1: %\documentclass[a4paper,draft]{article}
2: \documentclass[a4paper]{mye}
3: \usepackage{natbib}
4: \usepackage[latin1]{inputenc}
5: \usepackage{amsbsy}
6: \usepackage{amsmath}
7: \renewcommand{\vec}[1]{\boldsymbol{#1}}
8: \usepackage{graphicx}
9:
10:
11: \newcommand{\myfig}[1]{Fig.~\ref{#1}}
12: \newcommand{\myFig}[1]{Fig.~\ref{#1}}
13: \newcommand{\myeq}[1]{Eq.~(\ref{#1})}
14: \newcommand{\myEq}[1]{Eq.~(\ref{#1})}
15: \newcommand{\eiscat}{\mbox{EISCAT}}
16: \newcommand{\eg}{e.g.\ }
17: % I know this is really brokem but I have to get this printed now.
18: \renewcommand{\title}[1]{
19: \begin{center}
20: {\LARGE#1}\vskip 1.5em
21: \end{center}}
22: \renewcommand{\author}[2][Q]{\textbf{#2}\protect$^{#1}\protect$}
23: \renewcommand{\affil}[2][]{$^{#1}$#2\par}
24:
25: \begin{document}
26: \thispagestyle{empty}
27: % This is also broken, take away with ms option
28: %\twocolumn[%
29: {\title{Auroral field-aligned currents by incoherent scatter
30: plasma line observations in the E region}
31: \vskip 1em
32: \flushleft
33: \begin{center}
34: \author[1,2]{Ingemar~Häggström},
35: \author[2]{Mikael~Hedin},
36: \author[1]{Takehiko~Aso},
37: \author[2]{Asta~Pellinen-Wannberg}
38: \author[2]{and Assar~Westman}
39: \vskip 1em
40: \affil[1]{National Institute of Polar Research,
41: 1-9-10 Kaga, Itabashi-ku,\\ Tokyo 173-8515, Japan.}
42: \affil[2]{Swedish Institute of Space Physics, Box 812,
43: S-981 28 Kiruna, Sweden.}
44: \end{center}
45: \vskip 3em
46: }
47: %\journal{APUAR}
48:
49: \date{} %\renewcommand{\baselinestretch}{2}\normalsize
50:
51:
52: %\begin{document}
53: %\maketitle
54: %\printfigures
55:
56:
57: \begin{abstract}
58: \noindent
59: The aim of the Swedish-Japanese \eiscat{} campaign in
60: February 1999 was to measure the ionospheric parameters inside and
61: outside the auroral arcs. The ion line radar experiment was
62: optimised to probe the E-region and lower F-region with as high
63: a speed as possible. Two extra channels were used for the plasma line
64: measurements covering the same altitudes, giving a total of 3
65: upshifted and 3 downshifted frequency bands of 25~kHz each. For most
66: of the time the shifted channels were tuned to 3 (both), 4 (up), 5.5
67: (down) and 6.5 (both) MHz.
68:
69:
70: Weak plasma line signals are seen whenever the radar is probing the diffuse
71: aurora, corresponding to the relatively low plasma frequencies. At
72: times when auroral arcs pass the radar beam, significant increases
73: in return power are observed. Many cases with simultaneously up and
74: down shifted plasma lines are recorded. In spite of the rather
75: active environment, the highly optimised measurements
76: enable investigation of the properties of the plasma
77: lines.
78:
79:
80: A modified theoretical incoherent scatter spectrum is used to
81: explain the measurements. The general trend is an upgoing
82: field-aligned suprathermal current in the diffuse aurora,
83: There are also
84: cases with strong suprathermal currents indicated by large
85: differences in signal strength between up- and downshifted plasma
86: lines. A full fit of the combined ion and plasma line spektra
87: resulted in suprathermal electron distributions consistent with models.
88: \end{abstract}
89:
90:
91: \section{Introduction}
92: \label{sec:introduction}
93:
94:
95: The incoherent scatter spectrum consists mainly of two lines, the
96: widely used and relatively strong ion line and the very weak easily
97: forgotten broadband electron line. There is also another line present,
98: the plasma line, due to scattering from high frequency electron
99: waves, namely Langmuir waves. From the downgoing and upgoing Langmuir waves,
100: two plasma lines can be detected by the radar. The frequency shift
101: from the transmitted signal is the frequency of the scattered
102: Langmuir wave plus the Doppler shift caused by electron drift. Plasma lines
103: can be used to measure the electron drift and hence the line-of-sight
104: electric current. The problem in ion line analysis with the
105: uncertainty of the radar
106: constant can be solved by the plasma line frequency
107: determination and when that is done
108: the speed
109: of measurement can be significantly increased by including the plasma
110: line in the ion line analysis. However, since the frequency of the
111: plasma lines is not known beforehand, and the frequency is varying
112: with height, it is difficult to measure them with enough resolution.
113:
114:
115: There have been a number of reports on plasma line measurements and
116: their interpretation. Most of them have discussed the frequency
117: shift from the transmitted pulse and the scattering has mainly been
118: from the F-region peak, \eg \citet{showen1979rs}, \citet{kofman1993jgr} and
119: \citet{nilsson1996jatp}. The latter two showed also that the simple
120: formula for the Langmuir wave frequency,
121: \begin{equation}
122: \label{eq:1}
123: f^2=f_p^2 (1+3k^2\lambda_D^2)+f_c^2\sin^2\alpha
124: \end{equation}
125: where $f_p$ is the plasma frequency, $k$ the wave number $f_c$ the
126: electron gyro frequency, $\lambda_D$ the electron Debye length and
127: $\alpha$ the angle between the scattering wave and the magnetic field,
128: is valid to within a few kHz and thus enough to set the radar system
129: constant. To be able to
130: deduce any electron drifts, or current, out of the positions of the lines
131: these authors also show that \myeq{eq:1} is not sufficient, and it is
132: necessary to carry out more accurate
133: calculations. \citet{hagfors1981jgr} had also to go to further
134: expansions in deriving the ambient electron temperature from the plasma lines.
