Pages 645-654

OBSERVATIONS OF THE SOLAR WIND, THE BOW SHOCK, AND UPSTREAM PARTICLES WITH THE WIND 3D PLASMA INSTRUMENT

R.P. Lin1, D.E. Larson1, R.E. Ergun1, J.P. McFadden1, C.W. Carlson1, T.D. Phan1, S. Ashford1, K.A. Anderson1, M. McCarthy2, R. Skoug2, G.K. Parks2, H. Reme3, J.M. Bosqued3, C. d'Uston3, T.R. Sanderson4 and K.-P. Wenzel4

1Space Sciences Laboratory, University of California, Berkeley, E-mail:boblin@SSL.berkeley.edu
2Geophysics Program, University of Washington, Seattle
3Centre d'Etude Spatiale des Rayonnements, Toulouse, France
4Space Science Department of ESA, ESTEC, Noordwijk, Netherlands

ABSTRACT

The 3-D Plasma and Energetic Particle Instrument on the WIND spacecraft provides high sensitivity 3-D electron and ion measurements from solar wind plasma up to approximately 300 keV energy. We review a few of the many new results, including: 1) the detection of a quiet-time population (the "superhalo") of electrons extending up to ~102 keV energy; 2) solar impulsive electron events extending down to ~0.5 keV energy; 3) the remote sensing of the jump in magnetic field and the electric potential of the Earth's bow shock, using angular distributions of backstreaming electrons; and 4) the detailed distribution functions for the electrons producing solar type III radio bursts.

INTRODUCTION

For many years the solar wind has been regarded as a regime essentially independent from the energetic particle populations found in interplanetary space. The particles in the intervening energy range, from just above solar wind plasma to a few hundred keV, play a key role in the varied plasma and energetic particle phenomena observed to occur in the interplanetary medium (IPM) and upstream from the Earth's magnetosphere. Because of dynamic range considerations, previous instruments designed to measure the solar wind plasma ions and electrons lack the sensitivity to detect these suprathermal particles except during highly disturbed times.

The 3D Plasma and Energetic Particles Experiment (Lin et al., 1995) on the WIND spacecraft is designed to bridge the gap between solar wind plasma and energetic particle measurements by providing high sensitivity, wide dynamic range, good energy and angular resolution, full 3-D coverage, and high time resolution over the energy range from a few eV to  300 keV for electrons and  6 MeV for ions. Here we review some of the results for electrons from the first year of operation: 1) the detection of a quiet-time "super-halo" component of interplanetary electrons from ~1 keV to ~102 keV; 2) the discovery that the spectrum of impulsive solar electron events extend down to as low as ~0.5 keV energy; 3) the remote probing of the Earth's bow shock; and 4) detailed studies of the electron beam/ plasma wave interaction responsible for solar type III radio bursts. In addition, studies of the topology of magnetic clouds ejected from the Sun and of the structure of the magnetopause (Larson et al., 1997; Phan et al., 1997) are presented in these proceedings.

Fig. 1. Electron fluxes measured by the EESA-L detector on WIND. The top panel gives BN, the angle between the magnetic field vector and the model shock normal; gaps indicate when the field does not intersect the shock. The next two panels plot the average electron flux streaming out from the Sun (averaged over pitch angle 0° <  < 54°) and streaming back towards the Sun (124° <  < 180°), for 14 energy channels (right). The three lowest panels show the electron flux as a function of pitch angle at 780 eV, 56 eV and 13 eV. All fluxes have been computed in the solar wind plasma rest frame. (from Larson et al. 1996b).

PROBING THE EARTH'S BOW SHOCK

To illustrate the type of 3-D data obtained by this experiment, we use backstreaming electrons in the Earth's foreshock region as an example. Figure 1 (from Larson et al., 1996b) shows observations obtained with the Electron ElectroStatic Analyzer-Low (EESA-L) of the WIND 3-D Plasma and Energetic Particles instrument (see Lin et al. 1995 for description) in the Earth's foreshock. This detector measures electrons from ~10 eV to 1.1 keV with 15 logarithmically spaced energy steps and full 4p angular coverage in one spacecraft rotation (3 seconds). The data are sorted on board the spacecraft into 88 angular bins, each with roughly 22° x 22° resolution. These data are converted into units of distribution function and transformed into the solar wind rest frame using the solar wind velocity obtained from the Proton ElectroStatic Analyzer-Low (PESA-L) ion detector. A correction is made for an estimated spacecraft potential of ~8 V. Pitch angles for each angular bin are calculated using magnetic field data from the WIND Magnetic Field Instrument (MFI) [Lepping et al., 1995].

