Spatial distribution of relativistic electron precipitation during a radiation belt depletion event



[1] We present observations from the NOAA-15 MEPED telescopes during a radiation belt depletion event on January 19–20, 2000 to investigate the spatial extent of electron precipitation during this interval. Precipitation mapped to the equatorial plane was confined to radial distances less than ∼6.5 Earth radii, indicating that precipitation was not the direct cause of the decrease in trapped flux observed by GOES. We found an enhanced day–night magnetic field asymmetry during the event, suggesting that magnetopause losses may have been responsible. Precipitation at lower L-values was observed by POES on the dusk passes (18:30–21:00 MLT), but not on the dawn passes, and was observed in conjugate hemispheres. These observations suggest that both precipitation and magnetopause losses were acting during this flux depletion event.

1. Introduction

[2] The relativistic electron flux in Earth's outer radiation belts is known to be highly variable [e.g., Baker et al., 1986]. Several acceleration processes have been proposed to explain observed enhancements of the flux, including shock acceleration [e.g., Li et al., 1993; Kress et al., 2007], radial transport [e.g., Elkington et al., 2003; Ukhorskiy et al., 2006], and internal acceleration by wave-particle interactions [e.g., Shprits et al., 2008]. Particles are lost through precipitation into the atmosphere [e.g., Lorentzen et al., 2001], likely caused by pitch-angle scattering into the atmospheric loss cone by plasma waves (see review by Millan and Thorne [2007]), or loss through the magnetopause when the magnetospheric configuration no longer supports closed particle drift paths [e.g., Ukhorskiy et al., 2006; Ohtani et al., 2009]. Outward radial diffusion can cause particles initially at low L-values to be lost to the magnetopause [Shprits et al., 2006]. The role played by each of the acceleration and loss mechanisms proposed is not yet established; a combination of processes is likely to be acting, but their relative importance remains unquantified.

[3] The most distinct losses appear as catastrophic depletions of the radiation belts, first reported by Frank [1965], and later studied extensively at geosynchronous orbit [e.g., Green et al., 2004; Onsager et al., 2007]. These events occur even in the absence of geomagnetic storms, but are associated with a southward turning of the IMF following a period of quiet during which a cold, dense plasma sheet forms [Onsager et al., 2007]. An increase in solar wind pressure appears to enhance the losses.

[4] The rapid depletions in the relativistic electron flux were originally attributed to the “Dst effect” in which particles adiabatically move in response to the slowly changing magnetic field. Spacecraft instrumentation measuring at a fixed location and energy will see a decrease in flux as particles move outward, adiabatically decreasing their energy, in response to the decreasing magnetic field. However, it has been shown that real losses of electrons do occur at some time during these events, and atmospheric precipitation has been proposed to play a major role [Green et al., 2004; Millan et al., 2007]. Recent analysis by Ohtani et al. [2009] brings into question the role of precipitation versus magnetopause losses for these events. They found that the day–night asymmetry of the magnetic field is enhanced during the majority of depletion events, leaving electrons at geosynchronous orbit on open drift paths.

[5] Millan et al. [2007] reported observations of relativistic electron precipitation during a radiation belt depletion event observed at GOES and GPS on January 19–20, 2000. Precipitation was observed by the MAXIS balloon payload at IGRF L = 4.7 between 21:20–01:00 UT (1920–2240 MLT) during a small, isolated substorm. They found that precipitation could account for the rate of decrease in flux observed by GPS at L = 4.7, however, MAXIS observed precipitation at only one location, and thus could not definitively determine whether precipitation caused the observed decrease in flux measured by GOES.

[6] We present observations made with the Medium Energy Proton and Electron Detector (MEPED) on the POES satellite NOAA-15 during the depletion event of January 19–20, 2000. The same instrument was recently used by Horne et al. [2009] to examine the statistical distribution of precipitation during geomagnetic storms. Since the MEPED proton telescope responds to >1 MeV electrons, these observations allow us to investigate the spatial distribution of precipitation during a particular depletion event in order to determine the role of precipitation for this case.

2. Observations and Analysis

2.1. Instrument Description and Response

[7] In this study, we use data from the NOAA-15 satellite which is part of the Polar Operational Environmental Satellite (POES) constellation operated by NOAA. Two other POES satellites (NOAA-12 and NOAA-14) were operational during this period, but data were not available during the times of interest. NOAA-15 is three-axis stabilized, and was launched in 1998 into a 833 km sun-synchronous polar (98.7° inclination) orbit. Figure 1 shows the ground track for NOAA-15 between 21:30–23:59 UT on January 19, 2000. Also shown is the location of the MAXIS balloon.

Figure 1.

Geographic location of NOAA-15 on January 19, 2000 21:30–23:59 UT. The color scale indicates countrate in the P6 proton channel, and the MAXIS balloon location is shown in red.