135: The fact that so many reports deal with the F region peak is due to
136: the altitude profile shape of the plasma line frequency, which according to
137: \myeq{eq:1} will also show a peak around that height. The measurements
138: can thus be
139: made relatively easily using rather coarse height resolution but good
140: frequency resolution and only detect the peak frequency. Measurements
141: using the same strategy, but at other heights, have been made with a
142: chirped radar by matching the plasma line frequency height gradient and the
143: transmitter frequency gradient
144: \citep{birkmayer1986jatp,isham1993jgr}. This technique allows
145: determination of the Langmuir frequency with very high frequency
146: resolution, but do not use the radar optimally, since the chirped
147: pulse cannot be used for anything else than plasma line measurements.
148:
149: The enhancement of
150: the plasma lines, which occurs in the presence of suprathermal
151: electron fluxes \citep{perkins1965pr}, either photoelectrons or
152: secondaries from auroral
153: electrons, has been investigated by \citet{nilsson1996anngeo}, where
154: they also calculate predictions of plasma line strength for different
155: incoherent scatter radars and altitudes. They also show that the power
156: of the plasma line is rather structured with respect to ambient
157: electron density, depending on fine structure in the suprathermal
158: distributions due to excitations of different atmospheric
159: constituents.
160:
161:
162: Incoherent scatter plasma lines in aurora are more difficult
163: to measure since the variations in the plasma parameters are strong
164: with large time and spatial gradients. Reports of auroral plasma lines in the
165: aurora are also more rare, and most of them are based on too coarse
166: time resolution
167: \citep{wickwar1978jgr,kofman1980jgr,oran1981jgr,valladares1988jgr},
168: with resolutions ranging from 30 seconds up to 20 minutes. The
169: enhancements over the thermal level were high, but consistent with
170: what could be expected of model calculations of suprathermal electron
171: flux. They also tried to calculate currents and electron temperature
172: from the frequency shifts of the plasma lines but with very large
173: error bars.
174: \citet{kirkwood1995jgr} used the \eiscat{} radar
175: and the filter bank technique
176: and recorded much higher intensities of the plasma lines, since they
177: got down to resolutions of 10 seconds, and showed also
178: that the plasma-turbulence model proposed by
179: \citet{mishin1994jgr} was not consistent with the data, but could be
180: explained by reasonable fluxes of suprathermal electrons.
181:
182:
183: In this paper we present data obtained with the high resolution
184: alternating code technique \citep{lehtinen1987rs}, as
185: was also done for F-region plasma lines by \citet{guio1996anngeo}, with even
186: higher intensities due to the time resolutions of 5 seconds.
187: An interesting, but at the time of the experiment not realisable at \eiscat,
188: technique would have been the type of coded long pulses used by
189: \citet{sulzer1994jgr} for HF-induced plasma lines.
190: From relative strengths between up- and downshifted lines we detect a
191: general trend of upgoing field-aligned currents in the diffuse aurora
192: carried by the suprathermal electrons. We propose a generalisation of the
193: theoretical incoherent scatter spectrum, to include multiple shifted electron
194: distributions, and in one example we do a full 7-parameter fit of the
195: incoherent scatter spectrum, including the enhanced plasma lines
196: assuming Maxwellian secondary electrons, resulting in the first radar
197: measurement of its flux and a current carried by the thermal
198: electrons.
199:
200:
201: \section{Experiment}
202: \label{sec:experiment}
203:
204:
205: The measurements we present were collected by the 930~MHz \eiscat{} UHF
206: incoherent scatter radar, with its transmitter located at Ramfjordmoen
207: in Norway (69.6~°N, 19.2~°E, L=6.2). The signals scattered from the
208: ionosphere were received at stations in Kiruna, Sweden and Sodankylä,
209: Finland as well as at the transmitting site. General descriptions of the
210: radar facility are given by \citet{folkestad1983rs} and \citet{baron1984jatp}.
211: Local magnetic midnight at Ramfjordmoen is at about 2130~UT.
212: The aim, in the Swedish-Japanese \eiscat{} campaign in February 1999,
213: was to measure the ionospheric parameters inside and outside
214: the auroral arcs.
215:
216: For this a 3 channel ion line alternating code \citep{lehtinen1987rs}
217: experiment, optimised to probe the
218: E-region and lower F-region with as high a speed as possible, was
219: developed. The 16 bit strong condition alternating code with bitlengths of
220: 22~$\mu$s was used, giving 3 km range resolution, and with a sample rate of
221: 11~$\mu$s the range separations in consecutive spectra were 1.65~km.
222: \myFig{fig:txrx} shows the transmission/reception scheme of the first
223: 20~ms of the radar cycle.
224: The whole alternating code sequence takes
225: about 0.3~s to complete. During this period the incoherent scatter
226: autocorrelation functions (ACF) at the probed heights should not
227: change significantly for the alternating codes to work. In order to
228: keep this as short as possible, the short pulses, normally used for
229: zerolag estimation, were dropped and instead a pseudo zero lag,
230: obtained from decoding the power profiles of the different codes in
231: the alternating code sequence, was used. \myFig{fig:amb} shows the
232: range-lag ambiguity function for this lag centred around 0~$\mu$s in
233: the lag direction, but the main contribution to the signal comes from
234: around 7~$\mu$s. The more normal lag centred at 22~$\mu$s is also
235: shown for comparison.