The top panel of Figure 1 shows BN calculated extrapolating the locally measured magnetic field vector in a straight line to the model bow shock [Slavin and Holzer, 1981] scaled to match the bowshock crossing observed at ~0046 UT. Essentially all connections to the shock have BN greater than 90° (B points into the bow shock surface). WIND is located at XGSE ~ 22 RE and YGSE ~ -21 RE, about 15 RE upstream of the bow shock, but the features observed here are representative of other times.

The following panel, labeled ``Solar'', shows that the flux of electrons flowing outward from the Sun (the halo, or solar heat flux electrons) measured in the solar wind plasma rest frame, is relatively stable over this time period. The next panel, labeled ``Backstreaming'', shows that the electron flux streaming back toward the Sun turns on and off abruptly. The bottom three color panels present the pitch angle spectrograms for 780, 56, and 13 eV electrons. Backstreaming flux at 780 and 56 eV are detected when BN indicates a magnetic connection to the shock. During the disconnected periods, the 13 eV flux is greater at 180° than at 0° pitch angle, because the solar wind core electrons which dominate the 13 eV flux must have a small net motion back toward the Sun to balance the current of halo electrons moving away from the Sun (Feldman et al., 1975).  Spikes of backstreaming electrons, often extending to high energies ( ~150° - 180° flux at 780 eV), are observed near the foreshock boundary, when the field line is nearly tangent to the shock surface.

Fig. 2. Full 3-D angular distribution for electrons in the deep foreshock. Each plot represents the normalized flux on a surface of constant energy in the solar wind plasma rest frame. The Hammer-Aitoff equal area projection is used to display 4p steradians angular coverage. The true bin resolution is shown for the highest energy step (top left). Interpolation is used to smooth between the bin centers. The projection is rotated to place the -B direction (diamond) in the center of each plot. Pitch angles are displayed in the next plot (top right). The circles in each successive energy step separate the populations of escaping magnetosheath electrons (inner region), reflected electrons (within annulus) and solar wind electrons (outer region). (from Larson et al. 1996b).

Figure 2 displays the full 3-D electron distribution function when WIND was deep in the foreshock (1801 UT), well away from the foreshock boundary. Backstreaming electrons appear in the center of each plot whereas the solar wind heat flux will appear split between the left and right edges of each oval. The distributions are essentially gyrotropic; the asymmetries in gyro-angle are due to instrumental effects or statistical fluctuations. The most striking feature of each plot is the enhanced ring of flux centered about the anti-magnetic field direction. The angular diameter of this ring increases with decreasing energy.

The distributions are most easily analyzed by transforming to the deHoffman-Teller reference frame (dHTF), where the upstream plasma flow is parallel to the magnetic field direction and the motional electric field is zero. Electrons with a sufficiently large pitch angle will be magnetically mirrored by the increased magnetic field in the shock, and travel back upstream with a parallel velocity equal in magnitude to its incident value in the dHTF.

The shock velocity, VSB , in the dHTF is parallel to B by construction, and given by VSB = VSW - VHT, where VHT =  x (VSW x B)/B, and  is the shock normal. The parallel velocity of the reflected electrons in the solar wind plasma rest frame is then given by v||r = -v||i + 2VSB. Assuming Liouville's Theorem holds, the phase space density of the reflected population can be written in terms of the phase space density of the incident (solar wind) population: fr(v||,v) = fsw(-v|| + 2VSB,v), where the values of v|| and v are limited to those outside the loss cone. Electrons with pitch angles less than the mirror angle will pass through the shock, thus the reflected population should exhibit a loss cone. Electrons escaping from the magnetosheath would populate this loss cone. The two black circles shown in each panel divide angular space into three regions: escaping magnetosheath electrons in the center, reflected electrons in the annulus, and solar wind electrons outside.

