Coronal mass ejections, magnetic clouds, and relativistic magnetospheric electron events: ISTP

The role of high-speed solar wind streams in driving relativistic electron acceleration within the Earth's magnetosphere during solar activity minimum conditions has been well documented. The rising phase of the new solar activity cycle (cycle 23) commenced in 1996, and there have recently been a number of coronal mass ejections (CMEs) and related "magnetic louds" at 1 AU. As these CME/cloud systems interact with the Earth's magnetosphere, some events produce substantial enhancements in the magnetospheric energetic particle population while others do not. This paper compares and contrasts relativistic electron signatures observed by the POLAR, SAMPEX, Highly Elliptical Orbit, and geostationary orbit spacecraft during two magnetic cloud events: May 27-29, 1996, and January 10-11, 1997. Sequences were observed in each case in which the interplanetary magnetic field was first strongly southward and then rotated northward. In both cases, there were large solar wind density enhancements oward the end of the cloud passage at 1 AU. Strong energetic electron acceleration was observed in the January event, but not in the May event. The relative geoeffectiveness for these two cases is assessed, and it is concluded that large induced electric fields (9B/9t) caused in situ acceleration of electrons throughout the outer radiation zone during the January 1997 event.


Introduction
The Sun is increasingly likely to expel large clouds of material (called coronal mass ejections (CMEs)) during the peak of its 11-year activity cycle. These can move outward from the Sun with speeds in excess of 1000 krn/s [Kahler, 1992]. The shock waves preceding such plasma structures can accelerate interplanetary particles to high energies, sometimes up to several hundred million electron volts (MeV). If the shock waves and "magnetic clouds" associated with CMEs strike the Earth's magnetosphere, they can initiate major geomagnetic storms that can disrupt power systems, degrade communication links, and increase the probability of anomalous behavior of operational spacecraft on which our society increasingly relies [e.g.,  The International Solar-Terrestrial Physics (ISTP) program has put into place a large array of spacecraft and ground facilities for studying the space environment [e.g., Acura et al., 1995]. The observational resources offered by ISTP attained a rather complete configuration with the launch of POLAR in February 1996. The Sun reached its minimum sunspot activity in mid-1996 and began to exhibit characteristics of the new activity cycle (E. Hildner, private communication, 1997). Consequently, several CMEs have been observed on the Sun which subsequently reached the Earth's vicinity. In this paper, we examine two such CMEs and we consider the "geoeffectiveness" of their interaction with Earth's magnetosphere. In particular, the efficiency of relativistic electron acceleration in the outer magnetosphere due to the CMEs is reported here. Such acceleration is assessed in the context of another mode of solar wind-magnetosphere coupling which has been shown to be quite effective at electron acceleration, namely, high-speed solar wind stream events [Baker et al., 1994;Blake et al., 1997;Li et al., 1997a].

Recurrent Geomagnetic Storms
High-speed solar wind streams originate in solar coronal holes [Feldman et al., 1978]. Generally, transequatorial coronal holes are well developed in the declining phase of the solar cycle (rather than right at sunspot minimum). Longterm observations in the outer magnetosphere (at L-6.6) have demonstrated that energetic electron fluxes are strongly modulated by solar wind streams [Baker et al., 1986]. Lowerenergy (5 300 keV) particle fluxes track the solar wind variations quite closely and, as illustrated by Figure 1, appear to be the direct product of magnetospheric substorm activity [see Baker et al., 1978;Nishida, 1983, and references therein]. Higher-energy (_> 300 keV)particle fluxes in the outer trapping regions (L=4~7) are also modulated by the solar wind streams, but peak fluxes are typically delayed relative to substorm-related enhancements (see Figure 1). The highest-energy magnetospheric electrons show strong recurrence tendencies at the 27-day rotation period of the Sun [Williams, 1966;Paulikas and Blake, 1979] and are closely related to "recurrent" geomagnetic storms [Baker et al., 1997a].
On the basis of recent observations, it is concluded that even during near-minimum solar conditions (sunspot minimum) there are discernible coronal source regions and resultant solar wind streams which can produce intense magnetospheric particle acceleration. other large electron counting rate enhancement in the SAMPEX data which will be discussed here.