236: The range extents are rather similar
237: but the power is, of course, considerably lower for the pseudo zero lag.
238: Nevertheless, this is taken care of in the analysis and this lag is
239: rather important in events with high temperatures giving broad ion
240: line spectra or narrow ACFs. The transmitting frequencies were chosen to give
241: maximum radiated power for a given high voltage setting.
242:
243: \begin{figure}[htbp]
244: \begin{center}
245: \includegraphics[width=\columnwidth]{Fig1.eps}
246: \caption{Transmission (black) and reception (dark gray) scheme for part of
247: SP-SW-ALIS. A 16-bit alternating code, 22~$\mu$s bits, is cycled
248: over 3 frequencies. The interscan period is 9~ms and the total
249: cycle takes 300~ms. The plasma line channels were set to sample
250: the same range extent as the ion lines.
251: \label{fig:txrx}}
252: \end{center}
253: \end{figure}
254:
255: \begin{figure}[htbp]
256: \begin{center}
257: \includegraphics[width=\columnwidth]{Fig2.eps}
258: \caption{Range-lag ambiguity function for the first
259: two lags, a) ion line and b) plasma line. The range, given in
260: $\mu$s, can be converted to km with multiplication of 0.15. The
261: differences between the ion and plasma line ambiguity functions
262: are due to the use of different receiver filters.
263: \label{fig:amb}}
264: \end{center}
265: \end{figure}
266:
267: \begin{figure}[htbp]
268: \begin{center}
269: \includegraphics[width=\columnwidth]{Fig3.eps}
270: \caption{The frequency setup for the
271: experiment. The plasma line channels were fixed to 926 and 934.5~MHz,
272: giving "simultaneously" up and downshifted frequencies at 3
273: and 6.5~MHz. In addition there is an upshifted line at 4 and a
274: downshifted at 5.5~MHz.
275: \label{fig:freq}}
276: \end{center}
277: \end{figure}
278:
279: The monostatic plasma line part of the experiment used two
280: channels covering the same ranges as the ion line but with 3.3~km
281: range separations between the spectra. The frequency setup for the experiment,
282: illustrated in \myfig{fig:freq}, gives the possibility for 3 upshifted
283: and 3 downshifted bands. For 3~MHz and 6.5~MHz frequency shifts, both
284: the up- and downshifted plasma lines were measured. In addition there
285: was a 4~MHz upshifted and a 5.5~MHz downshifted band. The width of
286: these bands should have been set to match the bit length of the codes
287: used, 50~kHz, but unfortunately this was not the case and 25~kHz wide
288: filters were erroneously used. This gave naturally less signal
289: throughput, and in \myfig{fig:amb} the corresponding range/lag ambiguity
290: functions for the plasma line channels are included for the first two
291: lags. The decoding still works, giving just slightly increased
292: unwanted ambiguities, but above all the pseudo zero lag is moved out
293: to a larger lag value. This fact, with one exception, almost ruled out
294: the possibilities to
295: measure plasma line spectra because, as will be shown here later,
296: they are likely to be rather wide, due to large time and height
297: gradients of the plasma parameters. This lead the analysis to use mainly the
298: undecoded zerolag, which after integration over the different codes,
299: almost resembles the shape of a 352~$\mu$s (16$\times22~\mu$s) long pulse.
300:
301:
302: The experiment contained a large
303: number of antenna pointings in order to follow the auroral arcs, but
304: as this day was cloudy over northern Scandinavia the transmitting antenna was
305: kept fixed along the local geomagnetic field line. The remote sites,
306: receiving only ion lines,
307: were monitoring the same pulses as the transmitting site, and were
308: used to measure the drifts in the F-region, to derive the electric
309: field. Thus, these antennas intersected the transmitted beam at
310: the F-region altitude giving the best signal, for this day mostly at
311: 170~km.
312:
313:
314: \section{Measurements}
315: \label{sec:measurements}
316:
317:
318: \subsection{Ion line}
319: \label{sec:ion-line}
320:
321:
322:
323: The experiment started at 1900~UT on 14 February 1999 and continued until
324: 2300~UT. \myFig{fig:ionover} shows an overview of the parameters deduced
325: from the ion line measurements, which were analysed using the on-line
326: integration time, 5 seconds, in order to be compared to the plasma
327: line data. The analysis was done using the GUISDAP package
328: \citep{lehtinen1996jatp}, but a correction of 45\% of the radar system
329: constant used in the package had to be invoked to fit the
330: plasma line measurements according to \myeq{eq:1}.
331: This short integration time was possible due to the highly
332: optimised mode used, with all the transmitter power concentrated to
333: the E-region. In range, some integration was done, so that at lower
334: heights 2 range gates were added together and with increasing height
335: the number of gates added together increases to 15 in the F-region.
336:
337: There is a rather strong E-region from the start,
338: but no real arcs, and we interpret this as diffuse aurora. The peak
339: electron density shows some variation, but as time goes the
340: E-region ionisation decreases until 2050~UT, where it is almost gone.
341: The density
342: peak during this time was at around 120~km altitude and the lower edge
343: of the E-region
344: at 110~km, but at times the ionisation reaches down to 100~km. At 2050~UT
345: and onwards until 2240~UT the ionosphere above Tromsø became more
346: active and several auroral arcs passed the beam. Around some of the
347: arcs there are short-lived enhancements of electric field, seen as
348: F-region ion and E-region electron temperature increases. In the last
349: 10 minutes of the experiment the arc activity
350: disappeared and again there was diffuse aurora. From the field aligned
351: ion drifts it is evident that there is a rather strong wave activity in the
352: diffuse aurora until 2050~UT, while it is not so clear in the
353: continuation of the experiment. The last panel with the inferred
354: electron density from the pseudo zero lag shows the same features as
355: the fitted density panel but with highest possible resolution since no
356: height integration is made.
