

0. -50. -100. -0.1

0.0

0.1

Fig. 13.1. Resonance function deﬁned by (13.12). Real and imaginary parts of R are shown for various packet sizes (Reprinted with permission from American Geophysical Union)

with the Dawson integral, represents the reactive contribution (out of phase with the electric ﬁeld). When the packet is large so that ωL/v → ∞, Re(R) tends to the usual delta function of the Plemelj formula, πδ(kv−ω), and Im(R) tends to the principal part of 1/(kv − ω). Plots of real and imaginary parts of R are displayed in Fig. 13.1 for various sizes of packets. One clearly sees the tendency to a delta function for large kL and, conversely, the broadening of the resonance for a more localized wavepacket. The perturbed oscillating distribution has components in and out of phase with respect to the electric ﬁeld. Let us split fL of (13.10) into resistive and reactive terms, in a way that emulates the procedure performed by the wave correlator on rocket data. One obtains fL =

µ vp2 F (v) η(x) [Zi cos(kx − ωt) + Zr sin(kx − ωt)] kL v

(13.13)

where the acceleration factor A = eE0 /m has been rewritten in terms of µ using (13.8). The perturbation is thus either rather resistive or reactive depending upon the relative weight of Zi and Zr . Its amplitude is proportional to the wave amplitude, to the packet size, and to the slope of the distribution F . Figure 13.2 displays the perturbation in phase space by means of contours. Solid contours with gray shadings indicate a positive value, i.e. an

320

C.A. Kletzing and L. Muschietti e

n

Te

dis tance kx

30.

30.

20.

20.

10.

10.

0.

0.

-10.

-10.

-20.

-20.

-30.

-30.

-40.

-40.

-50.

-50.

0.90

0.95

1.00

1.05

1.10

-0.01

0.01

velocity v/vp

Fig. 13.2. Linear perturbed distribution as described by (13.13) with power-law model F (v) ∼ v −4 . Contours with gray shadings indicate enhancements at levels 0.006 (light) and 0.018 (dark). Dashed contours denote depletions at levels −0.006 and −0.018. The perturbation (normalized to F (vp )) features a chain of bunchellipses centered at the phase velocity vp . Panel marked δn at right displays the resonant density perturbation, split into its resistive (thick line) and reactive (thin line) components. Rightmost panel shows the potential φ of the Langmuir packet with characteristics: kL = 30, µ = 0.09

accumulation of electrons. Dotted contours indicate a negative value, i.e. a dearth of electrons. The bunching of the particles yields a chain of ellipses centered on the phase velocity vp and zeroes of the potential, which is shown in the rightmost panel. The panel marked δn displays both the resistive (thick line) and reactive (thin line) components of the density perturbation. It is clear that the linear perturbation is mostly resistive. In addition, from the maxima of δn being in phase with ∂φ/∂x > 0 we can conclude that the wave accelerates the electrons, whereby it is being damped. This is consistent with the power-law model, F (v) ∼ v −4 , we have chosen for drawing the plot. Note the slight tilt of the ellipses in Fig. 13.2. A consequence is that the result of a velocity-integration of the perturbed distribution critically depends on the integration window’s width and centering. For computing δn shown

13 Langmuir Wave-Electron Phase Correlation

321

in the mid-panel we used a 7% width on either side of vp . However, if the window is narrower and not centered on vp , the integration will emphasize the reactive component. Furthermore, the sign of the latter depends upon whether the window happens to be centered a little above or below vp . In Fig. 13.2 the parameter µ is small, µ = 0.09, hence the interaction is justiﬁably in the linear regime. However, when either the wave amplitude is larger or the packet is more extended, the linear solution to the Vlasov equation loses its validity and one must account for nonlinear corrections. These corrections to the orbits of resonant electrons δv and δx are diﬃcult to compute analytically. Instead, we resort to particle-in-cell simulations, where consistent interactions between ﬁeld and particles are automatically taken into account. The results of these simulations are shown in Fig. 13.3 which display the bunch-ellipses in a weakly nonlinear regime with µ = 1.2. Solid contours and shaded areas indicate a positive value, or an accumulation of electrons. Dotted contours indicate a negative value, or a dearth of electrons. Two points must be noted. First, the wave is here driven by a bump on the tail of the distribution function. Accordingly, resonant electrons are decelerated

Fig. 13.3. Bunch-ellipses in a weakly nonlinear regime, from a particle-in-cell simulation where the Langmuir packet is ampliﬁed by a bump on tail located at v = 10ve . Here µ = 1.2 and vp = 8.7ve . Accumulation locales (shaded in gray) have decelerated while dearth locales (dashed contours) accelerated. More in the text

322

C.A. Kletzing and L. Muschietti

by the wave and maxima of δn are in phase with ∂φ/∂x < 0. Second, since bunches of accumulated electrons are decelerated, the nonlinear correction to their orbit has δv < 0 and δx < 0. In contrast, bunches associated with a dearth of electrons are accelerated and thus have δv > 0 and δx > 0. Hence, the two types of bunch-ellipses are no longer aligned at v = vp as in Fig. 13.2. One can see in the plot that the solid contours have moved to the left while the dotted contours have moved to the right. In addition, their position in x drifts and slowly converges toward maxima of the potential, whereby their phase becomes more reactive.

13.3 Extended Wave Packet: A BGK Analysis The only nonlinear term in the Vlasov-Poisson system of equations is the acceleration term on the right hand side of (13.2). In Sect. 13.2, we chose to linearize this term by splitting the distribution in a large, homogeneous part F and a small, inhomogeneous part f , and then neglecting ∂f /∂v versus ∂F/∂v. We found that this was justiﬁed as long as µ 1. However, when the wave packet is so extended that a traversing electron can satisfy the resonance condition for a long time, this electron can be signiﬁcantly accelerated by the electric ﬁeld. Its orbit is then deeply altered. Therefore, the distribution function is strongly modiﬁed and the linearization procedure breaks down. The BGK method oﬀers another approach to solving the Vlasov-Poisson system of equations. Let us now imagine a very long wavepacket and examine the phase-space orbits of electrons in the so-called waveframe (frame moving with the wave phase velocity). These orbits are determined by the energy of the electrons ∂φ . (13.14) w = v 2 /2 − φ(x) where E(x) = − ∂x For this section, in order to simplify the notation, we introduce dimensionless units where length is normalized by λd , velocity is normalized by the electron ' thermal velocity ve = Te /m, and the electrostatic potential is normalized by Te /e. Electrons with w < 0 are trapped in local maxima of the wave potential. Electrons with w > 0 are untrapped and alternately accelerate and decelerate while passing over hill and dale of the potential. Now, any function of w where φ(x) is time-independent automatically satisﬁes Vlasov equation. The BGK approach exploits that property and assumes that the distributions are function of w only and that they are in a self-consistent steady state with the potential. This enables one to concentrate on solving the Poisson equation for a given model of wave ﬁeld, e.g. here a sinusoidal wave φ(x) = Ψ sin kx .