Fig. 3. Pitch angle distribution for the sample shown in Figure 2. Diamonds connected by dots show the approximate boundary between solar wind and backstreaming electrons. Triangles connected by dots separate flux due to reflected solar wind electrons from escaping magnetosheath electrons. Dashed lines show the predictions for reflected electrons assuming no loss cone. (from Larson et al. 1996b)

Assuming gyrotopy, the 3-D distributions shown in Figure 2 are used to calculate the phase space density
f(v||,v) shown in Figure 3. The angular width of the loss cone varies as a function of electron energy due to the effect of the cross-shock potential. By assuming conservation of energy and magnetic moment, it can be shown [Feldman et al. 1983, Fitzenreiter et al. 1990] that electrons with pitch angles less than the critical angle m, defined by

sin2(m) = B0/Bmax(1 + e/E)

will not be mirrored at the shock surface, whereas electrons with pitch angles greater than m will be reflected and return. Here B0 is the locally measured magnetic field strength and Bmax is the maximum field strength in the shock.  is the electrostatic potential across the shock (as measured in the dHT frame) and E is the particle's initial energy. For the example of Figure 3, the data fit to approximate values: Bmax/B0 = 5.0 and e = 86 eV.

Thus remote monitoring of the shock using 3D measurements of electron distribution can provide these parameters. Correlation to solar wind parameters may help to achieve a detailed understanding of the shock formation process.

SOLAR IMPULSIVE ELECTRON EVENTS

The steady-state solar wind electron population is dominated by a core with temperature kT ~10 eV, containing  95% of the plasma density and moving at about the solar wind bulk velocity, plus ~5% in a hot, kT ~ 80 eV, halo population carrying heat flux outward from the Sun, often in the form of highly collimated strahl [Feldman et al., 1975]. At energies of ~ keV and above, impulsively accelerated solar electron events occur at the Sun, on average, several times a day or more during solar maximum. As these electrons escape they produce solar and interplanetary type III radio bursts through beam-plasma interactions [see Lin, 1990 for review].

Fig. 4. Electron fluxes from ~100 eV to  100 keV for 27 December 1994. The solar electron event begins at ~1100 UT at ~100 keV, with velocity dispersion evident down to 624 eV. Two smaller events, at ~1100 UT and 1620 UT, are visible below ~6 keV. The dip at ~1500 UT at low energies is due to the close approach to the Moon, resulting in a plasma shadow. (from Lin et al. 1996)

Figure 4 shows the first observation [Lin et al., 1996] of solar impulsive electron events spanning the entire energy range from solar wind to suprathermal particle (few eV to hundreds of keV). A solar impulsive electron event begins at 1100 UT, easily identified by its velocity dispersion, e.g., the faster electrons arriving earlier, as expected if the electrons of all energies were simultaneously accelerated at the Sun and traveled the same distance along the interplanetary field to reach the spacecraft. The solar event can clearly be identified down to the 0.908 keV and even the 0.624 keV channels. A second, much weaker impulsive electron event is seen beginning about 1620 UT in the 8.77 keV channel. Another very small event may be starting at ~1120 UT at ~6 keV energy.

Fig. 5. Omnidirectional electron spectra (with pre-event electron fluxes subtracted) are shown for various times during the 27 December 1994 event, and averaged over the entire event. The pre-event electron spectrum shows the solar wind electron core and halo components, as well as a ``super halo'' extending to  100 keV. (from Lin et al. 1996)

The 3-D angular distributions show that the electrons in the main event are streaming outward from the Sun, with pitch-angle distribution highly peaked, within  30°, along the magnetic field. Figure 5 shows electron differential flux v. energy spectra at different times during, and integrated over the entire impulsive event, as well as the pre-event solar wind spectrum, with core, halo and ``super-halo'' (discussed later). The event spectra have the pre-event spectra subtracted. The peak in the primary event spectra progresses to lower energies with time, although contributions from the smaller events are also evident. The event-integrated spectrum displays a peak at  1 keV, with significant flux at ~0.5 keV. No electrons are detected at 422 eV or below for this impulsive event. Above the peak, the spectrum is similar to those reported previously above ~2 keV [Potter et al., 1980], and can be fit to a power-law shape dJ/dE = AE, where E is the electron energy in keV and A and  are constants. The best fit gives  =3.0 from ~1 keV to 40 keV, steepening to  =4.4 above ~40 keV.

Because at coronal temperatures electrons are not gravitationally bound while protons are, an ambipolar electric field (the Pannekoek-Rosseland field [Pannekoek, 1922; Rosseland, 1924]) is set up with a total potential drop of about 1 kV from the base of the corona to 1 AU. This potential varies inversely with distance from Sun center, and it accelerates protons outward and decelerates electrons. Thus, the peak in the spectrum of the electrons just as they escape the corona would be up to ~1 keV more than measured at 1 AU, e.g., ranging from  1 up to ~2 keV for the event of Figure 4 depending on the height of the acceleration.

The fact that the event spectrum extends down to such low energies indicates that at least some of the electron acceleration must occur high in the corona, since the range of ~ keV energy electrons in ionized hydrogen, due to Coulomb collisions, is short compared to the column depth through the corona. Assuming that the initial accelerated electron spectrum is power-law with the same exponent as seen at energies above the peak, the maximum overlying column density can be calculated [Lin 1974]. For a peak at ~1.5 keV, the column density must be less than ~9 x 1017 cm-2. This value implies that the lowest energy electrons must have been accelerated at altitudes of ~ 1 Rearth, for the typical active coronal density models derived from radio observations [Dulk and McLean, 1978], or ~ 0.2 Rearth for the quiet equatorial corona at sunspot minimum [Saito et al., 1977].

Even though the first year of the WIND observation, late 1994 to late 1995, is near solar minimum, tens of impulsive solar electron events have been seen, with many detected down to ~ 0.5 keV. The coronal flare acceleration process thus appears to produce a power-law spectrum extending down to  1.5 keV or lower, compared to a coronal thermal electron energy of kT ~ 0.1 keV (~ 106 K). Integrating over energy spectrum and over the duration of the event, and assuming that the cone of propagation for the electron event of Figure 4 is ~ 40 °, we estimate a total energy of  3 x 1026 ergs in escaping electrons. Thus, at least that much energy was released in the coronal flare process at the Sun.

Fig. 6. Electron differential flux spectrum from ~5 eV to  100 keV, measured at a very quiet time, in the absence of any solar particle events, by the Wind 3D Plasma and Energetic Particle experiment (Larson et al., 1996a). The diamonds, triangles, and squares indicate the three different detectors used to accommodate the wide range of fluxes over this energy range. The dashed lines give fits to Maxwellians for the solar wind core and halo.

QUIET TIME ELECTRONS

Unlike ISEE-3, the WIND spacecraft was launched near the minimum in the solar activity cycle. Thus, there are substantial periods free from energetic solar particle events and streams. Figure 6 shows the WIND omnidirectional electron spectrum from ~ 5 eV to ~ 102 keV measured during such a quiet period on February 22, 1995. Because of the extremely wide range of electron fluxes, three separate detectors--EESA-L, EESA-H, and SST--are required to measure this spectrum. The solar wind plasma Maxwellian core dominates from ~5 to ~50 eV; the solar wind halo takes over from ~ 102 eV to ~ 1 keV. The halo is believed to be due to the escape of coronal thermal electrons which have a temperature of ~ 106 K. Note, however, that the halo spectrum departs significantly from isothermal at energies above ~ 0.7 keV.

A third, much harder component is evident beginning above ~2 keV and extends to  102 keV, which we have denoted the ``super-halo'' [Larson et al., 1996a]. The spectrum of the ``super-halo'' appears to be approximately power-law with exponent  ~ 2.5. If this ``super-halo'' is solar in origin, (a more detailed analysis will be required to confirm this), it would imply that electrons of such energies must be continuously present at, and escaping from, the Sun.

It should be noted here that for exospheric models of the solar wind (Maksimovic et al., 1996) the presence of a significant non-thermal tail to the coronal electron population is sufficient by itself to accelerate the solar wind. The mechanism for producing this non-thermal population is unknown, but it is tempting to speculate that it is related to the mechanism which heats the corona (see, for example, Scudder, 1992). Further analysis and correlations with solar coronal observations are required to resolve the origin of these particles and their relationship to coronal heating.

Fig. 7. (a) The omnidirectional electron fluxes at 96 s resolution as measured by EESAH. The center energies are listed on the right. The fifteen minute period flux modulations are marked. (b) The omnidirectional electron fluxes as measured by the SST-Foil detector. Again, the center energies are listed on the right. (c) and (d) Spectrograms of the ratio (F/F0) of electron fluxes (F) to a reference flux (F0) determined by averaging the electron fluxes over the period ~00:00 UT to ~04:00 UT prior to the first event. The vertical axis represents energy while the darkness of shade represents electron flux. Superimposed on the plot are linear fits of the arrival time of the electrons versus their inverse velocity. The event start times and path lengths as determined by the linear fit are marked on the plot. (e), (f), and (g) Solar type III radio bursts as observed by the Wind Waves instrument (Bougeret et al., 1995). The RAD2 panel displays the radio emissions from 1 MHz to 14 MHz (linear frequency axis) with the darkness of shade representing power relative to cosmic background (logarithmic scale) that spans approximately a factor of thirty. The TNR (Thermal Noise Receiver) panel displays the frequency range from 4 kHz to 250 kHz. The grey scale is in absolute units of log(mV/m - Hz-1/2) as measured by antennae with ~50 m effective baseline. (h) The electric field wave power in the frequency band (19 kHz to 41.5 kHz) encompassing the local Langmuir frequency. (from Ergun et al. 1996)

ENERGETIC ELECTRON DISTRIBUTIONS AND PLASMA WAVE OBSERVATIONS FOR SOLAR TYPE III RADIO BURSTS

Four solar impulsive electron events were detected by WIND on April 2, 1995 (Ergun et al., 1996). Figure 7a (top panel) plots the omnidirectional electron fluxes at 96 s resolution as measured by EESAH. The center energies of the channels range from 140 eV to 27.1 keV, with the lowest energy traces on top. Immediately below (Figure 7b) are the omnidirectional electron fluxes as measured by the SST-Foil detector. The impulsive electron events can be clearly seen in the electron fluxes, with the strongest beginning at ~ 1110 UT with the 182 keV electrons and extending down in energy to ~ 600 eV.

A persistent feature in all of the events were ~ 15 minute period modulations of the electron fluxes (see Figure 7a) that appeared as the low-energy (~ 1 keV to ~ 4 keV) electrons arrived. Similar flux modulations can be seen in several other solar impulsive electron events observed by the WIND satellite and by the ISEE 3 satellite (see Lin et al., 1981, Figure 3). These modulations showed no measurable velocity dispersion, and the modulations of the second event are accompanied by modulations in the magnitude of the magnetic field that had a similar period, suggesting some type of hydromagnetic instability is occurring.

Figure 7c,d are spectrograms of the ratio (F/F0) of the event electron fluxes (F) to the pre-event flux (F0), determined by averaging from ~00:00 UT to ~04:00 UT. Superimposed on the spectrograms are linear fits of the arrival time of the electrons versus their inverse velocity. In all of the events, velocity dispersion of the electrons is consistent with the expected Archimedean spiral length (~ 1.15 AU) for the measured solar wind velocity (~ 370 km/s).

All four of the solar impulsive electron events were associated with solar type III radio bursts (Figure 7e,f,g) observed by the Wind Waves instrument (Bougeret et al., 1995). The radio bursts for events 3 and 4 extended up to 14 MHz, the maximum frequency of the Waves experiment. The lowest frequency radio electromagnetic emissions appeared to have a cut-off near the local second harmonic. The narrow-banded emissions that persisted through out the day varying between 20 kHz and 40 kHz (see Figure 7g) are locally generated Langmuir waves; the power in Langmuir emissions is shown in Figure 7h. The series of Langmuir bursts between ~11:45 UT and ~12:45 UT coincide with the arrival of ~ 12 keV through ~ 2 keV electron fluxes of event 3. The bursts at ~15:25 UT coincided with the arrival of ~ 6 keV electron fluxes of event 4. Figure 8 is an expanded view of Figure 7h between 11:45 UT and 12:40 UT. The nonlocal solar type III radio emissions contributed to the background power of ~ 1 µV/m. The local Langmuir bursts rise 1 to 2 orders of magnitude above the background, reaching ~ 50 µV/m.

Fig. 8. (a) An expanded view of the wave power near the local plasma frequency for April 2, 1995 (see Figure 7). The sharp peaks generally indicate local Langmuir emissions. The strongest series of wave emissions occurred from ~11:47 to ~11:55 UT (shaded). Below (b) are a series of 96 s average reduced parallel distribution functions fr(v||) on a log-log scale plotted in pairs. The top curve of each pair represents the +B direction, the bottom plot the -B direction. The vertical axis is the distribution function. The horizontal axis represents both velocity and time. The electron distributions are positioned such that the lowest velocity point (~ 7 times 103 km/s) indicates the beginning time of the 96 s average. A velocity scale for the first distribution is located below left. During the time of strongest wave emissions (shaded), the distribution functions were unstable or marginally stable (plateaued) while the later distributions were clearly stable. (from Ergun et al. 1996)

For this interval, the full 3-D electron distributions averaged over 96 sec, were reduced to 2-D distributions
f(v, v||) by assuming gyrotropy. The electron beam angular width was narrow, typically ~ 16° at 6 keV, similar to that reported for the ISEE 3 observations (Lin et al., 1981). Integrating over v, we obtain the reduced, one-dimensional, electron distribution functions, . The f(v||) are plotted in Figure 8 on a log-log scale in pairs. The horizontal axis represents both time and velocity, the velocity axis for the first distribution is immediately below. The electron distributions are positioned such that the lowest velocity point indicates the beginning time of the 96 second averaging period. The upper curve is in the +B direction (anti-sunward traveling electrons); the lower curve is the -B direction. At low velocities, the gap between the upper and lower traces reflects the anisotropic solar wind strahl electrons which carry heat flow away from the Sun. At higher velocities (>104 km/s), the separation between the two traces is due to the solar impulsive electron event which was already underway in the leftmost distribution.

During the Langmuir bursts from ~11:47 UT to ~11:55 UT (highlighted in grey) the reduced electron distributions appear to have been a marginally stable, or slightly unstable, at a beam velocity of ~ 6 times 104 km/s (10 keV). Throughout the rest of the period the reduced electron distributions were stable. The observed positive slopes at that parallel velocity are only marginally significant with

fr/v||  10-27 s2cm-5, corresponding to a growth rate of ~ 0.03 s-1, far less than the maximum growth rate of ~ 0.1 s-1 to ~ 1 s-1 seen in the ISEE 3 event reported by Lin et al. (1981). That event also had positive slopes that endured for long (>10 min) periods. Such strong, persisting positive slopes have not yet been seen by WIND.

The reduced, one-dimensional distribution functions were almost always stable ( fr/ v|| < 0). If kinetic relaxation from Langmuir waves dominated the electron evolution, the distributions are expected to have been plateaued, especially during periods of Langmuir emission. Velocity dispersion and adiabatic focusing (traveling into region of lower ) should continually act to destabilize the distributions. It is significant, then, that most of the distributions were stable with negative slopes. These observations indicate that, in addition to quasilinear relaxation, other processes may be contributing to the evolution of the electron beam. One possibility is that kinetic nonlinearities from oblique waves may play a role in electron evolution. Whistler waves have been reported to be associated with Langmuir emission in the interplanetary medium (Kennel et al., 1980; Thejappa, Wentzel, and Stone, 1995), and may play such a role here.

ACKNOWLEDGMENTS

This research was supported in part by NASA grant NAG5-2815.

REFERENCES

Bougeret, J.-L., M.L. Kaiser, P.J. Kellogg, R. Manning, K. Goetz, S.J. Monson, N. Monge, L. Friel, C.A. Meetre, C. Perche, L. Sitruk, and S. Hoang, Space Sci. Rev., 71, 231 (1995).

Dulk, G.A., and D.J. McLean, Coronal magnetic fields, Solar Physics, 57, 279 (1978).

Ergun, R.E., D. Larson, R.P. Lin, J.P. McFadden, C.W. Carlson, K.A. Anderson, L. Muschietti, M. McCarthy, G. Parks, H. Reme, J. M. Bosqued, C. d'Uston, T.R. Sanderson, K.P. Wenzel, M. Kaiser, R.P. Lepping, S.D. Bale, P. Kellogg, and J.-L. Bougeret, Wind spacecraft observations of solar impulsive electron events associated with solar type III radio bursts, Astrophys. J., in preparation (1996).

Feldman, W. C., J. R. Asbridge, S. J. Bame, M. D. Montgomery, and S. P. Gary, Solar wind electrons, J. Geophys. Res., 80, 4181 (1975).

Feldman, W. C., R. C. Anderson, S. J. Bame, S. P. Gary, J. T. Gosling, D. J. McComas, M. F. Thomsen, G. Paschmann, and M. M. Hoppe, Electron velocity distributions near the Earth's bow shock, J. Geophys. Res., 88, 96 (1983).

Fitzenreiter, R. J., J. D. Scudder, and A. J. Klimas, Three- dimensional analytical model for the spatial variation of the foreshock electron distribution function: Systematics and comparisons with ISEE observations, J. Geophys. Res., 95, 4155 (1990).

Kennel, C.F., F.L. Scarf, F.V. Coroniti, R.W. Fredericks, D.A. Gurnett, and E.J. Smith, Geophys. Res. Lett., 7, 129 (1980).

Larson, D.E., R.P. Lin, K.A. Anderson, S. Ashford, C.W. Carlson, D. Curtis, R.E. Ergun, J.P. McFadden, M. McCarthy, G.K. Parks, H. Reme, J. M. Bosqued, J. Coutelier, F. Cotin, C. d'Uston, K.P. Wenzel, T.R. Sanderson, J. Henrion, and J.C. Ronnet, Solar Wind 8 Conf. Proc., submitted (1996a).

Larson, D.E., R.P. Lin, J.P. McFadden, C.W. Carlson, K.A. Anderson, L. Muschietti, M. McCarthy, G. Parks, H. Reme, J. M. Bosqued, C. d'Uston, T.R. Sanderson, K.P. Wenzel, R.P. Lepping, Probing the Earth's bow shock with upstream electrons, Geophys. Res. Lett., 23, 2203 (1996b).

Larson, D.E., et al., this proceedings (1997).

Lepping, R. P. et al., The WIND magnetic field investigation, Space Sci. Rev., 71, 207 (1995).

Lin, R.P., Non-relativstic solar electrons, Space Sci. Rev., 16, 189 (1974).

Lin, R.P., D.W. Potter, D.A. Gurnett, and F.L. Scarf, Astrophys. J., 251, 364 (1981).

Lin, R.P., Electron beams and Langmuir turbulence in solar Type III radio bursts observed in the interplanetary medium, in Basic Plasma Processes on the Sun, ed. E. Priest and V. Krishan, p. 467, Internat'l Astron. Union, The Netherlands (1990).

Lin, R. P., et al., A three-dimensional plasma and energetic particle investigation for the Wind spacecraft, Space Sci. Rev., 71, 125 (1995).

Lin, R. P., Solar impulsive electron events, CESRA Proc., in preparation, (1996).

Maksimovic, M., V. Pierrard, and J. Lemaire, Astron. Astrophys., submitted (1996).

Pannekoek, A., Ionization in stellar atmospheres, Bull. Astron. Inst. Neth., 1, 107 (1922).

Phan, T.-D., et al., this proceedings (1997).

Potter, D.W., R.P. Lin and K. A. Anderson, Impulsive 2-10 keV solar electron events not associated with flares, Astrophys. J., 236, L97 (1980).

Rosseland, S., Electrical state of a star, Monthly Notices Royal Astron. Soc., 84, 720, (1924).

Saito, K., A.I. Poland, and R.H. Munro, A study of the background corona near solar minimum, Solar Phys., 55, 121 (1977).

Scudder, J.D., Astrophys. J., 398, 299, (1992).

Slavin, J. A., and R. E. Holzer, Solar wind flow about the terrestrial planets: 1. Modeling bow shock postition and shape, J. Geophys. Res., 86, 11401 (1981).

Thejappa, G., D.G. Wentzel, and R.G. Stone, J. Geophys. Res., 100, 3417 (1995).