[8] The MEPED instrument module is part of the Space Environment Monitor (SEM) carried on each NOAA POES spacecraft, and consists of eight detectors measuring electrons and protons between 30 keV and >200 MeV [Evans and Greer, 2004]. The instrument consists of two pairs of orthogonally oriented telescopes to measure electrons and protons respectively, and a set of omnidirectional detectors. For this study, we use data from the 0-degree (zenith) proton and electron directional telescopes, which have a field of view centered within ≤20° of the local magnetic field for latitudes corresponding to the bulk of the outer radiation belts (L = 4–7), and thus measure particles in the atmospheric loss cone.

[9] The electron telescope measures electrons in the E1 (>30 keV), E2 (>100 keV), and E3(>300 keV) energy channels. The proton telescope consists of two stacked silicon detectors with an applied magnetic field across the collimator to exclude low energy electrons. However, it is well known that high energy electrons can penetrate this magnetic shield (D. Evans, private communication, 2008). The details of the instrument response to energetic electrons will be presented in a later paper. Here, we simply identify the presence of high energy electrons in the field-aligned proton telescope data by examining both the P5 and P6 energy channels which measure 2500–6900 keV and >6900 keV protons respectively. In the absence of high energy protons (found in the SAA and during solar proton events), the count rate in P5 is at background levels, and any counts detected above background in the P6 proton channel can be attributed primarily to electrons with energies ∼1 MeV. We also examined the P3 channel of the proton telescope to screen for contamination of the electron channels.

2.2. NOAA-15 Observations of Electron Precipitation

[10] Figure 2 (top) shows count rate versus time in the P5 (light grey) and P6 (dark grey) proton channels on January 18–21, 2000. The P6 count rate is offset by 1 count/sec in order to make the P5 count rate along the x-axis more visible. Also shown (Figure 2, bottom)is the >2 MeV electron flux measured by GOES-8 (black) and GOES-10 (grey) over the same time interval following Millan et al. [2007]. The P6 channel shows a significant increase in count rate at ∼21:45 UT on Jan. 19, persisting until about 1:00 UT on January 20. There is no increase in count rate detected in the P5 channel, indicating that the increase in P6 is due to energetic electrons. The X-ray burst reported by Millan et al. [2007] started near the same time at 21:40UT and also ended at ∼1:00 UT, indicating that POES was observing the same precipitation.

Figure 2.

Sixteen-second resolution count rate versus time in the P5 (light grey) and P6 (dark grey) MEPED proton channels of the zenith-oriented telescope on January 18–21, 2000 for IGRF L > 2 to exclude the SAA (top), and >2 MeV electron flux measured by GOES-8 (black) and GOES-10 (grey) during same time interval (bottom).

[11] The NOAA-15 location was mapped to the magnetic equator using the TS05 magnetic field model [Tsyganenko and Sitnov, 2005]. Figure 3a shows 2-second resolution count rates in the P6 channel versus the radial distance mapped to the equator (in Earth radii) for all spacecraft passes between 19 Jan 21:00 UT and 20 Jan 01:00 UT. The P6 count rate was elevated above background on three passes, indicated by the times in the upper right corner. In all cases, the observed relativistic precipitation maps to radial distances less than 6.5 Earth radii. The GOES spacecraft orbits, which are inclined relative to the magnetic equator, were similarly traced, and map to distances of 6.9–7.1 RE for GOES-8 and 6.6–6.7 RE for GOES-10.

Figure 3.

(a) Count rate in the P6 energy channel versus location mapped to the equatorial plane using the TS05 field model for all spacecraft passes between 19 Jan 21:00 UT and 20 Jan 01:00 UT. Colored time labels are for three passes with count rate above background. (b) Northern hemisphere pass on (left) duskside and (right) dayside showing count rate in >30 keV electron (yellow), >100 keV electron (orange), >300 keV electron (red) and P6 (green) energy channels. Geographic latitude and magnetic local time for each pass is indicated.

[12] To investigate the uncertainty in mapping the precipitation to the equator, we also performed the field line tracing using the T89 and T96 magnetic field models. For T89, we used input values of Kp from 1 to 4 (compared to the largest observed value of Kp = 3 during the interval), and for T96, solar wind data from OMNIweb were used. Using the T89 model, precipitation was confined to 5.5–6.0 RE for Kp = 1 and 5.4–6.3 REfor Kp= 4, and for T96 from 5.0–6.2 RE. These models map the precipitation to smaller radial distances than the TS05 model, likely because of the fixed location of a partial ring current near dusk in the storm-time model which leads to a more stretched magnetic field. For all of the models, the precipitation maps to locations inside of geosynchronous orbit. Although the mapping depends on the choice of field model, GOES is always mapped to a larger radial distance than the observed precipitation.

[13] During this interval, NOAA-15 directly measured precipitation over a broad range of magnetic local times (0500–0900 and 1700–2200 for L = 4–8). Moreover, because of its location near the South Atlantic Anomaly (SAA) for southern hemisphere passes, NOAA-15 will also observe particles scattered into the atmospheric drift loss cone at other local times; drift loss cone particles precipitate near the SAA where the magnetic field is weakest and particles mirror at lower altitude. Thus we expect any intense precipitation that could cause a major depletion in trapped electrons to be detected by NOAA-15.

[14] Figure 3b shows NOAA-15 count rates for one duskside (∼18:30 MLT) pass (Figure 3b, left) and one dayside (∼08:15 MLT) pass (Figure 3b, right) in the northern hemisphere. Low energy precipitation is evident in the >30 keV electron channel (yellow) for both dawn and dusk passes. However, energetic precipitation, indicated by elevated counts in the >100 keV (orange), >300 keV (red) and P6 (green) channels, was observed only on the dusk pass. Energetic precipitation was restricted to dusk passes between 18:30–21:00 MLT throughout the interval. The location of the SAA at dusk local times during this event could explain the local time dependence. However, Figure 3b shows precipitation observed in the northern hemisphere, conjugate to the SAA. The count rate on this pass was the highest observed during the entire interval, higher than that observed in the southern hemisphere, suggesting that a significant fraction of the precipitating particles were scattered locally at dusk.

3. Discussion

[15] The POES observations presented here show that precipitation did not extend out to the location of GOES during the depletion event of January 19–20, 2000. Thus, while precipitation may explain the decrease in trapped flux observed by GPS as discussed by Millan et al. [2007], it cannot directly explain the decrease observed by GOES at higher L.

[16] To explain the depletion in the >2 MeV trapped electron flux, the particles were either lost to the magnetopause, were radially transported inward where they were precipitated, or they were somehow decelerated to lower energies below the 2 MeV threshold of the GOES integral energy channel. Millan et al. [2007] showed that the decrease in flux was observed down to ∼400 keV, therefore, this third possibility would require particles to be rapidly decelerated from >2 MeV to energies below 400 keV.

[17] Ohtani et al. [2009] examined the ratio of the magnetic field strength at the sub-solar magnetopause (Br0) to the measured north–south GOES H-component (BH), and found a ratio greater than one for the majority of electron loss events. They suggest this indicates particles on open drift paths since equatorially mirroring particles are expected to drift along contours of constant magnetic field strength. We calculated the ratio Br0/BH for the event presented here using 1-minute resolution ACE data from OMNIWeb for the solar wind dynamic pressure, and the magnetic field measured by GOES-8. Prior to the depletion event (18:00–20:000 UT on Jan. 19), the ratio was 0.6. As GOES-8 moved into dusk, it began to observe stretching of the magnetic field [Millan et al., 2007]. The solar wind dynamic pressure also increased, reaching a maximum of 5.5 nPa at 22:57 UT. At this time, Br0/BH = 1.7, indicating particle drift orbits that intersect the magnetopause. For comparison, we also examined the ratio on January 18 when the spacecraft was at the same local time but one day prior to the depletion event, and found Br0/BH = 0.6.

[18] Our 3D test particle model shows that during this interval, GOES is in the drift-orbit bifurcation region [e.g., Öztürk and Wolf, 2007], and moves into the region where particles are on open drift paths. Particles that exhibit drift-orbit bifurcation near the open-closed drift orbit boundary can also be lost to the magnetopause within several drift periods.

[19] Inside geosynchronous orbit, precipitation was likely caused by wave-particle scattering of the particles into the atmospheric loss cone [e.g., Millan and Thorne, 2007]. Precipitation was observed only on the dusk passes of POES, consistent with observations of duskside precipitation reported previously [Millan et al., 2002], possibly due to scattering by EMIC waves [Lorentzen et al., 2000]. The precipitation observed here extended down to energies of a few hundred keV which is also consistent with the energies reported by Millan et al. [2007] for the same event. Experimental results indicate that waves very near the ion cyclotron frequency are not uncommon and could be resonant with electrons at energies as low as a few hundred keV [Ukhorskiy et al., 2010].

[20] The observations presented here suggest that both precipitation and magnetopause losses were acting during this flux depletion event. Strong precipitation was observed at low L-values, but the enhanced day–night asymmetry of the magnetic field near geosynchronous orbit indicates magnetopause losses also played a role. Depending on the resulting radial gradient in phase space density due to precipitation losses at lower L-value and magnetopause losses beyond GEO, radial transport may carry particles inward where they are precipitated or outward where they are lost to the magnetopause. Future work must quantify the relative importance and interaction between these processes.


[21] This work was supported by NSF ATM–0457561 and NSF ATM–0540121. POES data were provided by the NOAA Space Weather Prediction Center.