At approximately 2200 UT on 19 May 1998, the PanAmSat Corporation's Galaxy 4 spacecraft experienced a failure in its attitude control system (Automatic Pointing Control, APC). Unfortunately, the backup system also had failed, either at that same time or earlier, so that the operators were unable to maintain stable Earth-link [Space News, 25-31 May 1998, p. 3]. The Galaxy 4 spacecraft is a heavily used communication satellite at geostationary orbit; its sudden failure caused the loss of pager service to some 45 million customers as well as numerous other communication outages [USA Today, p. 1, 21 May 1998].

Analysis by PanAmSat and Hughes continues as to the exact cause of the Galaxy 4 failure [Space News, ibid., p. 18]. Whenever there is an operational problem with space hardware, it is advisable to examine as broadly as possible the space environmental conditions prior to and at the time of the problem. Using a wide array of space data sets, we have analyzed the magnetospheric and solar wind conditions during the April-May 1998 period. There was large solar and magnetic activity in early May, but geomagnetic conditions were very quiet by 19 May. Moreover, the Galaxy 4 failure occurred when the satellite was in the afternoon local time sector. Thus, surface charging of the spacecraft was not a likely factor. However, strong evidence is found that highly relativistic electron [HRE] fluxes were substantially elevated above average conditions for a period of about two weeks prior to the 19 May failure of Galaxy 4.

Long-duration HRE enhancements have in the past been convincingly associated with spacecraft operational failure [e.g., Baker et al., 1994, 1996]. In considering an association of spacecraft failures with the relativistic magnetospheric electron population, it is worthwhile to recall the mechanism that would be at play. As shown in previous work [e.g., Vampola, 1987; Baker et al., 1987], high fluxes of energetic electrons can lead to "deep dielectric" charging of spacecraft subsystems. In this process, energetic electrons bury themselves in dielectric material such as thermal control blankets, electronic boards, coaxial cables, insulation, etc. Eventually, if the buried charge builds up more rapidly (due to continued irradiation) than it leaks away (due to low dielectric conductivity) then there can be an electrostatic discharge event (ESD) [Vampola, 1987]. Such a discharge can damage or destroy a sensitive circuit or subsystem and the result can be a spacecraft failure. It is therefore relevant to determine the characteristics of the space environment around the time of the Galaxy 4 failure.

It should be noted that whether the incident on 19 May was, or was not, due to a "space weather" effect, it nonetheless shows the vulnerability of our society to individual spacecraft failure. The vast number of users affected by the loss of just one spacecraft shows how dependent our society is on space technology and it shows how fragile our communication systems can be [Newsweek, p. 45, 1 June 1998]. The Galaxy 4 failure had a large impact because the spacecraft was optimally located over the central U.S. and could best handle digital pager signals. Therefore, 80% of all pager traffic was directed through it. Increasingly, phones, TV, radio, bank transactions, newspapers, credit card systems, etc., all depend upon satellites for some part of their link rather than being all ground-based. It seems very inadvisable to have such complex, societally significant systems susceptible to single-spacecraft failures. This seems particularly true as we are approaching the peak of the 11-year solar activity cycle.

The typical major communication spacecraft has an estimated value of $200-250 million. There are over one hundred such spacecraft in operation today and whole new constellations of low-to mid-altitude satellites are being emplaced at this very moment. The invested cost of space assets is truly staggering. As we approach the next solar activity maximum (in 2000-2001), it will be most important to see how robust these assets really are. Given the recent record of space environmental disturbances, we would expect many more highly disruptive spacecraft failures. We would suggest that space systems must be made highly immune to the space environment and backup systems must be transparently available to cover space system failure whatever their cause.

A rather detailed view of the magnetospheric particle population is obtained from the Solar, Anomalous, and Magnetospheric Particle Explorer (SAMPEX) spacecraft which is in a high-inclination (82°), low-altitude (~600 km) orbit. As such, it samples magnetic field lines threading essentially the entire magnetosphere every ~100 min [see Baker et al., 1994 and references therein]. SAMPEX carries sensor systems capable of measuring very energetic ions and electrons of both solar and magnetospheric origin. Figure 1a is a plot of particle fluxes measured by a large array of solid-state detectors (Heavy-Ion Large Telescope, HILT) onboard SAMPEX. The HILT channel used here has an electron energy threshold of 3 MeV; this channel also has sensitivity to >5 MeV protons (which is a point especially important to bear in mind during solar particle events and in the inner radiation belt).

