Institute of Geophysics and Planetary Physics, University of California, Los Angeles
Los Angeles, CA, 90024-1567
Originally published in:
Adv. Space Res., 15, (4)5-(4)16, 1995.
The rudimentary measurements of the instruments on Pioneer Venus over a 14 year span have provided a strong framework for the interpretation of the observations with a more sophisticated instrument package during the Galileo Venus flyby in February 1990. In some cases the Galileo results provided independent confirmation of earlier inferences. In other cases completely new data were obtained. Nevertheless, because of limitations of the instruments and the trajectory and thermal restraints some outstanding questions were not addressed. Much has been learned but there is still much more to do.
The Pioneer Venus and Galileo missions provide two extremes of Venus studies. The Pioneer Venus Orbiter has circled Venus in an elliptical 24-hour orbit since December 5, 1978, executing over 5000 circuits before it entered the Venus atmosphere in October 1992. The Galileo mission obtained only a few hours of data on February 10, 1990 as Galileo flew by Venus at a distance at closest approach of 3.67 Venus radii planetocentric. The Pioneer Venus instruments were low mass, relatively simple instruments of 1970's vintage. The Galileo instruments were sophisticated instruments of modern design. Both missions provided important observations concerning the interaction of the solar wind with Venus.
In this paper we examine 5 aspects of the Pioneer Venus and Galileo observations. We will look at measurements of the solar wind, the backstreaming of charged particles from the bow shock, the waves that these backstreaming particles generate, and the nature of VLF waves in the night ionosphere. Two sources have been proposed for these latter waves: in-situ instabilities as part of the solar wind interaction or atmospheric generation by electrostatic discharges, i.e., lightning. Many of the Galileo results to be discussed can be found in articles in the special issue of Science magazine of September 27, 1991. The Pioneer Venus results are discussed at greater lengths in the special issue of Space Sciences Reviews entitled Venus Aeronomy /1/.
|Fig. 1. The interaction of the solar wind with a neutral atmosphere of an unmagnetized planet. The neutral atmosphere (upper left) is ionized by solar EUV and UV radiation (upper right). In the absence of a magnetic field in the solar wind plasma, the flow would be absorbed by the planet leaving a wake (lower left). A magnetized solar wind cannot penetrate the highly conducting ionosphere and is diverted around the planet forming a magnetic barrier, bow shock and magnetotail (lower right).|
As illustrated in Figure 1, the solar radiation falling on the dayside neutral atmosphere of Venus ionizes it. Some of this ionization is transported to the night side by convection. Other ionization on the nightside arises from the collision of energetic electrons with the atmosphere. The solar wind, if it were not magnetized, might interact with this atmosphere and be absorbed. However, the magnetized solar wind is deflected by the highly electrically conducting ionospheric plasma. Since the solar wind flows faster than the speed of compressional waves in the plasma, a bow shock is formed that slows the flow and deflects it about the planet. The velocity of the compressional wave in a magnetized plasma, the fast magnetosonic mode, is anisotropic. It travels faster perpendicular to the magnetic field than along it. This asymmetry is expected to produce a shock with an asymmetric cross section /2,3/.
|Fig. 2. The variation of the Venus bow shock location as measured at the terminators by Pioneer Venus. Also included are various measures of solar activity.|
The initial measurements of the location of the Venus bow shock by Pioneer Venus produced a surprising apparent contradiction of earlier Venera 9 and 10 measurement of the location of the Venus bow shock, but the continued monitoring through the solar cycle provided the answer to that apparent contradiction. Figure 2 shows the location of the Venus bow shock measured from 1979 to 1991 by Pioneer Venus together with various measures of the EUV output of the sun /3/. The initially observed location of the bow shock was far removed from that seen by Venera 9 and 10 but the position of the bow shock varies significantly as solar activity increases and decreases. The earlier Venera measurements of the Venus bow shock /2/ shown by the asterisk in Figure 2 were not at odds with those of Pioneer Venus. They were simply obtained when solar activity was low.
|Fig. 3. The Galileo trajectory past Venus on Feb 10, 1990 in solar-oriented cylindrical coordinates. The irregular curved line is the location of the bow shock predicted from a model whose cross section is fixed but whose orientation varies with the direction of the interplanetary magnetic field.|
|Fig. 4. The elliptical cross section of the bow shock found by Pioneer Venus. The dotted line shows the cross section for quasi-parallel shocks and the heavy line for quasi-perpendicular shocks /3/.|
When Galileo flew by Venus in 1990, the trajectory just grazed the bow shock at a distance from -10 to -5 Venus radii (Rv) behind the planet /4/. This is shown in Figure 3. The dashed lines show two models of the location of the bow shock for different orientations of the interplanetary magnetic field /5/. The jagged line gives the estimated location of the bow shock based on an empirical model, controlled by the IMF direction. Considering the variability of the position of the bow shock as solar activity varies, shown in Figure 2, the near coincidence of the Galileo flyby trajectory and the Venus bow shock sketched on Figure 3 is quite simple. As discussed above, the velocity of the compressioned wave in a magnetized plasma is anisotropic. Thus the cross section of the shock should be elliptical. That this is indeed seen is illustrated in Figure 4 which shows the observed cross-section of the bow shock according to PVO observations /3/. Thus, the Galileo data confirm both our expectations and the Pioneer Venus observations of an asymmetric shock.
|Fig. 5. Schematic drawing of the ion pickup process. Neutral atoms high in the atmosphere are ionized by solar radiation and then are accelerated into cycloidal paths by the solar wind electric field. Energy per charge flux spectra show pickup ions detected close to the planet and close to the magnetotail where the flow is much slower than in the solar wind.|
Although the solar wind interaction confines the Venus ionosphere to a relatively thin layer surrounding the planet as sketched in Figure 1, the neutral atmosphere is not so confined. The upper atmosphere of Venus in fact extends via a hot oxygen exosphere well out into the region of the solar wind interaction /6/. These atoms are subject to ionization by three processes: photoionization, charge exchange and impact ionization. When they become ionized they are accelerated by the electric field of the solar wind as illustrated in Figure 5 which shows sample ion trajectories and plots of energy per charge ion flux spectra observed near the planet /7/ and near the magnetotail tail boundary /8/. In these two regions at least, the ions can be detected as a second peak on the spectrum just below that upper cut off of the Pioneer Venus plasma analyzer. Unfortunately, the PVO plasma analyzer has limited capability for detecting these ions because of its energy range, sensitivity and data compression algorithm. It is optimized for detecting the solar wind. Farther from the planet the ions should be accelerated up to the solar wind velocity and have energies over 50 keV.
|Fig. 6. Stacked E/Q spectra versus time together with the simultaneously measured magnetic field strength and components showing that the pickup ions appear along the edge of the region identified as the magnetotail in the magnetic field measurements /9/.|
Despite its limitations, we have been able to obtain some information about the ion pick up process at Venus from this instrument /9,10/. Figure 6 shows the relationship of the observation of pick up ions down the tail of Venus to the entry into the tail. The ions are observed as the proton flow slows and the spacecraft enters the region identified as the tail proper from the magnetometer data. Moreover, the region of observed ion pick up is controlled by the direction of the interplanetary magnetic field and is asymmetrically distributed as expected because the pick up process itself is asymmetric as illustrated in Figure 5. This asymmetry has been mapped as shown in the bottom panel of Figure 7 /10/.
|Fig. 7. Schematic drawing of the gasdynamic model used to determine the ion pickup rate. The lower panel shows the contours of constant ion flux and magnetic tail cross section in the Z-Y plane perpendicular to the solar wind flow /10/.|
The model of the gas dynamic interaction originally developed by J. R. Spreiter and colleagues /11/ has been improved recently to include the effects of mass loading and the closure of the flow behind the planet. When the mass loading rate is adjusted to produce a distortion in the magnetic field similar to that observed a mass loading rate of about 2 x 1024 O+ ions per second is obtained /10/.
The extended energy range, the mass discrimination and the greater sensitivity of the Galileo instruments gave much promise for the detection and characterization of these pickup ions. However, the geometry of the Galileo trajectory and the constraints on the look directions of the Galileo ion detector in the intense thermal environment of Venus made the detection of any ions difficult and despite a careful search, none were seen /12/.
|Fig. 8. Venus foreshock geometry /19/.|
In front of the Earth's bow shock energetic charged particles are found streaming back toward the sun along field lines connected to the bow shock, called the foreshock as illustrated in Figure 8. These particles are thought to be the product of many different processes. Electrons are thought to be accelerated at the bow shock along the first tangent field line at the perpendicular shock by the "fast Fermi" mechanism where the electrons see a rapidly advancing magnetic mirror as the magnetic flux tube is carried through the shock. Electrons are also heated behind the shock and the hotter particles can escape upstream back into the solar wind. Ions too become heated at the shock and the hotter particles can escape upstream. They can be reflected by the electric and magnetic fields at the shock and return upstream and they can be accelerated as they gradient drift along the shock front. Irregularities in the solar wind, drifting toward the shock, can act as scattering centers for Fermi acceleration also. Thus, we expect for many reasons to see particles streaming back into the solar wind.
|Fig. 9. Evidence for pickup ions in the measurements of the PVO plasma analyzer /13/. The bottom 3 traces show the E/Q flux spectrum and the azimuthal and polar angle of the flow. The uppermost flux peak can be seen to come from a direction much different than that of the solar wind. The upper four panels show the interplanetary magnetic field and that the spacecraft remained in the undisturbed solar wind for the entire period.|
The only instrument on Pioneer Venus that could reasonably be expected to detect these particles directly was the plasma analyzer. Although as we will discuss in the next section, these particles could be detected indirectly by the waves generated in their upstream passage. Figure 9 shows upstream ions detected upstream of the bow shock on day 212 of 1979 /13/. The key to their identification is that they are at an energy higher than that of the solar wind helium flux and that the direction of arrival of the particles is quite different than that of the solar wind. Perhaps surprisingly, all upstream ions detected were behind the nose of the shock in the region labelled the parallel streaming foreshock in Figure 8. Thus the drifting particles moved largely with the solar wind flow. Further, these event are rare. Only 4 such events were detected in a survey of 65 orbits. This observation is in contradistinction to the observation that upstream waves are essentially just as frequent at Venus as at Earth. Thus the scarcity of upstreaming particles seems to be due to instrumental limitations and not to a fundamental difference in the solar wind interaction at Venus.
|Fig. 10. Three dimensional velocity distribution of ions upstreaming from the bow shock into the solar wind observed by the Galileo plasma analyzer /12/. The phase space densities are computed with the assumption that the ions are protons. The coordinate +V3 is parallel to B. The V2 axis is parallel to B x Vs where Vs is the antisolar direction.|
|Fig. 11. Comparison of the energetic ion count rates with the foreshock geometry parameters during the Galileo flyby /14/. Depth is the distance in front of (negative) or behind (positive) the tangent field that line the solar wind has travelled before reaching the spacecraft. The upper distance is the distance from the tangent point on the shock to the point where the solar wind crossed the tangent field line. The bottom two traces are the magnetic field strength and the angle between the magnetic field and the shock normal at the point where the magnetic field line through the spacecraft intersects the shock.|
The Galileo instruments confirmed this suspicion as they observed copious upstream particles with their greater sensitivity /12,14/. Figure 10 shows the 3-dimensional velocity distribution of ions upstreaming from the bow shock when the spacecraft was along field lines connected to the shock and well in front of it. Figure 11 shows the response of the energetic particle detector at 4 different ion energies from 22 keV to over 120 keV. Ions are seen when the magnetic field connects solidly to the bow shock at all energies. Since Venus has no radiation belts to act as a reservoir for energetic particles, these particles must have been accelerated either at the bow shock or in the upstream region. This is important for the acceleration of cosmic rays since some had called into question the efficiency of Fermi acceleration in enabling particles to be accelerated from solar wind energies of about 1 keV up to 100's of keV.
The Venus foreshock contains many of the same upstream wave phenomena as the terrestrial foreshock /15/. The geometry of the foreshock is illustrated in Figure 8. Low frequency waves with periods of about 30 seconds or longer are found in the region of the ion foreshock. Upstream whistler mode waves with periods of close to 1 sec are seen in both the ions and electron shocks where the whistler mode ray path can reach the spacecraft from the shock /16,17/. The former waves are believed to result from instabilities of the backstreaming ions are then carried back toward the shock by the solar wind. These waves are in general weaker than terrestrial waves, in part because the region for wave growth is smaller. The low frequency waves measured by the Galileo magnetometer /4/ were similar to those detected previously by the Pioneer Venus magnetometer /15/. The latter waves are believed to arise as an instability in the shock ramp and to propagate at velocities greater than that of the solar wind.
The Venus foreshock provides a good environment for studying both these wave phenomena. First, because the size of the obstacle and the bow shock are quite constant under typical solar wind conditions, the foreshock geometry is quite stable. Second, because the region of wave growth is restricted the low frequency waves do not grow to as large an amplitude. Thus linear and quasi-linear theories are more applicable in the Venus foreshock than in the terrestrial foreshock. This property has been exploited to show that the Vlasov kinetic dispersion relation is much more accurate in high beta plasmas than the Hall-MHD approach /18/.
|Fig. 12. The intensity of Langmuir oscillations (30 kHz) and downshifted oscillations (5.4 kHz) versus foreshock depth as measured by the Pioneer Venus plasma wave experiment. The ninth decile indicates the amplitude exceeded by 10% of the signals /19/.|
Electrons are found backstreaming throughout the ion foreshock but they also extend further into the incoming solar wind as illustrated in Figure 8 because their backstreaming velocity is greater than that of the ions. Hence, they are swept back less than the ions. The region to which only electrons have access is called the electron foreshock. The energy of the backstreaming electrons varies with location in the electron foreshock so that the electrons closest to the tangent field line are most energetic. This is reflected in the intensity of waves generated by these electrons. Figure 12 shows PVO measurements of the electric field intensity of Langmuir oscillations at the electron plasma frequency (solid lines) and electrostatic "down shifted" waves at lower frequencies (dashed lines) versus the distance the solar wind has travelled past the foreshock boundary, i.e., the foreshock depth /19/. The Langmuir waves are strongest at the tangent field line where the electrons are most beam-like. The downshifted waves are strongest deep in the ion foreshock.
The Galileo plasma wave instrument was much more sophisticated than the PVO plasma wave device. Instead of the 4 narrow band channels of PVO, the Galileo instrument had both a sweep frequency analyzer and a waveform capture mode. Langmuir emissions and downshifted emissions were both detected but in far greater detail than on Pioneer Venus /20/. These emissions have a considerable amount of fine structure, with large shifts both upward and downward in frequency by as much as 20 kHz, together with monochromatic packets lasting only a fraction of a second, strongly suggestive of soliton-like structure.
|Fig. 13. Altitude profile of the Venus atmosphere showing the cloud layers.|
Venus is completely covered with a thick deck of clouds. As illustrated in Figure 13, this cloud deck is at high altitudes and is very thick, about 17 km thick. The cloud particles probably consist of sulfuric acid and not water. Thus the clouds may be quite different from terrestrial clouds. Nevertheless, many observers felt there might be lightning in the Venus clouds and both Soviet and American spacecraft were instrumented to detect lightning. These efforts have been recently reviewed /21/ and we will not repeat these details here. The Soviet measurements were made principally on the Venera 11-14 landers and consisted of measurements of electromagnetic waves in the VLF range. They found copious signals that could be due to lightning. The American measurements were made principally with the Pioneer Venus VLF plasma wave instrument. It too found copious VLF signals at low altitudes on the nightside of Venus. Two types of signals were detected: signals which had characteristics of whistler node waves (polarization, direction of propagation and frequency) and waves that could not be whistler mode waves. The former waves appeared to propagate long distances both across the surface of the planet and with altitude. The latter waves seemed to be restricted in local time and altitude and hence appeared to be good tracers of the source of the waves. The former waves behaved as one would expect for electromagnetic waves: efficient propagation in the Earth-ionosphere wave guide and little attenuation in the ionosphere. Nevertheless, the interpretation of these waves as due to Venus lightning was not accepted by all researchers and much controversy ensued.
|Fig. 14. Local time profile of the occurrence rate of activity at VLF frequencies in the night ionosphere of Venus /21/. The solid line denotes waves below the electron gyrofrequency that may be whistler mode waves. The dashed line shows how often the magnetic field strength is greater than 15 nT. An interval is defined to be active if it contains more than one burst of noise in 30 seconds.|
Figure 14 shows the local time distribution of these two types of signals. The lower frequency waves seen at all nighttime hours is the signal we interpret as whistler mode, generated somewhere near dusk, possibly in the afternoon sector, but able to propagate far from the initiating atmospheric discharge. The higher frequency signals, above the local electron gyrofrequency, are thought not to be propagating signals but may represent the ionospheric response to the near-field of the cloud. The fall off in occurrence near the dusk terminator is most probably associated with attenuation in the Venus ionosphere so these data imply that the lightning discharges extend well into the dayside.
|Fig. 15. The spectrum of the nine impulsive events detected during the Galileo Venus flyby /20/. To be counted as an event the spectral density had to exceed the in-flight determined receiver noise level by more than 4 standard deviations.|
Pioneer Venus had limited frequency coverage. It could detect only a small portion of the VLF spectrum, and was insensitive to radio waves which could more readily pass through the Venus ionosphere. Galileo on the other hand, as we mentioned above, had a very comprehensive radio and plasma wave investigation. Despite its brief stay in the vicinity of Venus and its distant flyby, it still detected several radio frequency bursts whose signature was unlike anything else but lightning /20/. The spectrum formed by these bursts is shown in Figure 15. The waves have approximately the right strength, occurrence rate and spectrum to be consistent with a source of lightning similar to that on Earth. Thus, Galileo was able to confirm the earlier inferences about the existence of lightning on Venus. However, it was able to do little about refining our knowledge of the rate of occurrence of Venus lightning.
The above comparisons of the results of the Pioneer Venus and Galileo missions at Venus show that the observations were quite complementary. The long term monitoring with rather rudimentary instruments set the stage for the brief flyby with the more sophisticated package of the Galileo spacecraft. Pioneer Venus, for example, found that the bow shock position varied markedly over the solar cycle and that it has an elliptical cross section whose orientation was controlled by the interplanetary magnetic field. Galileo, meanwhile, serendipitously skimmed the bow shock moving in and out of the magnetosheath as the interplanetary magnetic field orientation rotated. Pioneer Venus could measure backstreaming ions only under special circumstances but the Galileo instruments had the sensitivity, energy range and look direction to detect them readily. Pioneer Venus had a very simple plasma detector that could detect the presence of the various plasma waves but could do little to properly characterize them. Galileo in contrast had a very comprehensive plasma wave package which give complete frequency coverage and provide snapshots of plasma wave waveforms which will enable detailed comparison with theory. However, Galileo did not approach Venus closely enough to provide data on all the wave phenomena detected by Pioneer Venus. The comprehensive plasma wave package was very important for the study of the occurrence of lightning. Although a strong case had been made from the data on earlier missions, the Galileo passage provided a completely new set of data, independent from the earlier data, but totally consistent with it.
However, some areas were not advanced much by the Galileo observations. Pioneer Venus had only limited capabilities for observing pick up ions. Galileo because of instrument viewing constraints and the geometry of the trajectory was not able to measure pickup ions at all. Pioneer Venus had used the magnetic measurements to perform an extensive survey of upstream waves. It was found that the Venus foreshock was an excellent laboratory for the study of waves and the comparison with theory. Moreover, the stability of the foreshock geometry allowed studies to be undertaken that were difficult at Earth. For example, it was clearly demonstrated that the 1 Hz upstream whistlers resulted from processes at the shock and were not due to in-situ instabilities. These studies all required an extensive database and the brief Galileo flyby added little in this area.
Finally, we note that the combination of Pioneer Venus and Galileo measurements has made a tremendous advance in our understanding of both how the solar wind interacts with Venus and whether lightning occurs in the clouds of Venus. Nevertheless, some problems remain to be addressed in much more detail. We do not know how much atmosphere is lost per second to the solar wind nor the composition of these ions. This is a critical question for the evolution of the Venus atmosphere. Second, we do not know the overall frequency of occurrence of Venus lightning and what happens in the dayside atmospheres. All of our orbital data is from the nightside of Venus. We trust that the exploration of Venus has not come to an end and that new opportunities will arise to settle these questions.
We would like to than J. G. Luhmann, R. J. Strangeway, D. Orlowski, G. Crawford, and C.-M. Ho for their assistance with the Pioneer Venus analysis and M. G.Kivelson, R. J. Walker, and K. Khurana for their advice on the Galileo observations. This work was supported by the National Aeronautics and Space Administration through research grant NAG2-501 and through the Jet Propulsion Laboratory grant JPL 958510.
1. C. T. Russell, Venus Aeronomy, Kluwer Academic, 1991. See also Space Science Rev. 55, 1-4 (1991).
2. S. A. Romanov, Asymmetry of the region of the solar wind interaction with Venus according to data of Venera 9 and 10, Kosmich. Issled. 16, 318 (1978).
3. C. T. Russell, E. Chou, J. G. Luhmann, P. Gazis, L. H. Brace, and W. R. Hoegy, Solar and interplanetary control of the location of the Venus bow shock, J. Geophys. Res. 93, 5461-5469 (1988).
4. M. G. Kivelson, C. F. Kennel, R. L. McPherron, C. T. Russell, D. J. Southwood, R. J. Walker, C. M. Hammond, K. K. Khurana, R. J. Strangeway and P. J. Coleman, Magnetic field studies of the solar wind interaction with Venus from the Galileo flyby, Science 253, 1518-1522 (1991).
5. K. K. Khurana and M. G. Kivelson, A variable cross section model of the bow shock of Venus, J. Geophys. Res. submitted (1992).
6. A. F. Nagy and T. E. Cravens, Hot oxygen atoms in the upper atmospheres of Venus and Mars, Geophys. Res. Lett. 15, 433 (1988).
7. J. D. Mihalov and A. Barnes, Evidence for the acceleration of ionospheric O+ in the magnetosheath of Venus, Geophys. Res.Lett. 8, 1277 (1981).
8. J. D. Mihalov and A. Barnes, The distant interplanetary wake of Venus: Plasma observations from Pioneer Venus, J. Geophys. Res. 87, 9045 (1982).
9. K. R. Moore, D. J. McComas, C. T. Russell and J. D. Mihalov, A statistical study of ions and magnetic fields in the Venus magnetotail, J. Geophys. Res. 95, 12005-12018 (1990).
10. K. R. Moore, D. J. McComas, C. T. Russell, S. S. Stahara and J. R. Spreiter, Gasdynamic modeling of the Venus magnetotail, J. Geophys. Res. 96, 5667-5681 (1991).
11. J. R. Spreiter, A. Summers and A. Y. Alkane, Hydromagnetic flow around the magnetosphere, Planet Space Sci. 14, 223 (1966).
12. L. A. Frank, W. R. Paterson, K. L. Ackerson, F. V. Coroniti and V. M. Vasyliunas, Plasma observations at Venus with Galileo, Science 253, 1528-1531 (1991).
13. K. R. Moore, D. J. McComas, C. T. Russell, and J. D. Mihalov, Suprathermal ions observed upstream of the Venus bow shock, J. Geophys. Res. 94, 3743-3748 (1989).
14. D. J. Williams, R. W. McEntire, S. M. Krimigis, E. C. Roelof, S. Jaskulek, B. Tossman, B. Wilken, W. Studemann, T. P. Armstrong, T. A. Fritz, L. J. Lanzerotti and J. G. Roederer, Energetic particles at Venus: Galileo results, Science 253, 1525-1528 (1991).
15. C. T. Russell, T. I. Gombosi, M. Horanyi, T. E. Cravens, and A. F. Nagy, Charge exchange in the magnetosheaths of Venus and Mars: A comparison, Geophys. Res. Lett. 10, 163-164 (1983).
16. D. S. Orlowski, G. K. Crawford and C. T. Russell, Upstream waves at Mercury, Venus and Earth: Comparison of the properties of one Hertz waves, Geophys. Res. Lett. 17, 2293-2296 (1990).
17. D. S. Orlowski and C. T. Russell, ULF waves upstream of the Venus bow shock: Properties of one-Hertz waves, J. Geophys. Res. 96, 11,271-11,282 (1991).
18. D. S. Orlowski, C. T. Russell, D. Krauss-Varban and N. Omidi, Critical test for Hall-MHD model: Application to low frequency waves at Venus, J. Geophys. Res. submitted (1992).
19. G. Crawford, R. J. Strangeway and C. T. Russell, VLF emissions in the Venus foreshock: Comparison with terrestrial observations, J. Geophys. Res. submitted (1992).
20. D. A. Gurnett, W. S. Kurth, A. Roux, R. Gendrin, C. F. Kennel and S. J. Bolton, Lightning and plasma wave observations from the Galileo flyby of Venus, Science 253, 1522-1525 (1992).
21. C. T. Russell, Venus lightning, in Venus Aeronomy, 317-356, Kluwer Academic Publishers, Dordrecht, 1991; see also Space Sci. Rev. 55, 317-356 (1991).