357:
358: \begin{figure*}[htbp]
359: \begin{center}
360: \includegraphics[width=\textwidth]{Fig4.eps}
361: \caption{Summary of results from the ion line experiment. All
362: panels shows the parameters in a altitude versus time fashion
363: with the antenna directed along the geomagnetic field line.
364: Panels from top: Electron density, electron temperature, ion
365: temperature, line-of-sight ion drift velocity and electron
366: density based on only the returned power. The data were analysed
367: with 5~s integration due to the active conditions.
368: \label{fig:ionover}}
369: \end{center}
370: \end{figure*}
371:
372: \subsection{Plasma line}
373: \label{sec:plasma-line}
374:
375: Since the analysis of the plasma lines was forced to handle the
376: undecoded zerolags of the alternating codes, it was necessary to
377: analyse their profile shape. \myFig{fig:planal} shows how this analysis
378: was performed. From the fitted parameters of the ion line, electron
379: density and temperature, a profile of the approximate Langmuir
380: frequency can be calculated using \myeq{eq:1}.
381: When probing at fixed frequency, there
382: will be scattered signal only from heights where the probing and
383: Langmuir frequencies match each other. The effect of the undecoded
384: zerolag is similar to the one where a normal long pulse is used, but
385: with a lower signal strength. So, the profile shape will be a
386: square pulse centred on the corresponding altitude, since no gating is
387: performed. Because the signal
388: strengths are rather weak compared to the system temperature, there is
389: a great deal of noise in the profile shape. To be able to
390: extract the altitude and signal power, a fitting procedure need to be
391: performed. For this a continous function and a good first guess is
392: needed and the measured plasma line profile was convolved with the
393: pulse shape to have a triangular shape and also showing a
394: peak close to the matching height.
395:
396: \begin{figure}[htbp]
397: \begin{center}
398: \includegraphics[width=\columnwidth]{Fig6.eps}
399: \caption{Description of the analysis
400: procedure. a) Langmuir frequency profile together with the
401: probing frequency of 3~MHz shift. b) The echo profile using a
402: short pulse. c) The echo profile using a long pulse. d) The
403: echo profile of a long pulse convolved with the pulse shape.
404: \label{fig:planal}}
405: \end{center}
406: \end{figure}
407:
408: The plasma line part of the experiment is overviewed in \myfig{fig:plover}.
409: The signal strengths shown should be compared with the UHF system
410: temperature of about 90~K. At first glance there is almost nothing in
411: the upshifted part, but a more careful look shows weak signals between
412: 1940 and 2030~UT and after 2250~UT, corresponding to the occurrence of
413: diffuse aurora. Similar echoes can also be seen in the downshifted
414: channel, and are due to plasma lines at 3~MHz offset from the
415: transmitted frequencies, according to the ion line measurements.
416: In the downshifted part after 2100~UT,
417: frequent events of rather strong signals in phase with auroral arcs
418: pass the beam. Most of these are from the 5.5~MHz shift, but some
419: of them also are due to plasma lines at 6.5~MHz. Such events are less
420: frequent and for most of them there are also signals in the upshifted
421: part.
422:
423: \begin{figure*}[htbp]
424: \begin{center}
425: \includegraphics[width=\textwidth]{Fig5.eps}
426: \caption{Overview of the plasma line measurements. The panels
427: show the undecoded upshifted (top panel) and downshifted
428: returns, given as signal strengths in Kelvin.
429: \label{fig:plover}}
430: \end{center}
431: \end{figure*}
432:
433: Due to the fact that the same channel is used for several frequencies,
434: there can be several altitudes that fulfill the matching condition
435: between Langmuir and probing frequency. This complicates the analysis
436: somewhat, and there have to be several fits with different numbers of
437: triangles superposed on each other. Some examples of this analysis are
438: shown in \myfig{fig:plex}, from only one plasma line to several both up-
439: and downshifted lines. Using a lower limit of 2~K signal power, the total
440: number of 5 second integration events with enhanced plasma lines for
441: this evening
442: was 256, and a total of 468 plasma line echoes were detected, divided
443: into 220 on 3~MHz, 19 on 4~MHz, 157 on 5.5~MHz and 72 on 6.5~MHz.
444:
445: \begin{figure}[htbp]
446: \begin{center}
447: \includegraphics[width=\columnwidth]{Fig7.eps}
448: \caption{Examples of measured convolved profiles (circles) and the
449: triangle(s) fitted (lines).
450: Included on the plots are the profiles of the
451: Langmuir frequency (black) calculated from the parameters
452: of ion lines fit. The value on the x-axis represent signal
453: strength (K) and frequency (MHz) respectively. The numbers
454: in the title shows the time of measurement and how many
455: echoes was detected for the up- (red) and downshifted
456: (green) frequencies. A lower limit of 2~K received signal
457: power (2\%~SNR) was used. a) Simultaneous up and
458: downshifted echoes at 3~MHz in diffuse aurora. b) An
459: upshifted 3~MHz echo in diffuse aurora. c) A strong
460: downshifted echo at 5.5~MHz in an auroral arc. d)
461: Simultaneous up and downshifted echoes at 6.5~MHz and a
462: downshifted 5.5 in an auroral arc.