(13.15)

Let Fe (w) and Ft (w) be the distribution functions of, respectively, the passing and the trapped electrons. Poisson equation reads

13 Langmuir Wave-Electron Phase Correlation

d2 φ = −ni + dx2

∞

Ψ

Fe+ (w) + Fe− (w) √ dw + 2(w + φ)1/2

Ψ

−φ

√ 2Ft (w) dw . (w + φ)1/2

323

(13.16)

Terms on the right hand side represent, in order, the density of the ions, which are supposed to form a constant background, the density contribution from the passing electrons, and that from the trapped electrons. The passing electrons have been split into those moving to the right, Fe+ (w), and those moving to the left, Fe− (w). The trapped electrons, by contrast, must be symmetric with the same ﬂux of right and left moving particles, Ft+ (w) = Ft− (w) = Ft (w), in a stationary situation. 13.3.1 Passing Electrons A model of distribution for the ambient electrons that is practical for computational purposes and representative of the observed distributions is given by 1 2 Fe± (v) = , (13.17) π [1 + (v ± vp )2 ]2 where v is an absolute number measuring the velocity from the wave frame, and the direction is selected by the ± sign. Note that this distribution is normalized to unity and becomes a power-law in v −4 at large velocities. In the presence of the wave it translates into 2 Fe± (w) = (13.18) 2 2 with w > Ψ . √ π 1+ 2(w − ψ)1/2 ± vp After integrating this expression for the densities of right and left moving particles, one obtains the total density of passing electrons as

√ √ ϕ (1 + ϕ − vp2 ) 2 ϕ 2 1 np = 1 − − arctan (13.19) π vp4 + 2vp2 (1 − ϕ) + (1 + ϕ)2 π vp2 + 1 − ϕ where the notation ϕ ≡ 2(Ψ +φ) is introduced. The density is maximum where the potential is minimum, at ϕ = 0, and monotonically decreases toward a minimum where the potential is maximum, at ϕ = 4Ψ . We will assume that 4Ψ 1, which enables us to expand the complicated expression above into the simpler √ √ 16 2(5vp2 − 1) 4 2 1/2 (Ψ + φ) − (Ψ + φ)3/2 (13.20) np (φ) = 1 − π(vp2 + 1)2 3π(vp2 + 1)4 The most important term, in ϕ1/2 , has a coeﬃcient that is related to the value of the ambient distribution at the wave phase velocity ((13.17) with v = 0). This is in agreement with the intuitive'idea that electrons which move slowly along the separatrix (located at v = 2Ψ (1 + sin kx)) are strongly aﬀected by the potential and thus contribute the most to the density variations. By contrast, fast-moving electrons far from the separatrix hardly “notice” the potential and pass by quasi undisturbed.

324

C.A. Kletzing and L. Muschietti

13.3.2 Trapped Electrons We formally deﬁne the density of trapped electrons nt as an unknown function of φ through the expression Ψ √ 2Ft (w) dw . (13.21) nt (φ) ≡ (w + φ)1/2 −φ This integral equation can be inverted for Ft (w), which yields −w 1 d 1 nt (φ) dφ . Ft (w) = √ 2π −Ψ (−w − φ)1/2 dφ

(13.22)

The potential φ(x) is given by (13.15) and requires a net density perturbation ns (φ) = −k 2 φ with

ns = np + nt .

(13.23)

Substituting ns and np for nt in (13.22), one obtains after integration √ 4(5vp2 − 1) 2 2 − (Ψ − w)1/2 + (Ψ − w) (13.24) Ft (w) = 2 2 2 π(vp + 1) πvp π(vp2 + 1)4 with −Ψ ≤w 3.5 are preferred. The correlator was ﬂown on a rocket launched from Poker Flat, Alaska on March 4, 1988 which crossed several auroral arcs during the expansion phase of substorm. During the ﬂight, the parallel component of the electric ﬁeld reached amplitudes of 100 mV/m in the frequency band from 200 kHz to 5 MHz with a dominant frequency of 1.4 MHz indicating the presence of strong Langmuir oscillations. The output of the correlator showed several events with signiﬁcant correlation with electrons from the eight lowest energy channels which covered a range of energies from 380 eV to 3.2 keV. The correlations appeared ﬁrst in the higher energy channels and then moved to lower energy. Analysis of the data from this rocket ﬂight yielded ﬁve events with ﬂuctuations greater than 4σ. These ﬁve events, plotted as function of wave electric potential, are shown in Fig. 13.6 taken from Ergun et al. [9]. Each event is plotted with the associated error bar as determined from experimental uncertainties as well as the 1σ statistical error of the counts. All ﬁve events are located near 90◦ or 270◦ with respect to the wave electric

328

C.A. Kletzing and L. Muschietti

Fig. 13.6. Phase of bunched electrons for ﬁve events of correlations at levels greater than 4σ from Ergun et al. [9]. The phase of each event is plotted along with error bars derived from experimental uncertainties (Reprinted with permission from American Geophysical Union)

potential. This corresponds to 0◦ or 180◦ with respect to the wave electric ﬁeld. The frequency of all 5 events was 1.4 MHz which, when combined with the electron energies of 380 eV to 3.2 keV, yields wavelengths of 8 m to 20 m. Events 2 and 5 were such that more electrons were being accelerated by the wave than were being decelerated and thus corresponds to wave damping. They occurred during periods of weaker amplitudes of 10–20 mV/m. Events 1, 3, and 4 had more electrons decelerating than accelerating corresponding to wave growth. At these times the wave amplitudes were near 100 mV/m or greater. Following the analysis of Sect. 13.2, the value of µ can be calculated using the observed electric ﬁeld amplitude of 80 mV/m, the inferred wavelength λ = 12.8 m, and observed the plasma frequency of 1.4 MHz along with a value for kL. For these events, kL was estimated to be of the order of kL = 30 which yields a value of µ = 0.09. This is consistent with the linear analysis for a short wave packet which predicts the observation of the resistive component of the electron perturbation because in this regime, the resistive perturbation is generally larger than the reactive perturbation as illustrated in Fig. 13.2. 13.4.2 Measurements of the Reactive Component Further improvement in the measurement of the phase of electron bunching in Langmuir waves in the auroral zone was achieved using a new phase correlator developed at the University of Iowa (UI) in the late 1990’s. Figure 13.7 shows