Figure 1. (a) Color-coded SAMPEX electron fluxes measured for energies >3 MeV during April-May 1998. Data are plotted for L-values ranging from 1 to 8 versus Day of Year for 1998. (b) A plot similar to (a) but for the POLAR spacecraft. Both (a) and (b) show a major relativistic electron enhancement on DOY 124 (4 May). Times of spacecraft operational anomalies are indicated (1, 6, and 19 May) as discussed in the text.

The format of Figure 1a is such that particle flux is plotted according to the color bar to the right of the figure. The vertical axis of the figure is the McIlwain L- parameter which indicates the magnetic field line position: L is the distance in Earth radii (1 RE=6372 km) that a given magnetic dipole field line would cross the equatorial plane. The horizontal axis of Figure 1a is Day of Year (DOY) for 1998. SAMPEX data have been binned in 0.1-L intervals on a daily basis and are plotted from 1 April (DOY=91) to 31 May (DOY=151).

The data in Figure 1a show rather quiet conditions from DOY 91 to DOY 110. There was a modest, variable flux of electrons with E>3 MeV in the outer radiation zone between L~- 3.5 and L~- 6.0 during that interval. The "slot" region between L~2.2 and L~3.5 was quite evident with typically very low flux levels. The inner zone at L ~< 2.5 was also seen quite clearly in the sense that the HILT detectors were responding mostly to high energy ions in that part of the magnetosphere.

The nature of the radiation environment at low-altitudes changed dramatically on DOY 110. A solar energetic particle (SEP) event commenced on that day and persisted for the subsequent week or so. Very energetic ions (also identified unambiguously by other SAMPEX sensors) were seen with high intensities at L~> 5 and extending throughout the polar cap region. On DOY 115 (25 April) there was an abrupt increase in HRE fluxes deep in the outer zone peaking near L=4.0. The electron fluxes remained high for the subsequent several days. There then occurred another sequence of solar energetic particle events which were evidenced sporadically until at least DOY 130 (10 May).

Perhaps the most striking and notable event in this entire interval occurred on DOY 124 (4 May). On that day, there was a huge increase of the flux of HREs very deep in the magnetosphere (L~< 3). The slot region was filled and another radiation belt feature appeared at L~- 2.2 ± 0.2 . The relativistic electrons remained high throughout the outer zone for at least the subsequent two weeks. Electrons were seen to fill a broader region from L=3 to beyond L=5 over the next several day interval. The relativistic electron enhancement seen in Figure 1a was as intense, long-lasting, and spectrally hard as any event seen in the magnetosphere over the past several years.

Another view of the magnetosphere is offered by the POLAR spacecraft which is in a high-inclination, highly eccentric orbit (1.8 x 9 RE). As such, it cuts through the Earth's outer radiation zone at much higher altitudes than does SAMPEX. The POLAR payload includes the High-Sensitivity Telescope (HIST) which measures very energetic (E>0.3 MeV) electrons [Blake et al., 1995]. POLAR obtains high- resolution views of the outer electron belt every 17.5 hrs (the POLAR orbital period) on one inbound and one outbound pass. This allows assembly of L vs. time plots for POLAR that are closely analogous to those acquired with higher time resolution (but more limited pitch angle coverage) by SAMPEX.

Figure 1b shows POLAR/HIST data for electrons with E>2 MeV. The format is nearly identical to that for Figure 1a: The flux values are plotted as a function of L and time for DOY 91 (1 April) to DOY 151 (31 May). As is evident, the POLAR data show many of the same features that were present in the SAMPEX data. Unfortunately, POLAR does not survey the region below L~2.5, so we cannot use POLAR in the inner part of the magnetosphere. Where SAMPEX and POLAR L- values do overlap, the data illustrate the global, coherent nature of HRE events and confirms that the magnetosphere was populated by a very intense flux of relativistic electrons from early May until the end of the month.