463: \label{fig:plex}}
464: \end{center}
465: \end{figure}
466:
467:
468: \section{Theory}
469: \label{sec:theory}
470:
471:
472: In order to relate the plasma line measurements to physical quantities
473: it is necessary to investigate the spectrum of the incoherent scatter
474: process. The Nyquist theorem approach, derived in a long series of papers
475: by \citet{dougherty1960procrsl,dougherty1963jgr}, \citet{farley1961procrsl},
476: \citet{farley1966jgr} and finally \citet{swartz1979jgr}, arrives at
477: \begin{equation}
478: % \begin{multline}
479: \label{eq:2}
480: \sigma(\omega) =\frac{N_e r_e^2 \sin^2\delta} {\pi} \cdot %\\
481: \frac
482: {\left|
483: y_e
484: \right|^2
485: \sum_i \frac{ \eta_i \Re (y_i) } {\omega-\vec k \cdot \vec v_i} +
486: \left|
487: jk^2\lambda_D^2 + \sum_i \mu_i y_i
488: \right|^2 \frac
489: { \Re (y_e) } {\omega-\vec k \cdot \vec v_e}}
490: {\left|
491: y_e + jk^2\lambda_D^2 + \sum_i \mu_i y_i
492: \right|^2},
493: % \end{multline}
494: \end{equation}
495: where
496: \begin{equation}
497: \label{eq:3}
498: \eta_i=\frac{n_i q_i^2}{N_e e^2},
499: \end{equation}
500: \begin{equation}
501: \label{eq:4}
502: \mu_i=\frac{\eta_i T_e}{T_i},
503: \end{equation}
504: and the index $e$ stands for electrons and $i$ for the different ion species.
505: $N$ and $n$ are the densities, $r_e$ the classical electron radius, $v$ the bulk velocity, $q$ the
506: charge, $e$ the electron charge and $T$ the temperature.
507: The complex normalised admittance function, $y$, contains most of the
508: physics with the plasma dispersion function and have as main arguments
509: the collision frequency and magnetic field.
510: The same result
511: was also reached by \citet{rosenbluth1962pf}, using the dressed particle
512: approach. In \myfig{fig:wideis} there is an example of the spectrum,
513: showing clearly the strong ion line around zero offset frequency and
514: the rather weak plasma lines at rather large offsets. The plasma lines
515: become enhanced by a photo electron or auroral electron produced
516: suprathermal electron distribution, and in order to simulate what this
517: extra distribution does to the spectrum a modification to the formula has
518: to be made. First, a rewriting of \myeq{eq:2} following
519: \citet{swartz1978rs} has to be performed
520: in order to separate the electron and ion contributions to the
521: spectrum:
522: \begin{equation}
523: %\begin{multline}
524: \label{eq:5}
525: S(f) = \frac{1} {\pi} \cdot %\\
526: \frac{ \left| \frac{N_e y_e}{T_e} \right|^2
527: \sum_i \frac{ n_i Z_i^2 \Re(y_i) } {f+k v_i/2\pi} +
528: \left| jC_D + \sum_i \frac {n_i Z_i^2 y_i}{T_i} \right|^2 \frac {
529: N_e \Re(y_e) } {f+k v_e/2\pi}}
530: {\left| \frac{N_e y_e}{T_e} + jC_D +
531: \sum_i \frac{n_i Z_i^2 y_i}{T_i} \right|^2},
532: %\end{multline}
533: \end{equation}
534: where
535: \begin{equation}
536: \label{eqcd}
537: C_D=\frac{k^2 \epsilon K}{e^2},
538: \end{equation}
539: $f$ is the frequency shift, $Z=q/e$, $\epsilon$ is the dielectricity
540: and $K$ is the Boltzmann constant. Here, a
541: normalisation of the spectrum has also been performed, so that the zero lag
542: of the corresponding ACF is the raw electron density and the vector
543: velocity is replaced by the line-of-sight velocity. Using a treatment
544: in analogy to the ion contribution, it is now possible to rewrite the
545: spectrum to support a number of Maxwellian electron distributions as
546: \begin{equation}
547: %\begin{multline}
548: \label{eq:6}
549: S(f) = \frac{1}{\pi} \cdot %\\
550: \frac{ \left| \sum_e \frac{N_e y_e}{T_e} \right|^2
551: \sum_i \frac{ n_i Z_i^2 \Re(y_i) } {f+k v_i/2\pi} +
552: \left| jC_D + \sum_i \frac {n_i Z_i^2 y_i}{T_i} \right|^2 \sum_e \frac {
553: N_e \Re(y_e) } {f+k v_e/2\pi}}
554: {\left| \sum_e \frac{N_e y_e}{T_e} + jC_D +
555: \sum_i \frac{n_i Z_i^2 y_i}{T_i} \right|^2}.%\\.
556: %\end{multline}
557: \end{equation}
558:
559: \begin{figure}[htbp]
560: \begin{center}
561: \includegraphics[width=\columnwidth]{Fig8.eps}
562: \caption{The wide frequency incoherent scatter spectrum. The parameters
563: used: $N_e=4\cdot10^{11}$~m$^{-3}$, $T_e=700$~K, $T_i=600$~K,
564: $v_i=v_e=0$~ms$^{-1}$ and collision frequencies $\nu_e$ and
565: $\nu_i$ using the MSIS90e model (Hedin, 1991) for 140~km
566: altitude. Logarithmic scale for the strength is used to be able
567: to emphasise the different lines of the spectrum.