13 Langmuir Wave-Electron Phase Correlation

329

VC O Mas ter C lock Divide by 16 C lock

Kept in phas e by P LL

Input Waveform

P has e B in at 1.6 Mhz

8/9

4/5

0/1

12/13

Fig. 13.7. The UI wave-particle correlator uses a phase-locked loop (PLL) locked to the measured waveform with a clock derived from master voltage-controlled oscillator (VCO) running at 16 times the frequency. This master clock subdivides the wave in 16 phase bins. Electron counts are sorted into the bins as they arrive to produce a map of phase bunching (Reprinted with permission from American Geophysical Union)

the principle of operation. The correlator used a voltage-controlled oscillator (VCO) running at 16 times the expected frequencies for Langmuir waves. The VCO clock signal was divided by 16 to provide a signal which was then aligned to the measured AC parallel electric ﬁeld through the use of a phase-locked loop (PLL). The PLL output a signal which indicated when the loop was properly locked and also the frequency of the locked signal which is then compared with the analog waveform data to ensure that the PLL was locked to signals of interest. Because the VCO master clock runs at 16 times the frequency of the wave to which the PLL is locked, it provides a highly accurate means of sub-dividing the measured wave into 16 phase bins. As electrons are counted by the detectors, they are sorted into the appropriate phase bin associated with their arrival time. Calibration of the detector, wave, and correlator electronics veriﬁed the accuracy of the phase bins and determined the timing delays through the system. The timing delays cause the absolute bin number which corresponds to 0◦ phase shift (as well as other phase angles) to shift as a function of input wave frequency. For example, at 1.6 MHz, 0◦ phase shift between the wave and the electrons occurs between correlator bins 8 and 9. The bottom of Fig. 13.7 shows the calibrated phase angles of the bins at a frequency of 1.6 MHz. The UI correlator was ﬂown on a rocket ﬂight which was launched from Poker Flat, Alaska in February 6, 2002. Strong Langmuir waves were observed as the rocket traversed an auroral form near the poleward boundary of the auroral precipitation. The Langmuir waves were associated with a burst of ﬁeld-aligned electrons at energies below 1 keV and well below the inverted-V

C.A. Kletzing and L. Muschietti

σ

330

Fig. 13.8. Two examples of signiﬁcant correlation of Langmuir waves with electrons at 468 eV. For each of the 16 phase bins, the count level is shown in terms of standard deviations away from the average number of counts that would be expected for a bin for the total number of counts received (Reprinted with permission from American Geophysical Union)

peak energy of 5 keV. During this period of Langmuir emission, two intervals of stronger emission with amplitudes of 60–200 mV/m were observed. Each of these intervals were of the order of 50 ms in duration. During both intervals, the ﬁeld-aligned wave power was 8.5–11 times that of the perpendicular power indicating ﬁeld-aligned waves. Because the payload telemetered continuous waveform data, it was possible to determine that the Langmuir waves were remarkably monochromatic with little or no variation in frequency. During each of the two larger amplitude bursts, the amplitudes varied slowly over hundreds of wave periods. During each of the two bursts of strong Langmuir waves, signiﬁcant waveparticle correlation was found as shown in Fig. 13.8. Each set of panels shows four consecutive sets of correlator measurements of electrons with energy of 468 eV and which were sampled every millisecond. Data are plotted as standard deviations away from the expected average count rate in each phase bin. To be regarded as signiﬁcant, we require a correlation level of at least 3.5σ. Although there is frequent variation of the order of 2σ shown in Fig. 13.8, these are not considered signiﬁcant because the probability of 2σ variation is once in 22 samples as discussed above. With this in mind, in the left set of panels, the ﬁrst and third panels show notable correlation. The ﬁrst panel shows a weakly signiﬁcant correlation with three adjacent phase bins with ∼2σ levels of correlation corresponding to a 3.3σ variation.

13 Langmuir Wave-Electron Phase Correlation

331

However, the third panel from the top has two consecutive phase bins with phase bunching above 3σ. The combined probability of two such bins occurring together exceeds that of a single 4σ event. The bins in which the phase bunching occurs are those at 90◦ with respect to the wave electric ﬁeld (positive away from the ionosphere) for the wave frequency of 1.6 MHz. The second example shown in the right set of panels in Fig. 13.8 shows a second event with similar high levels of signiﬁcance in the top three panels. Initially, a single channel with more than 4σ signiﬁcance occurs at 90◦ . In the second panel, this feature broadens and shifts to somewhat larger phase shift with four phase bins above 2σ representing a combined signiﬁcance of more than 4σ. The calculated probability of random occurrence of the counts represented by the four phase bins centered around 90◦ is one in 932,068 suggesting that this is a highly signiﬁcant correlation. The third panel shows further advance to larger phase angles, with most signiﬁcant set of signals now showing more than -4σ (5 channels at −2σ each) between 270◦ and 0◦ . Figure 13.9 shows three sequential distribution function plots from immediately before, during and immediately after the ﬁrst correlation event. The times given above each plot correspond to the center time of each 40 ms energy sweep. As can be seen, at the time of the correlation, a small, downward electron beam parallel to the magnetic ﬁeld is measured at the same energy as the correlated electrons. This is indicated by a small arrow at the velocity corresponding to 468 eV. The isolated beam is not present in the distribution function measured before the correlation or in the distribution measured after the correlation. Although the time resolution of these measurements is much lower than for the correlator, this suggests that the correlation arises from this beam. The brief increase in electron phase space density at the energy of correlated electrons illustrated in Fig. 13.9 was also seen for the second correlation event. In both cases the preceding and following energy sweeps did not show this feature, suggesting that the resonant electrons were only brieﬂy in the detectable energy range. From the energy of the correlated electrons (468 eV) and the frequency of the waves (1.6 MHz), we can derive the wavelength of the Langmuir waves as 8.2 m, similar to the results of the UC Berkeley measurements. The waveform data makes it possible to estimate the packet parameters to aid in the interpretation of correlation events. Figure 13.10 shows 15 ms of data which includes the interval of the second set of correlations. The interval during which the correlations were observed is indicated by a heavy line above the waveform data. As can be seen, the correlations occurred during the largest wave amplitudes. If we assume a background thermal energy of 0.2 eV and use the inferred wavelength of 8.2 m along with the observed Langmuir frequency of 1.6 MHz, then the group velocity for Langmuir waves, vg =