Figure 2. Electron fluences (cm-2) for 14-day summing periods from January 1997 through May 1998 measured by Spacecraft 1994-084 at geostationary orbit. Fluence values are normalized to peak values (which all occurred in May 1998) for the three energy ranges shown.

Figure 2 provides a broad, recent history of the energetic electron fluxes at geostationary orbit. The plot shows Los Alamos National Laboratory measurements (running sums) of fluences (electrons/cm2) for 14-day intervals beginning in January 1997 through May 1998. The two-week values are normalized to the peak fluence observed in each of three energy channels: 0.7-1.8 MeV, 1.8-3.5 MeV, and 3.5- 6.0 MeV. Figure 2 clearly shows that the fluences in each energy range peaked on or just before 19 May 1998. Fluences at all other times were less than 75% of the value in mid-May. For the 3.5-6.0 MeV channel the fluences for all other times were never greater than ~50% of the value seen in May 1998. This includes the period in January 1997 when the Telstar 401 spacecraft failed [Reeves et al., 1998]. The May 1998 was the longest duration and spectrally hardest electron event seen in the past three years.

Figure 3. (a) GOES daily flux values of electrons with E>2 MeV for the period 21 April to 20 May 1998. Dates of various spacecraft operational problems are noted including the Galaxy 4 failures on 19 May. (b) Similar to (a) but for protons with E>100 MeV [courtesy of NOAA].

The GOES series of spacecraft operated by the National Oceanic and Atmospheric Administration (NOAA) measure the space environment continuously at geostationary orbit (GEO). Figure 3a shows the daily average flux (electrons/cm2-sr- day) of E>2 MeV electrons from 21 April to 20 May 1998. It is seen that the electron flux was low (~104/cm2-sr-day) on 21 April, but the flux then rose progressively over the subsequent week or so, reaching a flux maximum on 29 April. The electron intensities then were lower for several days (1-4 May). The average electron intensity jumped up by about two orders of magnitude on 5 May and stayed high for the subsequent 10 days. On 16 May the electron flux diminished by a factor of 2-3, but it remained well above 107 until the end of the plotting sequence.

It is important to be aware not only of the GEO electron environment, but also of the energetic ions. Figure 3b shows the time history of the flux of protons (E>100 MeV) measured by the GOES sensors. The period covered is identical to that for electrons in Figure 3a. The data in Figure 3b show several discrete proton enhancements (consistent with those detected at high L-values by SAMPEX, as shown in Figure 2a): The main proton events were seen 21 April, on 2 May, and again on 6 May.

As noted in the Introduction, the most significant spacecraft operational anomaly of concern during the period under investigation was the Galaxy 4 failure on 19 May. This time has been annotated with arrows on Figures 1, 2, and 3. It is evident that the 19 May Galaxy failure occurred after a very long period of elevated HRE fluxes, but was not associated with any high-energy solar particle events.

Other spacecraft problems of note during the period under discussion included a major POLAR anomaly on 6 May and the probable failure of the Equator-S spacecraft on 1 May 1998. The POLAR event consisted of a loss of about six hours of data due to a processor problem onboard the spacecraft. This was almost certainly due to a single-event upset (SEU) of the type caused by ion impacts on the spacecraft memory [Guit, 1998]. We note that the solar particle event on 6 May was characterized not only by E>100 MeV protons, but also by remarkably high fluxes of very energetic iron nuclei (Richard Mewaldt, private communication, 1998) as seen by the Advanced Composition Explorer (ACE). The POLAR anomaly occurred within the first moments of the appearance of the very energetic solar particles (J.B. Blake, private communication, 1998).

The Equator-S failure on 1 May consisted of a loss of the spacecraft central processor [Max Planck website: www.mpe.mpg.de].The failed processor was actually the backup unit since the primary processor had failed in December 1997 (G. Paschmann, private communication, 1998). We have not analyzed the Equator-S situation in detail yet, but we note that the failure occurred after a week or more of elevated relativistic electron fluxes throughout the outer radiation zone (Figure 1).