High-Energy Electron Acceleration
Observations as in Plate 1 indicate that relativistic electrons increase in measured counting rate, often by a factor of 10 or more, throughout much of the outer magnetosphere on a timescale of order 1 day even during rather weak recurrent geomagnetic storm activity [Baker et al., 1997a;Blake et al., 1997]. Abrupt count rate enhancements occur in the outer magnetosphere, as well as at low L shells, deep within the magnetospheric cavity. Geomagnetic activity is controlled by the solar wind speed and by the interplanetary magnetic field (IMF) orientation [e.g., Nishida, 1983]. The combination of high Vsw and strong (southward) IMF drives intense substorm activity. This produces copious quantities of energetic electrons in the energy range 30~300 keV [Baker et al., 1978]. As a second step in the electron acceleration sequence, Baker et al. [1997b] noted that the highest-energy electrons and the hardest spectra occur some days after the substorm activity peaks. Thus the outer zone magnetospheric electrons behave in a quite coherent way. This coherence is manifested spatially, temporally, and spectrally. On the basis of previous observations, an acceleration sequence can be summarized as in Figure 2. The solid line shows the spectrally soft enhancement which is a prompt response to an increase of Vsw. The dashed line shows the spectral hardening as the solar wind speed decreases. The intense relativistic electron population appears some days after the leading edge of the solar wind stream has passed the Earth. Thus the highest relativistic electron flux is seen throughout the outer zone when Vsw is decreasing.
Although substorm disturbances provide an important seed population of energetic electrons needed for major relativistic electron events [Baker et al., 1997b], it apparently is the further extraction of energy from the solar wind stream as it buffets the magnetosphere that ultimately produces the higher-energy electron population [see Blake et al., 1997]. The mechanism that takes the lower-energy seed population and converts many of these subrelativistic electrons to highly relativistic particles has not yet been fully understood [see Li et al., 1997b).

CME-Generated Storms
In contrast to recurrent storms, large nonrecurrent geomagnetic storms develop at Earth as a result of such aperiodic solar events as the CMEs discussed above which are normally associated with eruptive prominences and other solar disturbances. It is sometimes, but certainly not always, found that the leading edges of CMEs are moving outward from the Sun at a speed much higher than that of the normal solar wind [see Burlaga, 1995]. Thus such fast CMEs can move rapidly through the ambient plasma of the heliosphere. Their outward motion can lead to great distortion of the IMF and, given a sufficiently high relative speed, CMEs can be drivers for strong interplanetary shock waves. The field compression and draping ahead of the magnetic clouds caused by CMEs often leads to stronger than normal magnetic fields at the leading edge of the structure and the high plasma flow velocity can produce a geomagnetic storm and particle acceleration due to a large magnetospheric field compression and distortion. CME events observed recently have not had particularly high speeds. However, when compared to recurrent solar wind stream events that have characterized the solar minimum, these recent CME/cloud events provide a contrasting view of how magnetospheric electrons may be rapidly accelerated.

CME/Magnetic Cloud Observations
Data are presented here for two magnetic cloud (CMErelated) events. The interplanetary measurements and the magnetospheric responses will be compared and contrasted using a wide range of data sets. Despite the apparent similarity of the two cases, their effectiveness in accelerating highenergy electrons is found to be strikingly different.

3.1
Case 1  the POLAR outbound pass toward the right in Plate 3 occurred during the time (-1400 UT)that the leading portion of the magnetic cloud passed over the Earth (see Figure 3a). The IMF B z was turning southward and substorm activity was strong. Large fluxes of relatively low energy electrons were seen in the outer zone (L=4-7). Trapped ("pancake") pitch disturbances were indicated by the high intensities of electrons (broadly) seen at roll angles near 0 ø and 180 ø. The large "loss cones" are evident for the trapped particle distributions and show up as diagonal bands of depleted flux.
These flux minima trace out the local magnetic field direction, as seen by the POLAR magnetometer (data not shown here). The moderate-energy electrons seen in Plate 3 were apparently produced, in part, by the substorm activity discussed above for this case, but there was clearly a population of energetic electrons present throughout the outer zone prior to the cloud's arrival (i.e., from 0930 to -1300 UT).
The flux versus L for various energy ranges for the relevant POLAR passes through the radiation belts can be compared as a function of time. Figure 6a shows HIST channel 10 (1.9-2.9 MeV)electron differential fluxes versus L for cuts that occurred from May 26 (-0200-1800 UT) through May 29 (-1800 UT). Each succeeding cut is offset by a factor of 10 from the previous pass in order to allow clear plotting of each profile. Each profile has the day/UT of the start of the L cut. It is evident from the data in Figure 6a that the magnetic cloud interval (1400 UT on May 27 to 1200 UT on May 29) did not change the radial flux profiles dramatically compared to the "precloud" levels (although some modest changes were seen orbit to orbit). The last cut on May 29 showed quite a drop in flux at the inner edge of the outer belt (L <_ 3.5).

3.2
Case 2. January 10-11, 1997    . It is seen that on January 8 and 9 there were very weak and narrow radiation belt projections at the SAMPEX (600-km) altitude. However, on January 10, with the magnetic cloud's arrival, the outer radiation zone projection was much broader and more intense, Note that the "collar" of bright red around the northern pole in Plate 2f was broader, more intense, and more azimuthally complete than for any of the May 1996 days. Hence the radiation belts became very elevated in the January case. The abruptness and the strength of the radiation belt acceleration at high (near-equatorial) altitudes is clearly demonstrated in Figure 6b. These flux versus L profiles for the POLAR/HIST sensor (similar to Figure 6a), show the immense difference in measured peak intensity and outer belt width before, and after, the magnetic cloud reached the Earth: Prior to -0800 UT on January 10, peak fluxes for 1.9-2.9 MeV electrons were at 3 _< L <_ 4 and were _<103 electrons (cm2-s-sr-MeV) -•. By -0200 UT on January 11, the peak intensity was at L > 4 and the peak flux was -5 x 105 electrons (cm2-s-sr-MeV) -•. Note also that relativistic electrons were near counting backgrounds beyond L--5 prior to the cloud passage but were extremely high even at L _> 6 after the cloud passage.
The speed and global coherence of the electron acceleration in the January event is shown in Plate 4. The top panel shows the E>3 MeV electron counting rates at various L values for the SAMPEX sensors for the period January 1-30, 1997. There were essentially no detectable electrons in this energy range prior to January 10. As the magnetic cloud reached Earth, the electrons jumped up in flux at a broad range of L values (3.5 _< L _< 5) almost simultaneously. The lower panel shows line plots of POLAR/HIST, Highly-Elliptical Orbit (HEO), and LANL (geostationary orbit) data in comparable energy ranges for the same time period. The HEO spacecraft [see Blake et al., 1997] is in a 60 øinclination orbit with 7 R E apogee; it makes two cuts through the radiation belts every 12 hours. All of the measurements show similar time behavior thus emphasizing the global, coherent electron acceleration that occurred.

Relativistic Electron Acceleration Mechanism
As discussed by Rostoker et al. [1995a], large-amplitude magnetic fluctuations throughout the magnetosphere imply a global wave field in which dB/dt is large. Such an induced electric field could be effective in accelerating electrons throughout the outer zone. Moreover, this could go on concurrently and continually as long as the wave field was present. The level and duration of large-amplitude, low-frequency wave activity for the January period was strikingly greater than for the May period. This is illustrated in Figure 7 where the high-pass filtered data from several CANOPUS magnetometer stations are shown for May 27 and for January 10. The amplitudes of the field fluctuations are commonly a factor of 5 greater in the January case. The largest amplitude waves seen by CANOPUS on January 10 occurred between ~1000 and -•1130 UT. This was the approximate time that SAMPEX and POLAR saw the relativistic electrons suddenly increase in flux (see Reeves et al., 1998). A significant difference between the May 1996 and the January 1997 cloud periods was the level of substorm activity produced within the magnetosphere as the leading edges of the magnetic clouds passed the Earth. This is clearly shown by the very different levels of AL in the two cases. As also shown by Figures 4a and 4b, the substantially different substorm activity levels changed very much the intensities and time variabilities of low-to-moderate energy electrons at geostationary orbit. The available quantity of such electrons in the outer magnetosphere could play an important role in determining the ultimate flux of relativistic electrons that are produced. In the picture presented by Baker et al. [ 1997b], there is a two-step process: First, substorm activity generates a low-energy "seed population" and then, in a second step, some portion of these substorm generated electrons are further energized (see Figure 2). The new and compelling evidence from the present analysis is that the second step of the acceleration is closely associated with large amplitude, low-frequency waves. As also suggested by the work of Blake et al. [1997], all the solar wind features must be in play (large, southward IMF and relatively high speed solar wind flow) to get relativistic electron acceleration.

Discussion
We have shown in this paper two magnetic clouds that, superficially, seem quite similar. However, when examined in detail, the January 1997 cloud event was much more effective at accelerating high-energy electrons. The January event led to much stronger substorm activity and ring current development and it also produced much stronger global fluctuations in the geomagnetic field. These fluctuations in B would imply a large-scale induced electric field which could accelerate further the electrons produced initially by the strong substorm events that occurred as the leading edge of the magnetic cloud passed the Earth. However, the details of how such in situ acceleration can occur have still to be worked out. Figure 8 shows a schematic summary of our inferences concerning the acceleration process. For the January 1997 period, the radiation belts were very weak prior to the magnetic cloud arrival. The relativistic flux levels were low and only a small range of L values was populated substantially. When the January magnetic cloud struck the magnetosphere, there was strong substorm activity associated with the leading edge (in which the IMF was strongly southward). As shown in Figure 8b, a high-density population of magnetospheric electrons was produced by the substorm activity and this population rapidly diffused inward and was further accelerated by the strong induced electric fields of the lowfrequency waves. Finally, and on quite a short timescale, the entire magnetosphere was filled with high-energy electrons. The scenario shown in Figure 8 is not fundamentally different from the one which has previously been discussed for recurrent storms associated with high-speed solar wind streams (see Figure 2). However, the January CME event produced higher-energy electrons much more quickly than typically seen in stream-associated events. This suggests that the January cloud somehow was more efficient at relativistic electron production than are normal high-speed streams.
As a first step to analyze the efficiency question, one can easily integrate the energy input (oe)parameter discussed above for the two cloud cases. Doing this for the entire cloud interval on May 27-29, one gets loedt~2.0x1016j. Limiting the calculation to only the southward IMF period, this integral equals-1.5x1016j. For the January case, the integral for the entire cloud is ledt-7.0x1016j, while for the southward IMF interval the integral is ~5.5x1016j for January 10. Thus, perhaps a bit surprisingly, the January cloud was only a factor of about 3 to 4 "stronger" in total electromagnetic energy input than the May cloud.
An obvious further issue concerning the outer zone electron population is what fraction of the total magnetospheric input energy gets converted to relativistic electrons. One can make a rough calculation of this "efficiency factor" as follows: Let us assume for calculational convenience that the outer zone is actually a toms of circular cross section. Then the cross-sectional area is given by Az=n rr2, where r r is the toms (minor) radius. The volume (V T) of this toms is roughly 2n r B A T, where rB is the central (major) toms radial dimensi6n. Thus the volume of the assumed belt of particles is V r =2•; 2 rr2 r B.
The geometrical quantities for the January cloud event can be estimated from the available particle measurements. Be- Taking the integral of oe as -5x1016j from above, this works out to be *li• -3.2 x 10 -4. Hence the electron acceleration efficiency was-0.03% in the January case. Notably, of course, the May cloud event must have had a very much lower (nearly negligible) acceleration efficiency.

Summary and Conclusions
This paper has compared and contrasted two magnetic cloud events observed by a large number of ISTP and affiliated spacecraft. It has also used a variety of ground-based data sets to assess the nature of the cloud's interactions with the magnetosphere. It was found that the May 1996 magnetic cloud was not effective in accelerating high-energy electrons in the outer radiation zone. On the other hand, the magnetic cloud event in early January 1997 was much more effective in producing large flux enhancements of relativistic electrons throughout the outer zone (L _> 3.5). We observed such large differences in "geoeffectiveness" despite the superficial similarity of the solar wind/IMF properties within the two clouds.
We conclude that the January cloud had the requisite features of strong southward IMF and high enough solar wind speed to accelerate high-energy electrons. The high-density spike following the cloud may also have played an important role for the later, long-duration enhancement in January 1997. The combination of interplanetary drivers is generally the same as those which result in effective electron acceleration during high-speed solar wind stream events. Thus any solar disturbance that produces suitable solar wind "input" conditions at 1 AU, whether from coronal holes, or from CMEs, can be very effective at enhancing the Earth's radiation belts. In particular, events that drive strong, lowfrequency wave activity clearly can accelerate electrons with BAKER ET AL.: MAGNE'IIC CLOUDS COMPARED 17,291 remarkable speed and efficiency [Rostoker et al., 1997]. This suggests that electron acceleration to high energies might have similar causes in other (e.g., solar or astrophysical) contexts.