3v 2 k ∂ω 3v 2 th = th , ∂k ωpe λfpe

(13.25)

332

C.A. Kletzing and L. Muschietti

Fig. 13.9. Sequential distribution functions from before, during, and after the correlation event shown in the left panel of Fig. 13.8. The velocity corresponding to the correlated electrons is indicated with an arrow in the middle panel and shows that a ﬁeld-aligned beam is present at this time (Reprinted with permission from American Geophysical Union)

gives the value of vg = 8.02 km/s = 8 m/ms. If the wave is traveling down the ﬁeld line, then the amount of the wave packet above the correlation observations extends from the time of the correlations in Fig. 13.10 to the end of the packet some 3 ms later. This yields a length of the wavepacket above rocket payload of 24.1 m at the time of the correlations and corresponds to a value of kL = 18.5. A rough average for the amplitude of the electric ﬁeld during this interval is 130 mV/m. Combining this with the estimate of kL gives a value of µ = 0.12, consistent with the linear theory and similar to the value used in Fig. 13.2. It should be pointed out, however, that the usual limitations of single point spacecraft measurements apply to this interpretation. With the data at hand, we cannot rule out a scenario in which the electron beam has suddenly appeared and the wave packet has grown over time at the spacecraft location. Indeed, such a scenario would be consistent with the quasi-exponential increase in amplitude of the packet in the 10 ms preceding the correlation observations. In this case, there is no way to ascertain the amount of wave packet above the payload, but the shape of the packet suggests that the correlations were observed shortly after the linear growth phase ended and some type of non-linear saturation began to operate. The group velocity calculation still applies, but now becomes a lower bound on the length of the packet. By observing the packet for 3 ms we know that it extended at least 24 m above the rocket, but it may have extended much further, and then disappeared due to temporal eﬀects in the driving electron distribution. A third alternative is that the payload may have moved into and then out of a pre-existing region

13 Langmuir Wave-Electron Phase Correlation

333

Fig. 13.10. Langmuir waveform data which includes the interval of the second set of correlation events. The correlation interval is indicated by the solid bar above the waveform

of Langmuir waves such that the observed wave envelope is determined by the spatial structure of the Langmuir waves along the rocket track. However, given the rocket velocity of roughly 1 km/s, this latter scenario would require a rather small wave packet and does not seem likely.

13.5 Discussion The electron correlator observations reported by the UC Berkeley were predominantly resistive, that is, in-phase or 180◦ out-of-phase with the wave electric ﬁeld. Although they had expected to see the reactive component, the observations of the resistive component prompted the analysis developed in Sect. 13.2 and led to the conclusion that they had observed electrons which were bunched by relatively short wave packets with kL of the order of 50–100. This result was also consistent with the value for µ that was determined which suggested that thee observed waves were in the linear regime. As shown in the middle panel of Fig. 13.2, for µ = 0.09, the perturbation to the density and hence, to the electron distribution, is dominated by the resistive component for small values of µ. Although the reactive component is present, as shown in Fig. 13.1, its bipolar character requires relatively narrow response electron detectors so as not to average the positive and negative perturbation together, yielding no perturbation. The detectors which made these resistive correlator measurements had ∆v/v 16% and as shown in

334

C.A. Kletzing and L. Muschietti

Fig. 13.1, this is broader than the expected perturbation relative to the phase velocity. The electron phase bunching observed by the UI group was predominantly 90◦ out-of-phase with the electric ﬁeld, suggesting a trapped population of electrons. The wave ﬁeld in which the phase-bunched electrons were observed was long-lived in terms of wave periods, monochromatic, and showed slow modulation of the wave envelope. As discussed above, using the assumption that the wave packet moved past the payload due the group velocity of the waves yields a low value of µ which would suggest that the linear analysis also applies to this case. In examining Fig. 13.2, it is seen that although the overall perturbation to the density has a greater resistive component than reactive component, the reactive component does exist. On closer examination of Fig. 13.2, however, it can be seen that for a narrow range of velocities of the order of ∆v/v = 5%, on either side of the phase velocity, the reactive component is more dominant. This is seen in the shifting of the perturbed distribution toward being in phase with the potential for velocities above the phase velocity and a shift toward 180◦ out-of-phase for velocities below the phase velocity. Because the UI electron detectors had a comparably narrow velocity response ∆v/v 5%, they are capable of capturing one or the other side of the perturbation and could measure a reactive perturbation even for a short wave packet. In the case that the envelope of the observed waves is predominantly due to temporal evolution, then it is likely that the wave packet extends a significant distance above the rocket payload. This is suggested by the fact that the electron distribution will be unstable over a wide range of altitudes and would be expected to grow waves over a region of many wavelengths in extent. The character of the wave envelope also suggests a transition away from the linear stage of growth. Taken together, these two arguments suggest that an alternative explanation is that the correlations indicate trapping and thus the BGK analysis is appropriate. As shown in the right hand panel of Fig. 13.3, when the electron-Langmuir wave interaction becomes nonlinear, the density perturbation shifts toward a purely reactive phase as the electrons become trapped. Indeed the bunching of the electrons observed by the UI group is such that the positive perturbation was that of trapped electrons. To resolve this ambiguity will require future experiments which provide adjacent energy channels with narrow response so that the full character of the perturbed distribution function can be revealed. This would allow one to see if both reactive pieces, that is the positive perturbation below the phase velocity and the negative perturbation above the phase velocity shown in Fig. 13.1, are observed side-by-side as the linear model would suggest or if only a single reactive perturbation is observed, consistent with the BGK analysis.