What solar and solar wind conditions led to the remarkable magnetospheric particle events we have identified here? The International Solar Terrestrial Physics (ISTP) constellation of spacecraft and the related scientific and operational platforms available today give us an unprecedented view of this matter. The Solar Heliospheric Observatory (SOHO) showed a large coronal mass ejection (CME) on 2 May 1998. SOHO and GOES sensors also observed a series of large solar flares that gave rise to the energetic particles that were subsequently detected by SAMPEX and other near-Earth spacecraft (see Figures 1 - 3).

Figure 4. Solar wind speed, interplanetary magnetic field (IMF) strength, and IMF north-south (Bz) component for Day 110 (21 April) to Day 140 20 May) of 1998 as measured by the WIND spacecraft [data courtesy of K. Ogilvie and R. Lepping].

The active sun also produced powerful streams of solar wind plasma that were detected upstream of the Earth. Figure 4 shows the solar wind speed (VSW), the total interplanetary magnetic field (IMF) strength (BIMF), and the IMF north-south component (Bz) for the period DOY 110-140 (1998). These data are from the WIND spacecraft. In the period from DOY 121 through DOY 140, we see four separate streams in which VSW reached peak values ~> 600km/s. Such streams are very effective at producing subsequent magnetospheric HRE events [Baker et al., 1994, 1996]. A particularly notable solar wind period occurred on Day 124 where VSW went to ~850 km/s. This is the highest solar wind speed that has been measured near 1 AU in the past several years (A. Lazarus, private communication, 1998).

When high solar wind speed occurs in combination with large BIMF, and especially when Bz is strongly negative, then we expect powerful geomagnetic activity and electron acceleration [e.g., Blake et al., 1997]. Indeed, the planetary magnetic index Kp reached 9 on Day 124 (4 May). The Dst index on that day reached -218 nT (a major geomagnetic storm) and the provisional auroral electrojet (AE) briefly exceeded 2500 nT (WDC-C2, Kyoto University). All of this indicial information is indicative of powerful, global geospace disturbances on 4 May. As we noted, the SAMPEX and POLAR data revealed a remarkably strong acceleration of relativistic electrons exceptionally deep in the magnetosphere. The new radiation belt formation as seen at L=2.2 is only rarely observed to occur and happens under extreme solar wind and geomagnetic conditions.

We have considered the space environmental conditions in April and May of 1998. We find evidence of highly disturbed solar, solar wind, and geomagnetic conditions in late April and early May. The combination of coronal mass ejections, solar flares, and high speed solar wind streams led to a powerful sequence of solar wind drivers of magnetospheric processes at the Earth. The result of the compounding solar wind disturbances was to produce a deep, powerful, and long- lasting enhancement of the highly relativistic electron population throughout the outer terrestrial radiation zone. Previous such enhancements of the HRE population have been shown to have caused spacecraft anomalies due to deep dielectric charging.

Based on the above results, it is possible that the Galaxy 4 spacecraft problem was caused, or at least exacerbated, by the energetic electron environment at geostationary orbit during May 1998. Given the long, intense duration of electron flux enhancement seen from early to late May of 1998, the circumstances were quite conducive to produce deep dielectric, (or bulk) charging [see Vampola, 1987].

In attempting to associate a given spacecraft anomaly with the space environment, it is often asked why did only this particular spacecraft have a problem? Why did not other spacecraft also fail at the same time? Moreover, since the typical geostationary orbit spacecraft has experienced numerous other HRE enhancements, why did it not fail earlier during another event?

Given the probabilistic nature of environmental conditions during catastrophic failure events, one cannot answer all of these questions in a totally definitive way. It can be argued in this case that the relativistic electrons were at a high flux value for a remarkably long time. Moreover, the energy spectrum in the May 1998 case was exceptionally hard as shown by the LANL data (Figure 2). However, even given these facts, many GEO spacecraft made it through without apparent problems during the May event interval. We note, on the other hand, that several GEO spacecraft experienced significant operational problems/anomalies. For example, the Geostationary Meteorological Satellites of Japan had more than a dozen operational anomalies in the period 4-7 May 1998 (T. Obara, Comm. Res. Lab., private communication, 1998). In the final analysis, and in previous cases [Baker et al., 1994, 1996; Reeves et al., 1998], it is quite likely that even under the most severe space environmental stresses only a few susceptible spacecraft will probably fail during any particular hostile space weather interval.

We should hasten to add that normally we cannot "prove" that any particular on-orbit spacecraft was seriously harmed by the space environment. In general, we can only make a plausiblility argument. In some instances such as the Anik E1 failure (in January 1994) laboratory testing virtually removed all doubt [Baker et al., 1994]. In other cases, we are left with only circumstantial evidence.

In summary, the solar and geomagnetic conditions in May of 1998 were very disturbed. The kinds of disturbances witnessed in this recent period are indicative of the types of events that may commonly occur during the approaching solar maximum. It will be most important to determine how robust space systems are to the multifaceted effects of the space environment over the next several years.

Acknowledgments. We thank numerous SAMPEX and POLAR colleagues for data used in this report. We also thank K. Ogilvie and R. Lepping for WIND data shown here. The GOES data were provided by NOAA and we thank H. Singer for very valuable comments. We also thank D. Wilkinson of NOAA for GOES data processing assistance. We appreciate very useful discussions with X. Li and M.A. Shea. This work was supported by NASA.

Baker, D.N., R.D. Belian, P.R. Higbie, R.W. Klebesadel, and J.B. Blake, Deep dielectric charging effects due to high energy electrons in Earth's outer magnetosphere, J. Electrostat., 20, 3, 1987.

Baker, D.N., S. Kanekal, J.B. Blake, B. Klecker, and G. Rostoker, Satellite anomalies linked to electron increase in the magnetosphere, Eos Trans. AGU, 75, 401, 1994.

Baker, D.N., et al., An assessment of space environmental conditions during the recent Anik E1 spacecraft operational failure, ISTP Newsl., 6 (2), 8-29, June 1996.

Blake, J.B., et al., CEPPAD, Space Sci. Rev., 71, 531, 1995.

Blake, J.B., D.N. Baker, N. Turner, K.W. Ogilvie, and R.P. Lepping, Correlation of changes in the outer-zone relativistic-electron population with upstream solar wind and magnetic field measurements, Geophys. Res. Lett., 24, 927-929, 1997.

Guit, W.J., Polar Spacecraft Problem, GSFC Memo, 6 May 1998. Reeves G.D., et al., The relativistic electron response at geosynchronous orbit during the January 1997 magnetic storm, J. Geophys. Res., in press, 1998.

Vampola, A.L., The aerospace environment at high altitudes and its implications for spacecraft charging and communications, J. Electrostat., 20, 21, 1987.

Figure 1. (a) Color-coded SAMPEX electron fluxes measured for energies >3 MeV during April-May 1998. Data are plotted for L-values ranging from 1 to 8 versus Day of Year for 1998. (b) A plot similar to (a) but for the POLAR spacecraft. Both (a) and (b) show a major relativistic electron enhancement on DOY 124 (4 May). Times of spacecraft operational anomalies are indicated (1, 6, and 19 May) as discussed in the text.

Figure 2. Electron fluences (cm-2) for 14-day summing periods from January 1997 through May 1998 measured by Spacecraft 1994-084 at geostationary orbit. Fluence values are normalized to peak values (which all occurred in May 1998) for the three energy ranges shown.

Figure 3. (a) GOES daily flux values of electrons with E>2 MeV for the period 21 April to 20 May 1998. Dates of various spacecraft operational problems are noted including the Galaxy 4 failures on 19 May. (b) Similar to (a) but for protons with E>100 MeV [courtesy of NOAA].

Figure 4. Solar wind speed, interplanetary magnetic field (IMF) strength, and IMF north-south (Bz) component for Day 110 (21 April) to Day 140 20 May) of 1998 as measured by the WIND spacecraft [data courtesy of K. Ogilvie and R. Lepping].