568: \label{fig:wideis}}
569: \end{center}
570: \end{figure}
571:
572:
573: \begin{figure}[htbp]
574: \begin{center}
575: \includegraphics[width=\columnwidth]{Fig9.eps}
576: \caption{Close-up of the ion and the plasma lines. Top panel
577: shows the electron distribution, middle panel the ion line and
578: bottom panel the plasma lines, where the frequency scale of the
579: downshifted line have been reversed. The figure shows the lines
580: with two different electron distributions, thick line for the
581: normal thermal distribution and thin line for the same
582: distribution together with an suprathermal distribution.
583: \label{fig:narrowis}}
584: \end{center}
585: \end{figure}
586:
587: \begin{figure}[htbp]
588: \begin{center}
589: \includegraphics[width=\columnwidth]{Fig10.eps}
590: \caption{The plasma lines based on an upgoing (thick line)
591: and downgoing 10~eV beam of suprathermal electrons at a
592: current of 1 $\mu$Am$^{-2}$ and temperature of 5~eV. The
593: thermal parameters are the same as in \myfig{fig:wideis}
594: \label{fig:plstrength}}
595: \end{center}
596: \end{figure}
597:
598: In \myfig{fig:narrowis} the effect on the plasma lines of a suprathermal
599: distribution with a reasonable density of $10^7$~m$^{-3}$ and width of 10~eV
600: is shown, being lower than the ionisation energy for most ions. The
601: plasma lines grow considerable and the integrated power over the
602: bandwidth used in the experiment becomes comparable to the power of
603: the ion line. It may also be noted that there is no effect at all seen
604: in the ion line. \citet{perkins1965pr} have shown similar
605: calculations, but when their method was based on large expansions to
606: allow non-Maxwellian distributions one can here more directly superpose a
607: few Maxwellian distributions to explain the measurements and even make
608: fits of the spectra taken to get estimates of the suprathermal
609: distributions, which is of great importance in auroral measurements.
610: A consequence of \myeq{eq:6} is that it is also possible to
611: derive the spectrum assuming currents carried by the suprathermal
612: distribution, since it allows different drift velocities on the
613: various distributions. Indeed, \myfig{fig:plstrength}, shows differential
614: strengths on the two plasma lines, with upgoing electrons enhancing mainly the
615: downshifted line and downgoing ones the upshifted line.
616:
617:
618: \section{Discussion}
619: \label{sec:discussion}
620:
621:
622:
623: Plasma line measurements in the active auroral ionosphere are not an
624: easy task, due to the large variations in the ionospheric parameters.
625: The Langmuir frequencies are largely dependent on the ambient electron
626: density, making the line move considerably as the density changes,
627: which it does on time scales of seconds. Moreover, the density height
628: gradient makes the lines very broad when measured over a specific
629: height interval, and at times even broader than the receiver
630: band. A chirped radar would solve only a part of the problem at the
631: cost of transmitter power.
632: These complications make it hard to draw any conclusions on the power
633: in the lines, as one does not know for sure the scattering volume or the
634: time duration of the scattering. Bearing this in mind and to at least
635: minimise these effects, one can nevertheless look at the different
636: distributions in height and power for the different lines to get an
637: idea of their nature, using the on-line integration time of 5 seconds.
638:
639: The altitude distributions for the different frequencies are shown in
640: \myfig{fig:scatterall}. One must note that the signal levels shown are not
641: corrected for range, as the scattering volumes are not known, so
642: signals from a higher altitude are in fact stronger than the
643: corresponding signal from a lower height. It is clearly seen that the
644: strongest signal is the 5.5~MHz line, followed by 6.5, 4 and 3~MHz.
645: The altitude distribution shows more or less the expected dependence
646: on range, but there are some exceptions: In one point at 188~km in the
647: 3~MHz band and for the 5.5~MHz band the strong values between 130 and
648: 140~km seem to be stronger than the others, even taking into account
649: the range effect. However, the number of points are too few to be used
650: as evidence on altitude effects. Most of the echoes are coming from around
651: 120-150~km altitude and \myfig{fig:simelpower} shows a simulation of the
652: expected strength of the plasma line for given background electron
653: density. It shows a peak at around 5.5~MHz and this is also what the
654: experiment shows. A more realistic suprathermal distribution will
655: decrease the returned power for a number of frequencies and one should
656: see the figure as an upper-limit estimate. Indeed, \citet{nilsson1996anngeo},
657: have made predictions of the expected strength of Langmuir
658: waves for different heights and carefully derived distributions. These
659: predictions are in rather good agreement with the present measurements
660: showing a strong peak between 5 and 6.5~MHz.
661: Although there is some uncertainty on the scattering volume, the
662: observed strengths of the plasma lines can be used to get some
663: estimates of the distribution of the suprathermal population.
664: Assuming a width of 6~eV, one need, to get to the measured
665: strengths for the 3~MHz case, a suprathermal density of about
666: $5\cdot10^8$~m$^{-3}$ which is 0.5~\% of the thermal population.
667: For the 5.5 and 6.5~MHz cases, it is enough with only
668: $1\cdot10^8$~m$^{-3}$, but the uncertainty of volume is even larger
669: here due to the active environment and the values should maybe be the same as
670: for the lower frequency offsets.
671:
672: \begin{figure}[htbp]
673: \begin{center}
674: \includegraphics[angle=90,width=\columnwidth]{Fig11.eps}
675: \caption{Scattered plots of echo
676: altitude and strength for the different frequency shifts. The
677: altitudes range from 100 to 210~km, E and lower F region.
678: \label{fig:scatterall}}
679: \end{center}
680: \end{figure}
681:
682: \begin{figure}[htbp]
683: \begin{center}
684: \includegraphics[width=\columnwidth]{Fig12.eps}
685: \caption{Simulation of the plasma line strength at 140~km altitude with an
686: added distribution of secondaries for different electron
687: densities from 0.5 to $10 \cdot 10^{11}$ m$^{-3}$. The envelope
688: in the lowest panel shows the total power of the plasma lines
689: versus the Langmuir frequency.