13 Langmuir Wave-Electron Phase Correlation

335

13.6 Conclusions Multiple spacecraft observations have conﬁrmed the ubiquitous nature of Langmuir waves in the presence of auroral electrons. Early observations have shown clear evidence that the electrons show variations consistent with bunching at or near the Langmuir frequency. Linear analysis of the interaction of a ﬁnite Gaussian packet of Langmuir waves shows that there are two components to the perturbation to the electron distribution function, one in-phase (or 180◦ out-of-phase) with respect to the electric ﬁeld called the resistive component and one which is 90◦ (or 270◦ ) out-of-phase with respect to the electric ﬁeld. For small wave packets, the resistive perturbation dominates. For longer wave packets, a non-linear analysis is appropriate which suggests that the electrons have interacted long enough to become trapped and the reactive phase becomes dominant. Rocket observations of the phase bunching of the electrons using waveparticle correlators have measured both components. The UI observations [Kletzing et al., 16] diﬀer from those of the UC Berkeley observations [Ergun et al., 7, 9] in that a purely reactive phase bunching was observed as compared to a predominantly resistive perturbation. The resistive phase results of the UC Berkeley group were interpreted as arising from a short wave packet. The UI observations of the reactive phase can be explained by either a long, coherent train of Langmuir waves or that the narrower velocity response of the UI detectors made it possible to capture only one side of the reactive component of the perturbed distribution function for a short wave packet in the linear regime. Future wave-particle correlator experiments should be able to resolve these questions by providing more examples with better velocity space coverage.

References [1] Bale, S.D.: Observation of the topside ionospheric mf/hf radio emission from space, Geophys. Res. Lett. 26, 667, 1999. [2] Bauer, S.J. and R.G. Stone: Satellite observations of radio noise in the magnetosphere, Nature 218, 1145, 1968. [3] Beghin, C., J.L. Rauch, and J.M. Bosqued: Electrostatic plasma waves and hf auroral hiss generated at low altitude, J. Geophys. Res. 94, 1359, 1989. [4] Boehm, M.H.: Waves and static electric ﬁelds in the auroral acceleration region, PhD thesis, University of California, Berkeley, 1987. [5] Boehm, M.H., G. Paschmann, J. Clemmons, H. H¨ ofner, R. Frenzel, M. Ertl, G. Haerendel, P. Hill, H. Lauche, L. Eliasson, and R. Lundin: The tesp electron spectrometer and correlator (F7) on Freja, Space Sci. Rev. 70, 509, 1995. [6] Bonnell, J.W., P.M. Kintner, J.E. Wahlund, and J.A. Holtet: Modulated langmuir waves: observations from freja and scifer, J. Geophys. Res. 102, 17233, 1997.

336

C.A. Kletzing and L. Muschietti

[7] Ergun, R.E., C.W. Carlson, and J.P. McFadden: Wave-particle correlator instrument design, In: R.F. Pfaﬀ, J.E. Borovsky, and D.T. Young (Eds.), Measurement Techniques in Space Plasmas: Particles, volume 102 of AGU Geophys. Monogr. Ser., p. 4325, AGU, Washington, D.C., 1998. [8] Ergun, R.E., C.W. Carlson, J.P. McFadden, J.H. Clemmons, and M.H. Boehm: Evidence of a transverse modulational instability in a space plasma, Geophys. Res. Lett. 18, 1177, 1991. [9] Ergun, R.E., C.W. Carlson, J.P. McFadden, D.M. TonThat, J.H. Clemmons, and M.H. Boehm: Observation of electron bunching during Landau growth and damping, J. Geophys. Res. 96, 11371, 1991. [10] Gough, M.P., P.J. Christiansen, and K. Wilhelm: Auroral beam-plasma interactions: particle correlator investigations, J. Geophys. Res. 90, 12287, 1990. [11] Gough, M.P. and A. Urban: Auroral beam/plasma interaction observed directly, Plan. Space Sci. 31, 875, 1983. [12] James, H.G., E.L. Hagg, and L.P. Strange: Narrowband radio noise in the topside ionosphere, AGARD Conf. Proc., AGARD-CP-138, 24–1–24–7, 1974. [13] Kelley, M.C. and G.D. Earle: Upper hybrid and Langmuir turbulence in the auroral e-region, J. Geophys. Res. 93, 1993, 1988. [14] Kellogg, P.J. and S.J. Monson: Radio emissions from the aurora, Geophys. Res. Lett. 6, 297, 1979. [15] Kintner, P.M., J. Bonnell, S. Powell, and J.E. Wahlund: First results from the freja hf snapshot receiver, Geophys. Res. Lett. 22, 287, 1995. [16] Kletzing, C.A., S.R. Bounds, J. LaBelle, and M. Samara: Observation of the reactive component of langmuir wave phase-bunched electrons, Geophys. Res. Lett., 32, L05106, doi:10.1029/2004GL021175, 2005. [17] McAdams, K.L. and J. LaBelle: Narrowband structure in hf waves above the electron plasma frequency in the auroral ionosphere, Geophys. Res. Lett. 26, 1825, 1999. [18] McAdams, K.L., J. LaBelle, M.L. Trimpi, P.M. Kintner, and R.A. Arnoldy: Rocket observations of banded stucture in waves near the langmuir frequency in the auroral ionosphere, J. Geophys. Res. 104, 28109, 1999. [19] McFadden, J.P., C.W. Carlson, and M.H. Boehm: High-frequency waves generated by auroral electrons, J. Geophys. Res. 91, 12079, 1986. [20] Muschietti, L., I. Roth, and R. Ergun: Interaction of Langmuir wave packets with streaming electrons: phase-correlation aspects, Phys. Plasmas 1, 1008, 1994. [21] Newman, D.L., M.V. Goldman, and R.E. Ergun: Langmuir turbulence in the auroral zone 2. nonlinear theory and simulations: J. Geophys. Res. 99, 6377, 1994. [22] Nicholson, D.R.: Introduction to Plasma Theory, Wiley, New York, 1983. [23] Reiner, M.J. and M.L. Kaiser: Complex type iii-like radio emissions observed from 1 to 14 mhz, Geophys. Res. Lett. 26, 397, 1999. [24] Reiner, M.J., M. Karlicky, K. Jiricka, H. Aurass, G. Mann, and M.L. Kaiser: On the solar origin of complex type iii-like radio bursts observed at and below 1 mhz, Astrophys. J. 530, 1049, 2000. [25] Samara, M., J. LaBelle, C. A. Kletzing, and S.R. Bounds: Rocket observations of structured upper hybrid waves at fuh = 2fce , Geophys. Res. Lett., submitted, 2004.