690: \label{fig:simelpower}}
691: \end{center}
692: \end{figure}
693:
694: The most interesting
695: thing with incoherent plasma lines is, of course, the possibility to
696: derive differential drifts between ions and electrons, and from these
697: deduce ionospheric currents. For this, one needs to measure the up-
698: and downshifted lines simultaneously. Although the time of the
699: measurement for both lines were not exactly the same in this
700: experiment, the time shifts between them are so small (3-6~ms)
701: compared to the total cycle time (300~ms) of the codes, that this
702: effect is of minor importance. In \myfig{fig:scatter36}, the strengths and
703: altitudes of the two concurrently recorded up- and downshifted plasma
704: lines are shown. In general, there are stronger up- than downshifted
705: lines for the 3~MHz case, whereas no such trend can be seen in the 6.5
706: MHz band. However, there are exceptions to these overall trends
707: and on occasions there are large differences in the signal strengths
708: between the lines. Almost all of the 3~MHz plasma lines were recorded
709: in diffuse aurora and this evident difference in signal power needs a
710: closer examination. When there is a drift of the thermal electrons, the
711: plasma lines shift, and when probing at a fixed frequency, the scatter
712: may not come from the same altitude for the up- and downshifted lines
713: respectively. The strength of the plasma line is also rather altitude
714: dependent due to the damping by the collisions of electrons with ions and
715: neutrals. But \myfig{fig:scatter36} shows no general height difference
716: between the up- and downshifted 3~MHz lines, so this difference in strength
717: cannot be explained by thermal electron bulk drifts. To simulate the effect of
718: current carried by suprathermal electrons, \myfig{fig:simfreqpower}
719: illustrates the strength of the 3~MHz plasma lines for Maxwellian
720: electron beams of different energies. With no net current the lines
721: are of almost the same strength, and the difference is mainly dependent on
722: where in the receiver band the lines are. However the experiment shows
723: stronger upshifted lines, thus it is evident that the diffuse aurora this
724: night contained fluxes of downgoing suprathermal electrons or, in other
725: words, there was an upgoing current carried by suprathermal electrons.
726: The average power ratio between up- and downshifted lines, 1.4, can be
727: used to estimate the amount of current carried by the suprathermals.
728: As before, using a density of $5\cdot10^8$~m$^{-3}$ and temperature of 6~eV,
729: and shifting this population to a current
730: density of 15~$\mu$Am$^{-2}$ brings to the observed ratios. This
731: current seems somewhat high, but not unrealistic.
732:
733:
734: For the 6.5~MHz bands there may be a slight difference with respect to
735: the altitude, and that is most likely due to thermal currents causing
736: frequency shifts of the plasma lines. This effect is not very clear,
737: but as lower heights have higher density, or Langmuir frequency, and
738: as the downshifted line is at a slightly lower height, this is most
739: probably an effect of a downgoing current carried by thermal
740: electrons. The correction due to heat-flow in the plasma dispersion
741: function, discussed by \citet{kofman1993jgr} and confirmed later also
742: by \citet{nilsson1996jatp} and \citet{guio1996anngeo}, but not taken into
743: account here, would also show the same effect in altitude difference
744: between the lines. The band widths used here, 25~kHz, are much wider
745: than the effect of heat-flow, which is less than or around 1~kHz in the
746: F-region and much lower in the E-region, so that cannot explain the
747: 6.5~MHz height differences. A recent paper by \citet{guio1998anngeo}
748: with proper calculation of the dispersion equation
749: investigates the heat-flow and finds that it is not necessary
750: to invoke the effect at all, but as this paper don't go to the same
751: extreme the heat-flow has to be considered.
752:
753: \begin{figure}[htbp]
754: \begin{center}
755: \includegraphics[angle=90,width=\columnwidth]{Fig13.eps}
756: \caption{Scatter plots of power and altitude for the two simultaneously
757: measured up and downshifted lines at 3 and 6.5~MHz. The
758: triangles and circles for the power of the 3~MHz case, in the
759: upper left, show lines detected before and after magnetic
760: midnight respectively --- no difference can be seen.
761: \label{fig:scatter36}}
762: \end{center}
763: \end{figure}
764:
765: \begin{figure}[htbp]
766: \begin{center}
767: \includegraphics[width=\columnwidth]{Fig14.eps}
768: \caption{Simulation of 3~MHz plasma line strength
769: at 140 km with an electron beam with varying mean energy from 0
770: to 20~eV and temperature of 10~eV, corresponding to a few
771: $\mu$Am$^{-2}$ field-aligned currents, depending on energy. The top
772: panel shows the ratio between plasma and ion line strengths for
773: upgoing beams and the bottom for downgoing beams.
774: \label{fig:simfreqpower}}
775: \end{center}
776: \end{figure}
777:
778: \begin{figure}[htbp]
779: \begin{center}
780: \includegraphics[width=\columnwidth]{Fig15.eps}
781: \caption{Measured spectra of the both plasma
782: lines and the ion line and the best 7-parameter fit of the
783: theoretical spectrum. The parameters fitted were $N_e=9.7 \cdot
784: 10^{10}$~m$^{-3}$, $T_e=631$~K, $T_i=697 $~K,
785: $v_i=-100$~ms$^{-1}$,
786: $v_e=657$~ms$^{-1}$ and a suprathermal distribution
787: with $n_e=0.8 \cdot 10^9$~m$^{-3}$ and $T=11$~eV.