13 Langmuir Wave-Electron Phase Correlation

337

[26] Sanbonmatsu, K.Y., I. Doxas, M.V. Goldman, and D.L. Newman: NonMarkovian electron diﬀusion in the auroral ionosphere at high Langmuir-wave intensities, Geophys. Res. Lett. 24, 807, 1997. [27] Spiger, R.J., J.S. Murphree, H.R. Anderson, and R.F. Loewenstein: Modulation of auroral electron ﬂuxes in the frequency range 50 kHz to 10 MHz, J. Geophys. Res. 81, 1269, 1976. [28] Stasiewicz, K., B. Holback, V. Krasnoselskikh, M. Boehm, R. Bostr¨ om, and P.M. Kintner: Parametric instabilities of langmuir waves observed by freja, J. Geophys. Res. 101, 21515, 1996. [29] Walsh, D., F.T. Haddock, and H.F. Schulte: Cosmic radio intensities at 1.225 and 2.0 mc measured up to and altitude of 1700 km, Space Res. 4, 935, 1964. [30] Weatherwax, A.T., J. LaBelle, M.L. Trimpi, and R. Brittain: Ground-based observations of radio emissions near 2fce and 3fce in the auroral zone, Geophys. Res. Lett. 20, 1447, 1993.

Index

absorption D-region 163 negative 65 resonant 246 acceleration 205, 253 ‘inverted-V’ 107, 129 auroral electron 205 in electric ﬁeld 64 in lower hybrid waves 258 primary region 110 regions of 64 transverse 286 upward ions 111 active sounding 30 adiabatic motion 64 AKR 106, 129, 205, 298, 305 source region 129 bounded source 66 conditions for generation 56 elementary radiation events (ERE’s) 132 ﬁne structure 132 narrow band 132 relation to continuum 43 source extension 59 source life time 59 source region properties 63 statistical properties 58 Alfv´en waves 110 kinetic 116, 119 shear 116, 205 ampliﬁcation path 72 angle of propagation 71

antenna distributed dipoles 192 eﬀective length 199, 201 electric dipole 299 multiple 298 orientation 206 Appleton-Hartree equation 147 atmosphere 142 aurora auroral hiss 60, 68, 156, 177, 191, 201, 212 power ﬂux 204 Auroral Kilometric Radiation (AKR) 55, 205, 212 auroral oval 106, 205 auroral regions 107, 298 acceleration region 107 downward current 107 upward current 107 auroral roar 212 auroral wave emission 314 background plasma 114 black aurora 112 dayside 106 electric ﬁeld 205 radar 142 visible 142 beaming angle 41 Bernstein modes 9, 24, 41 BGK modes 121 Boltzmann equation 90 brightness temperature 55

340

Index

bulk ﬂows

114

cavities auroral 178 cavity width 286 ionospheric 66 observation of 68 plasmaspheric 44 refractive index 27 small scale 60, 214 spherical 274 Cherenkov radiation 29, 156, 202, 212 chord length 274 chorus 175 Cluster spacecraft 298 CMA diagram 6, 192 coherence length 91, 205 collapse lower hybrid 274 time 274, 289 collision frequency anomalous 258 comets 252 condensation of electrons at low v 65 convection 133 correlation auroral hiss-HF backscatter 164 correlation function 88 multi-point function 88 whistler-HF backscatter 164 correlator wave particle 214, 313 cosmic rays 106 secondaries 106 showers 106 current bifurcation 136 ﬁeld-aligned 107, 134 Hall 135 Pedersen 110 sheet 251, 252, 259, 314 tail current sheet 106 thin tail sheet 135 cutoﬀ 235 high frequency 28 low frequency 129 upper frequency 141 X-mode 71

density depletion 274, 278 depth of 292 diﬀusion 253 dispersion maximum growth rate 79 of plasma waves 5 relation ﬁnite geometry 78 distribution beam 129, 213, 252 auroral electron 110 electron conic 255 electron shell 255 hole 63 horseshoe 64, 128, 130, 212, 298 loss cone 63, 128, 213 Maxwellian 110 nonthermal 213 passing 125 perturbation of 313 ring 128 shell 298 trapped 64, 125 two electron 116 wave 300, 308 DNT resonance 19 “donkey ear—hyperpage 45 double layer 115, 125, 127, 255 electrostatic shock 116 ramp 128 shear ﬂow driven 120 strong 116 dual payloads 16 echo delay times 10 echo train 156 R-X mode echo 180 whistler echoes 28, 158, 180 Z mode echo 180 Z-mode, X-mode 14, 22 electric ﬁeld ﬁeld-aligned 112 reconnection 255 electrons anisotropic 41 auroral 129 conics 255 density scale height 212

Index electron beams 255 hard spectrum 98 inertial length 205 magnetized 133 oscillations, cylindrical 20 passing 323 precipitating 156 reactive bunching 313, 328 relativistic 129 resistive bunching 313, 325 soft spectrum 99 sounder-accelerated 17 trapped 324 trapped magnetospheric 156 wave packet perturbation 317 Ellis radio window 217 energy conversion 251 redistribution 253 transport 251 ﬂuctuations density 298 electromagnetic 298 small scale 91 Freja spacecraft 298 frequency electron cyclotron 38, 129, 141 lower hybrid 178 plasma 128, 141, 213, 252, 254 upper hybrid 28, 213 frequency gap 68 Gendrin angle 150, 168 Geos Spacecraft 298 geospace 298 harmonics electron cyclotron height apparent 4 heliosphere 252

212

impact parameter 275, 278 impedance 19 incoherence 205 index of refraction 15 refractive surfaces 17

341

surface of 150 inhomogeneity 235 ﬁeld-aligned 15 random 88 transverse 246 instability beam-beam 125 Buneman 126 current 127 gyro-resonant 156 modulational 289 two stream, modiﬁed 256 two-stream 125, 126 kinetic 128 whistler mode 156 interaction beam-plasma 202, 313 coherent 202 interacting wave bands 132 Langmuir-electron 315 nonlinear wave 315 resonant 141 wave-particle 141, 212, 253 Interball spacecraft 298 interferometry 299 invariant embedding method 240 ﬁrst adiabatic 64 ion gyro radius 252 ion inertial length 252 ion-acoustic speed 116 ionogram 4 ionosonde 4 ionosphere air-ionosphere boundary 153 lower edge 153 ionospheric cavity, Calvert 65 ionospheric reﬂection 4 irregularities density, ﬁeld aligned 5, 142 FAI 142 large scale 147, 156 small scale 147, 154 transverse scale 142 large-scale 91 small scale 19, 91 irregularities, ﬁeld aligned 41 Korteweg-de Vries equation

117, 121

342

Index

laboratory experiments 252 Landau damping 205, 313 lightning 141, 151, 161 linear instability 56 linear process 129 magnetic ﬂux tube 255 magnetized planets 252 magnetosphere 141 cusp 106 magnetopause 40, 106, 252 magnetotail 252 plasmasheet boundary layer 259 tail 106 current sheet 106 tail acceleration region 133 maser electron cyclotron 56, 129, 212, 298 ﬁnite geometry 74 ﬁnite geometry eﬀect 65 measurements burst mode 307 multi-component 298 mode Bernstein-Green-Kruskal (BGK) 121 connection 68 conditions for 73 through cavity boundaries 68 conversion 41, 81, 82, 143, 165, 212, 235 conversion radiation 212 coupling 22, 56, 235 crossover region 235 eigenmodes 212 electrostatic 212 energy leakage 79 extraordinary (X) 4 extraordinary (X,Z) 77, 130 free space 4 growth rate 79 identiﬁcation 70 identiﬁcation of 69 internally unstable 68 mode equation extraordinary (X,Z) 67 ordinary (O) 67 ordinary (O) 4, 41 quasi-electrostatic 17, 149, 193

refraction inside source 72 transmission coeﬃcients upper hybrid 31 W-mode 141 whistler 141, 191 Z-mode ducting 15

81

Nonlinear Schr¨ odinger equation nonthermal mechanism 55

121

Ohm’s law 114 generalized 114, 255 particles auroral energy ﬂux 107 nonthermal 87 pitch angle scattering 165 relativistic 94 ring current protons 165 solar energetic 106 trajectory stochastic 89 phase space ﬂow 64 phase space holes electron holes 205 phase-space holes 116, 120 BGK modes 121 bipolar electric ﬁelds 122 electron holes 121 ion holes 121, 122 observations of 122 simulations 123 tripolar electric ﬁelds 122, 126 width 124 plasma blobs 275 collisionless 251 density 206 inversion method 27 dielectric tensor 88 ionospheric 144 overdense 214 plasmagram 22, 24, 180 plasmapause 22, 37, 145, 314 plasmasphere 144 diﬀusive equilibrium 145 plasmaspheric hiss 166, 173, 176 plasmaspheric notch 45

Index plasmons 121 Poeverlein construction 8, 153 Poisson distribution 277 polar region low altitude 28 Polar spacecraft 298 polarization 235 left hand L 6 linear 195 right hand R 6 ponderomotive force 114 population inversion 56 potential barrier 126 ﬁeld-aligned drop 111 S-shaped 116 U-shaped 116 Poynting ﬂux 300 pressure anisotropy 114 gradient 114 process coherent 115 propagation angle 12 anisotropic 148 dispersive 148 ducted 12, 154 Earth-ionosphere wave guide 158 guiding 142 horizontal 12 inter-hemispheric 22 inter-satellite experiments 208 non-ducted 156 oblique 131 parallel 6, 131 perpendicular 6, 131 ray bending 163 reﬂection 142 refraction 142 scattering 142 stratiﬁed medium 12 transmission cone 153 vertical 12 waveguide duct coupling 171 proton spurs 20 pseudo-potential 116 radar

4

aurora 142 radiation 252 bremsstrahlung coherent 102 incoherent 102 coherent 56, 103, 130 continuum enhancement 40 escaping 40 kilometric 37, 42 non-thermal (NTC) 37 terrestrial 212, 314 trapped 40 dipole 19 direct 212 ﬁne structure 212 impulsive 151 incoherent 205 incoherent Cherenkov 17 incoherent cyclotron 17 indirect 212 isotropic 72 narrow-band 37 radiating diagram 71 solar radio bursts 101 sources location 38 narrow 82 regions 213 synchrotron 41, 100 diﬀusive 100 transition radiation 102 VLF 122, 128, 151 wave guide theory 130 whistler chorus 156 discrete equatorial 156 midlatitude hiss 156 periodic 156 quasi-periodic 156 radiation belts 142, 145 Radio Plasma Imager 157 radio wave cutoﬀ 6 emission from cavities 81 fading 142 interferometry 299 interplanetary bursts 252 pulse 4

343

344

Index

solar bursts 252, 314 trans-ionospheric 142 ray optics 206 ray path 16 ray tracing 169, 299 receivers plasma wave 142 reciprocity principle of 191 reconnection 106, 252 astrophysical implications 264 asymmetric 252 current closure 134 diﬀusion 133 fast 255 ﬂux transfer events, FTE 258 Hall current 133 ion dynamics 133 outﬂow region 253 radiation from 256, 264 reconnection site 133, 255 separatrix 253 small scale 136 tailward 106, 133 three-dimensional 136 whistler 255 whistlers 261 X-line 253 reﬂection from hemispheres 22 conjugate 22 local 22 refractive index 5, 148 cavity 24 relativistic eﬀects 56, 65 resistance anomalous 115, 258 resolution limitations 252 resonance 6, 235 cyclotron 214 DNT 19 group resonance 206 higher nfce 20 inverse 258 Landau 142 lower hybrid 150, 157 lower oblique 191 resonance cone 150, 197, 205

sounder-stimulated 31 upper oblique 7, 191, 207 resonance condition 56 resonator 298 Sagdeev potential 116 separatrix region 253 shock 251 simulations, numerical 252 skin depth 135 ion 135 solar wind 252 solitary waves (ESW) 255 soliton radiation 56 solitons 259 sounding frequency 4 sounding technique 4 space weather 142 spectrum bandwidth 164 dynamic 68 frequency-time spectrogram of inhomogeneity 102 spectral broadening 164 spectral shape 89 stratiﬁed medium 8 horizontally 152 structure formation 252 substorm 55, 106, 129, 133 activity 142 temporal sampling 282 time evolution deterministic 274 top-hat 278 topside sounder 10, 217 transfer equation 81 transmitters VLF 141 transport coeﬃcients 253 trapped electrons 65 Tsyganenko model 23 turbulence 205, 251 energy density 115 magnetic 91 two point study 31 upper hybrid

131

Index band 31 frequency 31, 38 mode 31 resonance 38, 246 W-mode 28 sources 159 waves Alfv´en 110 ampliﬁcation 298 BEN 259 Bernstein 41 broadband emission 131 chirps 216 coherent Langmuir train 313 electron cyclotron 256 harmonics 38, 216, 263 electrostatic 38 electrostatic solitary (ESW) 254 extended packet 322 ﬁnite size packet 316 Gaussian packet 313, 318 incoherent 115 ion acoustic 116, 122 ion cyclotron 116 Landau damping 149

345

Langmuir 214, 254, 256, 259, 313 lower hybrid 149, 164, 256, 274, 277 damping 285 lower hybrid-drift 253 multiplets 214 plane wave analysis 303 polarization 259 saucer emissions 122 shock wave 251 soliton 116, 124 upper hybrid 216, 254, 314 VLF whistlers 125 wave bands 214 wave form 214 wave normal 148 normal angle 153 wave packet 150, 313 whistler 123, 214, 254 magnetospheric whistlers 156 multi-component 156 Z-inﬁnity 7 Z-mode 5, 41 ducting 15 fast 18 slow 16

Lecture Notes in Physics For information about earlier volumes please contact your bookseller or Springer LNP Online archive: springerlink.com

Vol.640: M. Karttunen, I. Vattulainen, A. Lukkarinen (Eds.), Novel Methods in Soft Matter Simulations Vol.641: A. Lalazissis, P. Ring, D. Vretenar (Eds.), Extended Density Functionals in Nuclear Structure Physics Vol.642: W. Hergert, A. Ernst, M. Däne (Eds.), Computational Materials Science Vol.643: F. Strocchi, Symmetry Breaking Vol.644: B. Grammaticos, Y. Kosmann-Schwarzbach, T. Tamizhmani (Eds.) Discrete Integrable Systems Vol.645: U. Schollwöck, J. Richter, D. J. J. Farnell, R. F. Bishop (Eds.), Quantum Magnetism Vol.646: N. Bretón, J. L. Cervantes-Cota, M. Salgado (Eds.), The Early Universe and Observational Cosmology Vol.647: D. Blaschke, M. A. Ivanov, T. Mannel (Eds.), Heavy Quark Physics Vol.648: S. G. Karshenboim, E. Peik (Eds.), Astrophysics, Clocks and Fundamental Constants Vol.649: M. Paris, J. Rehacek (Eds.), Quantum State Estimation Vol.650: E. Ben-Naim, H. Frauenfelder, Z. Toroczkai (Eds.), Complex Networks Vol.651: J. S. Al-Khalili, E. Roeckl (Eds.), The Euroschool Lectures of Physics with Exotic Beams, Vol.I Vol.652: J. Arias, M. Lozano (Eds.), Exotic Nuclear Physics Vol.653: E. Papantonoupoulos (Ed.), The Physics of the Early Universe Vol.654: G. Cassinelli, A. Levrero, E. de Vito, P. J. Lahti (Eds.), Theory and Appplication to the Galileo Group Vol.655: M. Shillor, M. Sofonea, J. J. Telega, Models and Analysis of Quasistatic Contact Vol.656: K. Scherer, H. Fichtner, B. Heber, U. Mall (Eds.), Space Weather Vol.657: J. Gemmer, M. Michel, G. Mahler (Eds.), Quantum Thermodynamics Vol.658: K. Busch, A. Powell, C. Röthig, G. Schön, J. Weissmüller (Eds.), Functional Nanostructures Vol.659: E. Bick, F. D. Steffen (Eds.), Topology and Geometry in Physics Vol.660: A. N. Gorban, I. V. Karlin, Invariant Manifolds for Physical and Chemical Kinetics Vol.661: N. Akhmediev, A. Ankiewicz (Eds.) Dissipative Solitons Vol.662: U. Carow-Watamura, Y. Maeda, S. Watamura (Eds.), Quantum Field Theory and Noncommutative Geometry

Vol.663: A. Kalloniatis, D. Leinweber, A. Williams (Eds.), Lattice Hadron Physics Vol.664: R. Wielebinski, R. Beck (Eds.), Cosmic Magnetic Fields Vol.665: V. Martinez (Ed.), Data Analysis in Cosmology Vol.666: D. Britz, Digital Simulation in Electrochemistry Vol.667: W. D. Heiss (Ed.), Quantum Dots: a Doorway to Nanoscale Physics Vol.668: H. Ocampo, S. Paycha, A. Vargas (Eds.), Geometric and Topological Methods for Quantum Field Theory Vol.669: G. Amelino-Camelia, J. Kowalski-Glikman (Eds.), Planck Scale Effects in Astrophysics and Cosmology Vol.670: A. Dinklage, G. Marx, T. Klinger, L. Schweikhard (Eds.), Plasma Physics Vol.671: J.-R. Chazottes, B. Fernandez (Eds.), Dynamics of Coupled Map Lattices and of Related Spatially Extended Systems Vol.672: R. Kh. Zeytounian, Topics in Hyposonic Flow Theory Vol.673: C. Bona, C. Palenzula-Luque, Elements of Numerical Relativity Vol.674: A. G. Hunt, Percolation Theory for Flow in Porous Media Vol.675: M. Kröger, Models for Polymeric and Anisotropic Liquids Vol.676: I. Galanakis, P. H. Dederichs (Eds.), Halfmetallic Alloys Vol.678: M. Donath, W. Nolting (Eds.), Local-Moment Ferromagnets Vol.679: A. Das, B. K. Chakrabarti (Eds.), Quantum Annealing and Related Optimization Methods Vol.680: G. Cuniberti, G. Fagas, K. Richter (Eds.), Introducing Molecular Electronics Vol.681: A. Llor, Statistical Hydrodynamic Models for Developed Mixing Instability Flows Vol.682: J. Souchay (Ed.), Dynamics of Extended Celestial Bodies and Rings Vol.683: R. Dvorak, F. Freistetter, J. Kurths (Eds.), Chaos and Stability in Planetary Systems Vol.685: C. Klein, O. Richter, Ernst Equation and Riemann Surfaces Vol.686: A. D. Yaghjian, Relativistic Dynamics of a Charged Sphere Vol.687: J. W. LaBelle, R. A. Treumann (Eds.), Geospace Electromagnetic Waves and Radiation

Our partners will collect data and use cookies for ad personalization and measurement. Learn how we and our ad partner Google, collect and use data. Agree & close