788: \label{fig:fullfit}}
789: \end{center}
790: \end{figure}
791:
792: Most of the above discussions on currents are only on directions,
793: but to get any quantitative numbers it is necessary to
794: look at the individual spectra. Of the 256 events of enhanced plasma
795: lines, there is only one that is good enough to investigate. All the
796: other are too broad either due to the Langmuir frequency gradient
797: smearing or the fact that the Langmuir frequency at the height in
798: question is varying
799: during the 5 second time slots. A further reason is that the
800: stationarity condition for the alternating code technique is not
801: fulfilled, due to changes of Langmuir frequency, and the spectra
802: change too much within the 300~ms cycle. Anyway, there is one, and
803: the up- and downshifted bands, together with the ion line band, are
804: shown in \myfig{fig:fullfit}. The shifts in this case are around 3~MHz as
805: deduced from the ion line analysis. These spectra were then fitted to
806: the theoretical spectrum in \myeq{eq:6} using 7 ionospheric
807: parameters, the thermal parameters for both ions and electrons
808: and the parameters of a non-shifted suprathermal electron distribution.
809: At the first glance the fit on the ion line seems rather poor, but the
810: fit was done as usual in the time domain with proper weighting on the
811: different lags of the ACF, and in the FFT process to produce
812: \myfig{fig:fullfit} these statistical properties are lost. The fit
813: looks actually better in the time domain, but the frequency domain was
814: chosen in the figure to be more informative.
815: The current density $j=N_ee(v_i-v_e)$, and with the fitted parameters, the
816: field aligned current carried by the thermal electrons, amount to
817: 12~$\mu$Am$^{-2}$ downward. This is opposite the general current seen in
818: the plasma line strength for the 3~MHz band, but the magnitudes are
819: comparable. For this case no current could be seen on the
820: suprathermals, and it is evident that the currents are rather
821: structured. Of course, there are many more cases of no plasma lines
822: than enhanced plasma lines in the diffuse aurora, and some cases of
823: only one plasma line either up- or downshifted. It is not surprising that
824: this single example of thermal currents do not follow the general trend.
825: Again, heat-flow is not
826: taken into account, but that should increase the current somewhat.
827: The deduced suprathermal distribution is in good agreement with those
828: \citet{kirkwood1995jgr} derived from the precipitating flux of
829: primaries; of course in the present case the distribution is
830: Maxwellian and without the fine structure due to the atmospheric
831: constituents, but the numbers are comparable.
832:
833:
834:
835: \section{Conclusions}
836: \label{sec:conclusions}
837:
838:
839: We have made measurements of plasma lines in the active auroral
840: E-region. During 256 periods of 5 second integration we found a total
841: of 468 plasma line echoes, divided into 220 on 3~MHz, 19 on 4~MHz, 157
842: on 5.5~MHz and 72 on 6.5~MHz. It may seem strange to try to measure
843: plasma lines at such a low frequency as 3~MHz giving low signal
844: levels, but in fact most of the echoes and the most interesting results
845: came from this frequency offset.
846:
847:
848: The strongest echoes were found at the 5.5~MHz line, and somewhat
849: weaker ones at 6.5~MHz, inside auroral arcs. One must, however, note
850: that the strength measured inside the arcs is mostly a low-limit
851: estimate due to the active environment. The integration period used, 5~s,
852: is rather long in auroral arc conditions, and changes typically occur on
853: shorter time scales. Therefore, effects of gradients in the
854: Langmuir frequency profile, and hence scattering volumes, have not
855: been taken into account.
856:
857:
858: The Holy Grail in incoherent scatter plasma lines is the possibility
859: to measure currents, and in this case, the field-aligned currents in
860: aurora. The simulations carried out here, the extended full incoherent
861: scatter spectrum with multi-Maxwellian distributions of electrons, show that
862: the strength of the lines is determined by the suprathermal part of
863: the electron distribution, and the frequency mainly by the thermal
864: part. For simultaneous up- and downshifted plasma line differences in
865: intensity we can deduce currents carried by suprathermals, and for
866: differences in frequency, currents carried by thermals.
867:
868:
869: The simultaneous up- and downshifted frequencies of the 3 MHz line in
870: the diffuse aurora show, on average, an upward field-aligned
871: suprathermal current during the two main periods when they were
872: detected, 1940-2030~UT and 2250-2300~UT. In the arcs in general,
873: there is an indication of downward thermal current as seen from the
874: altitudes of the 6.5~MHz echoes. Of course, no rule is without
875: exceptions, and there are cases where one line is much stronger than
876: the other or the other line is not at all enhanced, indicating strong
877: currents.
878:
879:
880: In the full 7-parameter fit of the incoherent scatter spectrum with the
881: ion line and the both enhanced plasma lines, we obtained a thermal
882: current with a
883: suprathermal distribution of electrons consistent with distributions
884: derived from precipitating fluxes.
885:
886:
887:
888: \begin{acknowledgements}
889: One of the authors (I.H.) was working under a contract from NIPR and
890: is grateful to the Director-General of NIPR for the support. We are
891: indebted to the Director and staff of \eiscat{} for operating the
892: facility and supplying the data. \eiscat{} is an International
893: Association supported by Finland (SA), France (CNRS), the Federal
894: Republic of Germany (MPG), Japan (NIPR), Norway (NFR), Sweden (NFR)
895: and the United Kingdom (PPARC).
896: \end{acknowledgements}
897:
898: \bibliographystyle{egs}
899: %bibliographystyle{plainnat}
900:
901: \bibliography{strings,a_z,new}
902:
903: \end{document}
904: