Click Here

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 113, A09212, doi:10.1029/2007JA013011, 2008

Full Article

for

Numerical calculations of relativistic electron drift loss effect

Kyung Chan Kim,1 D.-Y. Lee,1 H.-J. Kim,2 L. R. Lyons,3 E. S. Lee,4 M. K. Ozturk,5 1 and C. R. Choi
Received 29 December 2007; revised 13 May 2008; accepted 16 June 2008; published 9 September 2008.

[1] It has been suggested that drift loss to the magnetopause can be one of the major loss

mechanisms contributing to relativistic electron flux dropouts. In this study, we examine details of relativistic electrons drift physics to determine the extent to which the drift loss through the magnetopause is important to the total loss of the outer radiation belt. We have numerically computed drift paths of relativistic electrons guiding center for various pitch angles, various measurement positions, and different solar wind conditions using the Tsyganenko T02 model. We specifically demonstrate how the drift loss effect depends on these various parameters. Most importantly, we present various estimates of relative changes of the omnidirectional flux of 1 MeV electrons between two different solar wind conditions based on a simple form of the directional flux function. For a change of the dynamic pressure from 4 nPa to 10 nPa with a fixed IMF BZ = 0 nT, our estimate indicates that after this increase in pressure, the equatorial omnidirectional flux at midnight near geosynchronous altitude decreases by $56 to $97%, depending on the specific pitch angle dependence of the directional flux. The effect rapidly decreases at regions earthward of geosynchronous orbit and shows a general trend of decrease away from midnight. For a change of the IMF BZ from 0 nT to 15 nT with a fixed dynamic pressure of 4 nPa, the relative decrease of the omnidirectional flux at geosynchronous altitude on the nightside is much smaller than that for the pressure increase, but its effect becomes substantial only beyond geosynchronous orbit. Possibilities exist that our results may change to some extent for a different magnetospheric model than the one used here.
Citation: Kim, K. C., D.-Y. Lee, H.-J. Kim, L. R. Lyons, E. S. Lee, M. K. Ozturk, and C. R. Choi (2008), Numerical calculations of relativistic electron drift loss effect, J. Geophys. Res., 113, A09212, doi:10.1029/2007JA013011.

1. Introduction
[2] It has been suggested that there are several processes which can contribute to relativistic electron flux dropouts. They can be divided into two classes depending on whether the process of flux dropouts is adiabatic or non-adiabatic. The most well-known adiabatic process is a fully-adiabatic flux change (or the Dst effect) that leads to a global dropout during the storm main phase [e.g., Kim and Chan, 1997; Li et al., 1997]. During storm-times, relativistic electrons respond to the changing magnetospheric field because of the enhanced ring current by conserving all three adiabatic invariants. This reduces relativistic electron fluxes as measured at constant energy and location, but they are expected
1 Department of Astronomy and Space Science, Chungbuk National University, Chungbuk, South Korea. 2 Department of Astronomy and Space Science, Kyunghee University, Yongin, Gyeonggi, South Korea. 3 Department of Atmospheric Sciences, University of California Los Angeles, Los Angeles, California, USA. 4 Space Sciences Laboratory, University of California Berkeley, Berkeley, California, USA. : 5 Department of Information Technologies, Isk University, I stanbul, Turkey.

Copyright 2008 by the American Geophysical Union. 0148-0227/08/2007JA013011$09.00

to return to pre-storm levels when the storm recovers. However, there have been observations during storms where the reduced fluxes of relativistic electrons during the storm main phase never recovered to their pre-storm levels at the end of a storm [e.g., OBrien et al., 2001; Onsager et al., 2002; Reeves et al., 2003]. These suggest that other loss processes occurred during the storm-time leading to a permanent loss of relativistic electrons within the radiation belt. [3] In fact, there are several other, either adiabatic or nonadiabatic, mechanisms proposed to explain flux dropouts. They include the substorm growth phase-like flux dropout [e.g., Reeves et al., 1998], atmospheric precipitation by wave-particle interaction or by pitch angle scattering under severely curved magnetic field [e.g., Horne and Thorne, 2003; Green et al., 2004; Lee et al., 2006; Bortnik et al., 2006; Millan et al., 2007], and drift loss through the magnetopause [e.g., Li et al., 1997; Desorgher et al., 2000]. [4] This paper deals with the drift loss effect where electrons are lost permanently through the magnetopause when their drift shell is opened to the magnetopause [e.g., Li et al., 1997; Desorgher et al., 2000]. Recently, this process has been reported as an important cause of flux dropouts. Ukhorskiy et al. [2006] suggested, on the basis on the test particle simulation, that continuous intensification of the ring current during storm-times leads to a drift loss to the
1 of 14

A09212

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

magnetopause by opening previously closed drift shell. Bortnik et al. [2006] also reported that at high L-shells (L > 5) electrons can move outward aided by the Dst effect and then encounter the magnetopause. Shprits et al. [2006] suggested that an outward radial diffusion may be driven by the drift loss. However, in a recent paper, Green et al. [2004] studied three possible causes of the ensuing flux decrease. They suggested that drift loss to the magnetopause cannot explain the flux dropouts due to discrepancies between the location of the magnetopause and the region of decreased flux. It seems necessary to study in a systematic way conditions under which the drift loss effect can be a significant loss process. [5] The goal of this study is to understand details of MeV electrons drift physics and determine the extent to which, and the region where, drift loss contributes to the total loss (or flux dropout) significantly and to determine the solar wind conditions under which drift loss could be a dominant process. For this study, we have numerically computed guiding center trajectories of 1 MeV electrons for various pitch angles and solar wind conditions using a Tsyganenko model. The primary solar wind parameters that are expected to affect the drift loss process are dynamic pressure and IMF BZ. Lyons et al. [2005] suggested that the drift loss effect is expected to be effective because of a solar wind dynamic pressure enhancement whether or not there is a storm. Thus, in this study, we test the effect of both dynamic pressure and IMF BZ on the drift loss process. [6] The paper is organized as follows. First, in section 2 basic equations and models are introduced. In section 3 we use guiding center trajectory calculations for 1MeV electrons to show how the electrons drift loss depends on pitch angle, spacecraft location in terms of MLT, radial and vertical distances, and solar wind conditions. In section 4, we estimate omnidirectional flux changes based on the results in section 3. Discussion and conclusions are given in sections 5 and 6, respectively.

and adaptive Bulirsch-Stoer numerical scheme to solve the guiding center equations [Press et al., 1994]. We have used the guiding center equations by Brizard and Chan [1999] above rather than the conventional equations by Northrop [1963] because the conventional equations lead to nonnegligible numerical errors when electrons are located in the regions where the magnetic field curvature is very high. The guiding center expression by Brizard and Chan [1999] is based on phase space Lagrangian formulation to minimize such error and to ensure conservation of energy and phase space volume to second-order in the adiabatic expansion parameter (see Brizard and Chan [1999] for more details). 2.2. Background Magnetic Field Models [9] In this study, we have used the T02 model by Tsyganenko [2002a, 2002b] as the background magnetic field, which is static in time, and electric field is not included. In the T02 model one can vary the solar wind dynamic pressure PD, the IMF BZ, and other solar windrelated quantities as input parameters. It models more realistic situations than previous models do by taking into account several important factors such as a significant dawn-dusk asymmetry and by employing more flexible and realistic approximation for the field of all magnetospheric sources including magnetopause current, ring current, cross-tail current, and large-scale field aligned current. In the calculations discussed below, we have tried various combinations of the input parameters in order to obtain various models of the magnetosphere with zero tilt angle for which the guiding center equations of motion above have been solved. 2.3. Response of the Magnetosphere to Various Solar Wind Conditions [10] Figure 1 shows the magnetospheric magnetic field change, DB, in response to four different values of the solar wind dynamic pressure PD. Here DB means a difference in the magnitude between the total magnetic field and the purely dipolar magnetic field, so that positive and negative values refer to compression and depression, respectively. For all four cases IMF By = BZ = 0 nT. It is clear that as PD increases, the magnetopause moves inward, the dayside compression (DB > 0) increases, and some modest depression on the nightside increases, which must be due to a tail current increase with PD increase built into the T02 model. A higher PD value results in a larger day-night asymmetry as well as a significant dawn-dusk asymmetry. [11] Figure 2 shows the magnetospheric magnetic field change corresponding to four different IMF BZ values with PD = 4nPa fixed. We can see from Figure 2 that a more southward IMF BZ leads to a stronger depression on the nightside due to an increased tail current. This results in increasing day-night asymmetry. However, the effect is weaker than the dynamic pressure effect in Figure 1. However, it should be mentioned that the T02 model does not include the effect of the IMF on the magnetopause position and shape, which thus might lead to a weaker compression effect on the dayside than in reality. We suspect that in a real situation the IMF BZ effect might be somewhat more significant than the results presented in this paper.

2. Models for Test Particle Orbit Calculations

2.1. Guiding Center Equations of Motion of Relativistic Electrons [7] In order to trace relativistic electrons we have used the set of guiding center equations derived by Brizard and Chan [1999] as given by equations (1) and (2) below: The relativistic guiding center velocity, which includes the parallel motion as well as the drift motion, is expressed as
B* d~ pjj ~ R m^ b rB * qgB* gm Bjj dt jj 1

[8] The guiding center acceleration parallel to the local magnetic field is given by
dpjj m~ B* rB dt gB* jj 2
g 2 mv2

Here g = (1 v2/c2)1/2 is the relativistic factor, m = 2B ? is p b, the relativistic magnetic moment and ~ = ~ + qk r ^ B* B b B*. Bk* = ^ ~ We have used the Richardson extrapolation

2 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 1. T02 DB calculation as obtained by varying the dynamic pressure PD but with IMF BZ fixed to 0.

[12] Figures 1 and 2 shed some light on how relativistic electrons respond to changing solar wind conditions. The feature that a stronger PD and a more intense IMF BZ lead to a more intense day-night asymmetry implies a higher likelihood of drift shell shielding by the magnetopause,

which would enhance the chance that electrons could be lost through the magnetopause. Here, by drift shell shielding, we mean that the drift shell encounters the magnetopause, and thus it is used equivalently to the term magnetopause shadowing. To estimate this effect quantitatively using a

Figure 2. T02 calculation results as obtained by varying the IMF BZ but with the dynamic pressure PD fixed to 4 nPa.
3 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

started at 24 MLT at geosynchronous orbit. Figure 3a shows a portion of the drift paths for three different initial pitch angles which illustrate the three types: The outer most path (green) corresponds to initial a (pitch angle) = 70 and this orbit is open to the magnetopause; The intermediate shell (cyan) corresponds to initial a = 50, and this orbit undergoes bifurcation on the dayside; The inner most shell (dark yellow) corresponds to initial a = 30 and this orbit is closed and the particle is trapped. Figure 3b shows the pitch angle-dependent shell splitting effect [Roederer, 1970]. The orbits of electrons with pitch angle 90 60 are open to the magnetopause. These electrons are directly lost after drifting through the magnetopause. The orbit for electrons with pitch angle 50 is bifurcated on the dayside and the part of the orbit on the dayside is not shown in this equatorial view. Whether or not the particles that suffer from DSB are lost through the magnetopause depends on whether the particle moves on an open shell or on a closed shell in the absence of DSB. More detailed discussion on this point is given below. The rest of the orbits are for lower pitch angles and correspond to trapped orbits. 3.2. Dependence of the Drift Loss Effect on PD, IMF BZ, Pitch Angle, and Spacecraft Position [16] In order to evaluate the drift loss effect in a systematic way, we first launched 1 MeV electrons in the geosynchronous orbit equatorial plane every 2 hours in nightside MLT from 18 to 6 with different initial pitch angles from 10 to 90 separated by 10. The initial launching positions on the equatorial plane at 6.6 RE are displayed in Figure 4. Their orbits are traced for one full orbit or until being stopped when they encounter the magnetopause under T02 models with various solar wind conditions. It is assumed that the particles encountering the magnetopause are actually lost (The readers are referred to the study by Kim et al. [2005] and references therein for more discussion on this assumption). The orbit calculation determines the type of each orbit, namely, trapped, bifurcated, or open orbit, which depends on the particles pitch angle, MLT position, and solar wind condition. [17] The calculation results are summarized in Figure 5, which classifies drift path types in the parametric domain of the particles initial pitch angle and MLT value of the particles launching position. The results were obtained for four different values of PD from 4nPa to 20nPa with IMF BZ fixed to zero. In each plot, the horizontal axis represents the particles launching MLT on the nightside and the vertical axis the initial pitch angle at the launching position. The cases where electron orbits are open to the magnetopause are indicated as MP loss (Direct) (red). The cases where electron orbits suffer from the bifurcation at the dayside magnetopause but after then encounter the magnetopause at high latitudes are indicated as MP loss (after Bifurcation) (pink). This case is further demonstrated by the orbit in Figure 6, where the electron first undergoes bifurcation on the dayside then hits the southern hemispheric magnetopause (because the magnetopause is quite close due to the high dynamic pressure) after executing a new bounce motion in the southern hemispheric region. Similar MP losses can also occur in the northern hemisphere for other pitch angle particles. The cases where electrons suffer from bifurcation on the dayside but are trapped

Figure 3. Examples of 1 MeV electron guiding center orbits for PD = 10 nPa and IMF BZ = 0 nT in the T02 model. (a) Three types of drift paths. (b) Pitch angle-dependent shell splitting effect.

reasonable model magnetosphere is our main goal of this work as presented below.

3. Results of Guiding Center Drift Calculations for Various Solar Wind Conditions
[13] The drift loss effect in general depends on several factors including individual particle pitch angle, the radial, longitudinal and vertical positions of an observing spacecraft, and solar wind conditions. In this section, we will demonstrate this effect quantitatively. 3.1. Three Types of Drift Paths [14] We first introduce three typical types of drift path of relativistic electrons in the magnetosphere: (1) Trapped path, (2) Shabansky orbit or Drift-Shell Bifurcation (DSB), and (3) Open path [Shabansky, 1971; Ozturk, 2004; Ozturk and Wolf, 2007]. [15] Examples of the three types of drift path and their dependence on pitch angle are shown in Figure 3. These examples are for a 1 MeV electron in a T02 model with PD = 10 nPa and IMF BZ = 0 nT. In all cases, the particles were

4 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 4. Initial launching positions at geosynchronous orbit. (do not hit the magnetopause) are shown as Bifurcation (Trapped) (yellow) in Figure 5. In other words, bifurcation may or may not lead to the drift loss, and the two possibilities are distinguished as pink and yellow. We will discuss this issue further in section 5. The cases where electron orbits are simply closed are indicated by Permanently Trapped (blue). One can see from Figure 5 how the drift loss effect increases for higher values of PD: The drift loss effect propagates to lower pitch angles and wider MLT regions for higher values of PD. Note that this drift loss effect is due to both the direct shielding of drift shells by the magnetopause and the shielding by the magnetopause after drift-shell bifurcation.

[18] To demonstrate the IMF BZ effect on the drift loss, a diagram similar to Figure 5 is shown in Figure 7, where PD is fixed at 4 nPa and IMF BZ is varied from 0 to 15 nT. For this case, the drift loss effect is limited to equatorially mirroring particles for all four cases. We can see that the domain for the bifurcation effect significantly increases as the IMF BZ becomes more southward, but all these orbits are trapped orbit. We have found no case where the bifurcation causes the bifurcated drift shell to encounter the magnetopause at high latitudes, but there is a possibility that this result could change if we had tried the particles with a different bounce phase [Ozturk and Wolf, 2007]. This issue will be discussed further in the Discussion. Figure 7 suggests that we do not obtain as much drift loss by imposing a more southward IMF BZ as we do by increasing pressure in Figure 5. However we caution the readers in accepting this result because of a few factors to be considered carefully. First, the T02 model does not take into account the magnetopause response to the varying IMF values. In other words, the magnetopause does not move as the IMF BZ changes. If the IMF BZ effect on the magnetopause shape is correctly included, we expect that the drift loss domain will increase to some extent in both MLT and pitch angle compared to the results presented in Figure 7. Another factor is the strength of the southward IMF BZ. A more strongly southward IMF will lead to a more stretched tail magnetic field, and thus weaker equatorial magnetic field, which would enhance the chance for electrons to encounter the magnetopause. Detailed calculations based on an improved model than the currently used T02 model is necessary to verify this possibility quantitatively. Also a further effect that is expected from a very stretched tail field is scattering of pitch angle by increased magnetic field curvature, which could scatter particles into a smaller pitch

Figure 5. Representation of drift path types in parametric domain of the particles launching position (MLT) and initial pitch angle at the launching position for four different values of PD with fixed value of IMF BZ = 0.
5 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 6. MP loss (after Bifurcation) type of drift orbit. An 1-MeV electron was launched at 24 MLT with 60 pitch angle for PD = 10 nPa and IMF BZ = 0 nT. angle, reducing the possibility of drift loss. In Figure 7, we have not presented the results for IMF BZ < 15 nT because of two reasons. First, our numerical scheme becomes less reliable for an excessively stretched magnetic field as would be caused by a very strongly southward IMF condition due to increasing numerical errors with increasing magnetic field stretching. We have however confirmed that the results for IMF BZ = 15 nT are reliable. Second, we concern the fact that validity of guiding center approximation decreases with increasing stretching of magnetic field. [19] In Figures 5 and 7 we have explicitly shown the effect of drift loss to the magnetopause as a function of MLT but at fixed geosynchronous orbit. We now demonstrate the drift loss effect as a function of radial position, but at a fixed 24 MLT, in response to various solar wind conditions. For this study, we have computed the radial dependence of the trapping boundary for different pitch

angles as well as for various solar wind conditions. For the purposes of this paper, the trapping boundary is defined to be the outer most boundary where relativistic electron orbits are closed, and this may be different from definitions by some other authors [e.g., see Imhof et al., 1990]. The result is presented in Figure 8. [20] Figure 8 (top) shows curves for the trapping boundary at 24 MLT in the equatorial plane for various dynamic pressures with fixed IMF BZ = 0, and Figure 8 (middle) shows similar results for various IMF BZ values with fixed PD = 4 nPa. In both of these plots, the horizontal axis represents the (equatorial) radial distance at 24 MLT and the vertical axis the initial equatorial pitch angle at each radial launching position. Geosynchronous altitude is marked for reference. The domain below each curve represents the trapped region for a given solar wind condition. It is clearly seen that a stronger solar wind dynamic pressure PD and/or a more southward IMF BZ moves the trapping boundary closer to the Earth due to loss to the magnetopause. Also the trapping boundary for a higher pitch angle is located closer to the Earth than that for a lower pitch angle. This is simply due to the pitch angle-dependent drift shell splitting effect, where higher pitch angle electrons are more likely to encounter the magnetopause than are lower pitch angle electrons. However, we also see from Figure 8 (middle) that the drift loss effect for non-equatorial electrons (pitch angle <90) due to increasing southward IMF BZ is significant only beyond 6.6 RE. [21] Figure 8 (bottom) shows variations of trapping boundary thickness for various solar wind conditions. We here define the trapping boundary thickness as the difference in the radial distance of the trapping boundary between pitch angle of 20 and 90 as obtained from the first two images. The two curves represent the effects from the pressure and from the IMF BZ, respectively. One can see the overall tendency that the trapping boundary thickness

Figure 7. Similar to Figure 5, but PD = 4 nPa fixed and IMF BZ = 0, 5, 10, 15 nT.
6 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 8. Trapping boundary locations (at 24 MLT and in the equatorial plane) versus pitch angles (top) for various PD values with IMF BZ = 0 nT and (middle) for various IMF BZ values with PD = 4 nPa. (bottom) Trapping boundary thickness (see text for its definition).

decreases with increasing solar wind dynamic pressure. For the IMF BZ, there is a dramatic decrease when the IMF BZ changes from 10 to 15 nT. This is primarily due to the fact that the nightside field stretching is more pronounced when IMF BZ changes from 10 to 15 nT than when it

does between smaller values (in magnitude) as one can see in Figure 2. [22] The trapping boundary plots in Figure 8 are used below to give the pitch angle dependence of electron flux that would be measured by a spacecraft. We give an example here. Suppose that IMF BZ = 0, and the dynamic pressure is 10 nPa for which the trapping boundary is given as the red line in the top plot of Figure 8. Now consider a spacecraft sitting at the geomagnetic equator at radial distance 6.6 RE and 24 MLT. Then, according to Figure 8, electrons with pitch angle a > 53.4 at the equator would all be lost to the magnetopause while those with pitch angle a < 53.4 at the equator would be trapped. While we have obtained this result by launching electrons at 24 MLT and tracing them, it applies to any electrons arriving there and their orbital fate will be determined by the condition whether or not their pitch angle at the arrival point is greater than the critical angle 53.4. A caution is however needed at this point. Within one drift orbit of electrons the spacecraft can see electrons drifting from west onto 24 MLT even if their pitch angle at the arrival point is larger than the critical angle. However, eventually such electrons will all be lost and not be seen by the spacecraft over a longer time scale. [23] Figure 9 shows the range of pitch angles for electrons at the equator at a radial distance of 6.6 RE at 24 MLT to remain to be trapped. This pitch angle range becomes narrower for a stronger dynamic pressure as shown in Figure 9 (top). This effect is insignificant for the case of increasing southward IMF BZ shown in Figure 9 (bottom), but the readers are reminded of what we have already pointed out above regarding uncertainties in the IMF BZ effect. Note that the result here is for a point in the equatorial plane, and it changes to some extent for a location away from the equator as discussed below. [24] Since spacecraft orbits often traverse away from the equator, we compare what a spacecraft away from the equator can measure to that of an equatorial spacecraft. We consider a spacecraft located radially at 6.6 RE, but at ZSC = 0.5 RE, and display relevant computation results in Figure 10. First, the range of pitch angles in the equatorial plane for which electrons can travel up to the spacecraft position was obtained by first adiabatic invariant conservation. Figures 10a and 10d display the maximum equatorial pitch angles of electrons that can travel up to the spacecraft located at ZSC = 0.5 RE above the equator at 6.6 RE on the nightside for various solar wind conditions. The equatorial crossing points of the field lines on which the spacecraft resides are shown in Figures 10b and 10e. It is seen that either a stronger dynamic pressure or a more strongly southward IMF condition leads to narrower ranges of equatorial pitch angles on the nightside that can travel up to the spacecraft location. This is due to field line stretching that causes an increase of the field intensity above the equator and a reduction of the field intensity at the equator, resulting in enhancement of the field intensity ratio between the two regions as presented in Figures 10c and 10f. Depending on the specific vertical position of a spacecraft, the equatorial pitch angle range of electrons that can be seen by the spacecraft can therefore vary: As the vertical position of the spacecraft is raised, the field intensity ratio increases, and the pitch angle range that the spacecraft can detect becomes narrower.

7 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 9. Range of pitch angle for electrons at radial distance of 6.6 RE at 24 MLT in the equatorial plane to remain to be trapped. (top) For various PD but fixed IMF BZ = 0 nT. (bottom) For various IMF BZ but PD fixed to 4 nPa.

[26] In Figure 11 (middle and bottom), the dashed lines refer to the maximum equatorial pitch angle that a spacecraft located at 24 MLT at (XSC = 6.6 RE, ZSC = 0.5 RE) can see as determined by first adiabatic invariant conservation as in Figure 10. We emphasize that this is pitch angle at the equatorial crossing point of the field line on which the spacecraft resides. The solid lines refer to the other maximum equatorial pitch angle at the equatorial crossing point of the same field line, which is for electrons that survive drift loss as shown in Figures 8 and 9. Note that since the spacecraft is away from the equator, the equatorial crossing point of the spacecraft field line is larger than 6.6 RE. In Figure 8 (middle), it is clear that for all cases, except for PD = 4 nPa, the drift loss effect (namely the solid lines), rather than first adiabatic invariant conservation, determines the pitch angle range of electrons that the spacecraft away from the equator will measure. In other words, the pitch angle range that the spacecraft will measure becomes narrower by considering the drift loss effect than considering only the first adiabatic invariant conservation. For example, the dashed line predicts that, for PD = 10 nPa, IMF BZ = 0 nT, the pitch angle range of electrons that can reach the spacecraft location is <60 at the equatorial crossing point of the spacecraft field line if one does not consider the drift loss effect. However, the drift loss effect indicates that electrons only with a pitch angle of <48 at the same equatorial crossing point can survive drift loss. Therefore it is the drift loss effect that determines the actual pitch angle range that the spacecraft located at ZSC = 0.5 RE will see. In contrast, the IMF BZ effect is different as shown in Figure 11 (bottom). The pitch angle range that will be seen at the spacecraft away from the equator is determined by first invariant conservation for all BZ (the dashed line). However, as discussed above, it is possible that the IMF BZ effect that we have obtained with the current T02 model might change to some degree if the magnetopause position and shape in response to IMF BZ change were properly implemented into the model.

4. Calculation of Omnidirectional Flux and Flux Dropouts

[25] We stress however that the results in Figure 10 were obtained without considering the possible drift loss of electrons. What pitch angle range that a non-equatorial spacecraft can measure at a given vertical position should be determined by considering both the equatorial pitch angle range obtained from first adiabatic invariant conservation as in Figure 10 and the pitch angle range of electrons which survive the drift loss effect. This is more clearly illustrated in Figure 11 (top) where two cones of pitch angle range are sketched: If one represents an equatorial pitch angle range of electrons which do not undergo drift loss, then the other one automatically represents another range of equatorial pitch angle determined by first invariant conservation of electrons that can reach the spacecraft location. If the drift loss effect is large such that it leads to the inner cone, then it is the drift loss effect that determines the pitch angle range of electrons that can travel up to the spacecraft location. Examples of detailed computations for various solar wind conditions are given in Figure 11 (middle and bottom). [27] In the above we have found through test particle orbit calculations how the pitch angle range that can be observed by a spacecraft becomes narrower for different solar wind conditions and its dependence on observing spacecraft position. We have used this result in order to quantitatively estimate how and to what extent omnidirectional flux dropouts seen by a spacecraft can be affected by the drift loss. [28] Our main goal here is to calculate the relative change (i.e., percentage change) of omnidirectional flux due to the drift loss at a fixed spacecraft between two different solar wind conditions. To obtain an omnidirectional flux, one needs to specify a proper pitch angle function of directional flux, from which one can draw an omnidirectional flux. For this work, we use equation (3) below as the pitch angle function of the directional particle flux at the equator.
j0 a0 C0 sinn a0 3

8 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 10. (top) Cases wherein IMF BZ is fixed to zero. (bottom) Cases wherein PD is fixed to 4 nPa. (a, d) Maximum equatorial pitch angle below which electrons can travel by conserving the first adiabatic invariant up to a spacecraft located at ZSC = 0.5 RE above the equator at 6.6 RE as a function of PD, IMF BZ, and nightside MLT (see text for detailed explanation). (b, e) Equatorial crossing point of the field line that crosses the spacecraft positioned off equator as a function of PD, IMF BZ, and nightside MLT. (c, f) The ratio of the field intensity at the spacecraft location ZSC = 0.5 RE to that at the equatorial crossing point as a function of PD, IMF BZ, and nightside MLT. The corresponding omnidirectional flux is given by
Z J0 4p
0 ac0

Z j0 a0 sin a0 da0 4pC0

0

ac0

sinn1 a0 da0 4

Here B0 and B are the field intensities at the equator and spacecraft position, respectively, and the corresponding omnidirectional flux is given by
Z J 4p
0 ac

Z ja sin ada 4pC0

0

ac

The value of n has been reported to depend on L-shell [Vampola, 1998]. However, for our study, we assume that n and C0 are constant, implying that there is no gradient in the directional flux across magnetic flux tubes. We stress that the integral in equation (4) is over the pitch angle range for which electrons remain to be trapped over the time scale longer than one drift orbit period when observed by a spacecraft at a given position at the equator as discussed in the previous section. In other words, the integral range for the pitch angle extends from 0 to ac0, where ac0 is the critical pitch angle beyond which the electrons are lost to the MP. Without the drift loss, the integral would include all pitch angles from 0 to 90 degrees. [29] Next, by applying the Liouville theorem and conservation of 1st adiabatic invariant to equation (3), we can obtain the directional flux at a spacecraft located at an arbitrary position along the same field line as below.
ja C0 B0 =Bn=2 sinn a 5

B0 =Bn=2 sinn1 ada 6

Here ac refers to the pitch angle range that will be seen at the non-equatorial spacecraft position, and this is affected by critical pitch angle ac0 survived from the drift loss at the equator. [30] Then, the relative omnidirectional flux change in percentage seen by a spacecraft between two different solar wind conditions can be written as follows.
JSW 1 JSW 2 100 JSW 1 0 1 Ra n=2 B0 =BSW 2 0 c sinn1 adaSW 2 A 100 R ac @1 n1 n=2 B0 =BSW 1 0 sin adaSW 1

Here subscript SW1 and SW2 mean solar wind conditions 1 and 2, respectively. We use a simple model to demonstrate the pure drift loss effect on flux dropouts by assuming that n(r) = 1 and C0(r) is a constant for both solar wind conditions 1 and 2.

9 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

[31] We choose here the case where the dynamic pressure increases from 4 to 10 nPa with IMF BZ = 0 fixed. Also our calculation is done for spacecraft locations at different nightside MLTs from 18 to 6 and several different vertical positions. First, let us consider the relative omni-flux decrease at the equator. The corresponding expression can be obtained from equation (4) or (7) as follows.
Relative omniflux decrease% at equator   J0PD 10nPa 100 1 J0PD 4nPa

Figure 11. (top) Cones of two different pitch angle ranges of electrons that can travel up to a spacecraft situated above the equator (see text for details). (middle and bottom) The dashed lines represent maximum equatorial pitch angles below which a spacecraft at 24 MLT at (XSC = 6.6 RE, ZSC = 0.5 RE) can measure as determined on the basis of first adiabatic invariant conservation only. The solid lines represent the maximum pitch angle of electrons that survive from the drift loss effect at the equatorial crossing points of the field line that penetrates the spacecraft located at 24 MLT at (XSC = 6.6 RE, ZSC = 0.5 RE) (see more details in text).

The critical pitch angle ac0 in J0 of this equation would be different between the two solar wind conditions, which thus leads to the relative change in the equatorial omnidirectional flux. The result is shown in Figure 12 (top and middle). [32] In Figure 12 (top), the horizontal axis represents the spacecraft XSC-position at the equator at 24 MLT. The column graphs represent the trapped (or observable) pitch angle range for the two PD values, namely, white column for PD = 4 nPa and hatched column for PD = 10 nPa. It is seen that the observable pitch angle range becomes narrower for the higher pressure state, which is however significant for XSC = 6.5 RE and beyond. The narrower pitch angle range leads to a flux decrease for the PD = 10nPa compared to that for PD = 4nPa. This relative decrease is shown by the solid line curve. It is clearly seen that the drift loss effect exists as close as $5.3 RE, but the relative flux drop is weak to mild inside geosynchronous altitude. The relative flux drop becomes significant at and beyond geosynchronous altitude: The relative flux decrease is $60 70% near geosynchronous altitude and the maximum decrease is $87% at $7.2 RE. Of course, one can expect from Figure 8 that by considering a stronger pressure state than 10 nPa, the drift loss effect would penetrate into a region inward of $5.3 RE and the relative flux dropout would overall become more significant than shown in Figure 12. [33] Figure 12 (middle) shows the MLT dependence of the relative omnidirectional flux change showing results for spacecraft located at the equator at 3, 6, 18, 21 and 24 MLT. First, the overall profiles of omnidirectional flux decrease at 3 and 21 MLT are similar to that at 24 MLT but the drift loss effect starts to occur at slightly larger radial distances than at 24 MLT. It can be seen that the omnidirectional-flux decrease near geosynchronous altitude at 24 MLT is larger by approximately 10% than the decreases at 3 and 21 MLT. However, the difference is rather sensitive to measurement location: The flux decrease is larger at 24 MLT than at 3 and 21 MLT by a factor of $3 at $6.4 RE, for example. On the other hand, the relative flux decrease at 6 and 18 MLT inside geosynchronous altitude is not significant, 20% or less, and becomes substantial only at $7 RE or beyond. [34] Figure 12 (bottom) shows the relative omnidirectional flux change calculated for two vertical locations, ZSC = 0.3, and 0.5 RE, of a spacecraft at 24 MLT. For this calculation, equation (7) was used. For comparison, the relative flux decrease at the equator (ZSC = 0.0 RE) is also shown. The relative flux decrease at ZSC = 0.3, 0.5 RE is significant at and beyond XSC $ 6.3 RE, showing basically similar tendencies as for ZSC = 0.0 RE. For XSC $ 5.3 to 6.3 RE the relative flux drop seen by the non-equatorial spacecraft is quite small, as indicated by dotted lines in the

10 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 12. Relative omnidirectional flux decrease (%) as PD increases from 4 to 10 nPa with IMF BZ = 0 fixed. (top) The range of trapped (or observable) pitch angle (column graph) and relative omni-flux decrease (%) (solid line) seen by a spacecraft at 24 MLT at the equator as a function of spacecraft XSC-position. (middle) Profiles of the relative omni-flux decrease for different MLTs at the equator. (bottom) Profiles of the relative omni-flux decrease for various values of vertical location ZSC of a spacecraft located at 24 MLT.

figure. This is explained as follows. First, readers are reminded that we demonstrated in Figure 11 that there can be in general competition in determining the pitch angle range observable at a non-equatorial position between the drift loss effect at the equator and the first adiabatic invariant conservation. For XSC $ 6.3 RE and larger, it is the drift loss effect that determines the observable pitch angle range at the non-equatorial and the equatorial spacecraft positions, and thus drift loss leads to the large omnidirectional flux decrease for the higher PD for all the locations. Within the inner region (dotted line) inward of XSC $ 6.3 RE, what determines the observable pitch angle range at the non-equatorial spacecraft is the first adiabatic invariant conservation. However the field intensity ratio between the equator and the non-equatorial spacecraft location is not much different between the two solar wind conditions, thus leading to the minor relative decrease of the omnidirectional flux. [35] We have also performed similar calculations of the relative flux drop for two different values of IMF BZ, 0 and 15 nT, with a fixed PD = 4 nPa. The result is shown in Figure 13. One can see from Figure 13 (top) that the relative flux drop due to the drift loss effect becomes significant only somewhat beyond geosynchronous altitude for ZSC = 0.0RE. The results shown in Figure 13 (middle) indicate no significant difference for 3 and 21 MLT. The relative flux decrease at these nightside MLTs at geosynchronous orbit is only $15%, which is much smaller than that obtained by increasing the pressure from 4 to 10 nPa above. The relative flux decrease at 6 and 18 MLT is even lower and becomes significant only beyond $7.1 RE. The relative flux drop for the two non-equatorial locations is shown in Figure 13 (bottom). As for Figure 12, the dotted portion of the lines indicates the region where the first adiabatic invariant conservation determines the observable pitch angle range at the non-equatorial spacecraft, which determines the omnidirectional flux. For ZSC = 0.5 RE, this effect is nonnegligible even before the drift loss effect begins to play the role at XSC $ 6.8 RE: The relative decrease at XSC $ 6.6 RE is $35%. The result for ZSC = 0.3 RE is similar to that for the equatorial spacecraft. [36] So far we have assumed n = 1 in the pitch angle distribution function in equation (3) but previous studies have shown that n is highly dependent on local time, L-shells and energies due to different acceleration and loss processes of energetic electrons [e.g., Vampola, 1998; Selesnick and Blake, 2002; Gannon et al., 2007]. The lower the value of n, the less will be the relative number of equatorially mirroring particles, and the lower will be the resulting drift loss. Results as a function of n are displayed in Figure 14, which shows the relative flux decrease at 6.6 RE and 24 MLT between PD = 4 to 10nPa for three vertical locations ZSC as a function of the exponent n in the pitch angle distribution function. It is clearly seen that choice of a higher n leads to a higher relative flux drop. The case of n = 0 corresponds to the minimum value of the relative flux drop, which is $56%, and the flux decrease increases to $97% for n = 10. A future study is necessary to determine what n value is the most realistic one, but some observations have shown that n$1 is a reasonable value [e.g., Baker et al., 2001]. [37] It is interesting to note a recent study by Gannon et al. [2007] who, on the basis of CRESS MEA observation,

11 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

pitch angle electrons due to the pitch angle-dependent shell splitting effect, leading to a minimum flux around 90 on the nightside after 1 drift while the higher flux around 90 on the dayside may correspond to lower L-shells that are not affected by the drift loss. Fritz et al. [2003] also reported that butterfly distributions at higher L-shells are associated with drift loss through the magnetopause.

5. Discussion
5.1. Influence of Drift-Shell Bifurcation on Particle Loss [38] In section 3 we have mentioned the role of drift-shell bifurcation (DSB) on the orbits of particles which come close to the magnetopause. The causes and effects of DSB have been investigated in detail by Ozturk and Wolf [2007]. In the context of particle loss, DSB has two important consequences. First, a particle that undergoes DSB on a closed shell will not drift into the magnetopause. Their trapping will not be affected by DSB. In contrast, some particles that would drift directly into the magnetopause in the absence of DSB, that is, the particles on an originally open shell, are diverted away from the magnetopause by DSB. It may later drift into the magnetopause and be lost. This depends on the new second invariant it gains just after the bifurcation, or on subsequent drift orbits. However, all bifurcating particles on open shells will be lost eventually. Therefore DSB reduces the rate of drift loss. [39] The second consequence is related to the fact that DSB alters the second invariant. As Figure 6 shows, some particles drift into the magnetopause after bifurcation. However, another particle on the same initial drift shell but with a different bounce phase could have stayed inside the magnetosphere. It is computationally hard, though not impossible, to predict how a given particle will behave, as this will require following the rapidly-changing bounce phase. Such a test will be done in a future work. The particles that do not drift into the magnetopause after bifurcation will eventually return to the night side, make another drift around the Earth and undergo DSB once more. A fraction of such particles acquire a large second invariant and drift into the magnetopause. Therefore, in the long run, all such particles are lost. Drift-shell bifurcation puts such particles on hold and causes a slow release over many drift periods. The readers are referred to the work by Ozturk zturk and Wolf [2007] for more detailed [2004] and O discussion. In real situations, electrons can also be lost from the magnetosphere by other loss processes such as atmospheric precipitation by the wave-particle interaction before they encounter the magnetopause by the radially outward shift caused by DSB over the long run. The actual rates of loss, depending on magnetospheric conditions, can be available only from a more comprehensive study in future. 5.2. Energy Dependence [40] We have computed the drift orbits for 5 MeV electrons to check if there is a difference between different energies. We have confirmed that this higher energy gives the same overall results as those obtained for 1 MeV electrons. Minor differences exist in relation to the fact that high-energy electrons drift faster than lower energy electrons, but both drift along the same drift shell. The drift time

Figure 13. Same as in Figure 12 except that the two different solar wind conditions are chosen to be IMF BZ = 0 and 15 nT, respectively, with PD = 4 nPa fixed in both cases. reported that the pitch angle distribution of energetic electrons on the dayside are predominantly 90-peaked distributions (corresponding to a maximum flux around 90), while butterfly distribution (corresponding to a minimum flux around 90) are commonly observed at higher L-shells on the nightside. The observed butterfly distribution could be the result of drift loss of near-equatorial pitch angle electrons. As discussed above, this is simply because higher pitch angle elections, when launched on the nightside, are more likely to encounter the magnetopause than are lower

12 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

Figure 14. Relative omnidirectional flux decrease (%) at 6.6 RE and 24 MLT as a function of n, the parameter in the pitch angle distribution function in equation (3). calculated for PD = 4 nPa is about 10$14 min for 1 MeV and 2$4 min for 5 MeV, which depends on test particles pitch angle. Naturally, there should be a time difference in the loss process between 1 MeV and 5 MeV. For a given energy the drift speed is faster under the condition of a stronger pressure and a more southward IMF BZ, which is due to a stronger gradient B in the magnetosphere with increasing pressure and more southward IMF BZ. A major difference due to different energies occurs for electrons that experience the drift shell bifurcation [Ozturk, 2004; Ozturk and Wolf, 2007]. A set of particles on the same drift shell scatter over neighboring shells after bifurcation. At higher energies, the root-mean-square spread is larger. Therefore the likelihood of loss for a bifurcating particle is larger for a particle with a higher energy. The rate of loss is also larger.

6. Summary and Conclusion

[41] In this work we have estimated the extent to which drift loss contributes to the total loss (or flux dropout) within the outer radiation belts and the solar wind conditions under which the drift loss can be a significant process. We have determined explicitly the dependence of the drift loss effect on particle pitch angle and spacecraft location versus MLT, radial and vertical distances for various solar wind conditions using a Tsyganenko model. It is shown that for a stronger PD and a more southward IMF BZ, the dayside magnetosphere is more compressed and the nightside is more stretched due to a tail current, resulting in a more intense day-night asymmetry, which thus leads to drift loss to the magnetopause by drift shell distortion. More specific results are summarized as follows. [42] 1. The drift loss effect, which is in general more likely for a higher pitch angle and nearer the midnight region, spreads to low pitch angles and dusk and dawn MLT regions at geosynchronous orbit as the dynamic pressure increases for a fixed IMF BZ. [43] 2. This is much less significant when IMF BZ becomes more southward for a fixed dynamic pressure.

However, this conclusion is limited by the shortcoming that the Tsyganenko model used here does not reflect the magnetopause response to changing IMF values. [44] 3. The trapping boundary, which is defined here to be the outer most boundary where relativistic electron orbits are closed, differs between different pitch angles: For a given MLT and a given solar wind condition, its radial location is more outward for a lower pitch angle. As the dynamic pressure increases and/or IMF BZ becomes more southward, the trapping boundary for all pitch angles overall moves inward. [45] 4. The result in 3 implies that a spacecraft at a given location will see a narrower pitch angle range with increasing PD or increasingly southward IMF BZ due to the inward movement of the trapping boundary. On the basis of this idea, we have suggested a range of pitch angle for electrons to remain to be trapped as a function of PD and IMF BZ at a fixed location. For example, the range of pitch angle for electrons to remain to be trapped at 6.6 RE at 24 MLT reduces from 0 85.5 to 0 33.4 when the dynamic pressure increases from 4 to 20 nPa with a fixed IMF BZ = 0nT. [46] 5. We have also estimated the drift loss effect at locations away from the equator. Specifically, we have suggested a range of equatorial pitch angles for which electrons can travel up to the point (XSC = 6.6 RE, ZSC = 0.5 RE) at 24 MLT by considering both first adiabatic invariant conservation and the drift loss effect. The calculations were done for various PD for a fixed IMF BZ and vice versa. The results show that for PD values of 10 nPa or larger with IMF BZ = 0 nT, the pitch angle range is determined by the drift loss effect at the equator while it is the first adiabatic invariant conservation for PD = 4 nPa even if IMF BZ is as small as 15 nT. [47] 6. One important aspect that has been considered in the present work is effect related to bifurcation physics. By taking this effect into account in the particle tracing, we were able to correctly determine whether a bifurcating electron as approaching the magnetopause is simply lost into the magnetosheath or remains to be trapped after bifurcation. [48] 7. On the basis of all the results above, we have calculated omnidirectional flux and the relative decrease of the flux between two different solar wind conditions. [ 49 ] 7a. An example where the dynamic pressure increases from 4 to 10 nPa (with a fixed value of IMF BZ = 0) indicates a large relative flux decrease of $60 70 %, when measured by a nightside (at 24 MLT) spacecraft at the equator at geosynchronous altitude. This result is based on the assumption of the directional flux having the form of sinnao with n = 1. With other values of n, the relative flux decrease ranges from $56% at its minimum (when n = 0) to $97% (when n = 10). However, the effect rapidly declines inside geosynchronous orbit. [50] 7b. A qualitatively similar, but quantitatively different, result has been obtained for two different conditions of IMF BZ (IMF BZ = 0 and 15 nT with PD = 4 nPa fixed in both cases): The relative flux decrease between the two solar wind conditions is substantially lower, $15% at the equator, 24 MLT and geosynchronous altitude, which however increases rapidly outward. [51] 7c. For the two examples above, the drift loss effect at nightside MLT regions away from midnight starts to occur at larger radial distances than at 24 MLT and the

13 of 14

A09212

KIM ET AL.: RELATIVISTIC ELECTRON DRIFT LOSS

A09212

relative flux decrease becomes overall weaker when it is away from midnight. [52] 7d. Our results presented here are based on the T02 magnetospheric model. Use of a different model can of course lead to quantitatively different results to some extent. The extent to which the results can differ for different models should be determined in a future work using more than one realistic model. [53] It is difficult to fully explain the observed flux dropouts by considering only one loss process. A full understanding of observed flux dropouts requires consideration of all the possible loss and acceleration processes. For example, the drift loss effect is closely connected to the Dst effect and the substorm growth phase-like (dropout) effect, which are due to time-dependent global and local field stretching, respectively. That is, as mentioned in the introduction section, stretching will lead electrons to move outward, and thus enhances the chance for such particles to encounter the magnetopause. The same stretching can also cause scattering of the particles pitch angle into lower angles. Or similar scattering may occur by wave-particle interactions. Such pitch angle scattering may actually decrease the drift loss rate. All these effects need to be incorporated. Moreover, atmospheric precipitation should also be considered since observations have shown that this is an important mechanism at lower L-values. Therefore estimating combined effect of these processes and relative contributions of each process to flux dropouts under a given solar wind and magnetospheric condition is our ultimate long-term goal. [54] Acknowledgments. This work at Chungbuk National University was supported partly by a grant from Korea Astronomy and Space Science Institute in 2006 and also partly by a grant (R01-2007-000-10674-0) from the Korea Science and Engineering Foundation. Work at UCLA was supported in part by NSF grant ATM-0646233. [55] Amitava Bhattacharjee thanks Mary Hudson and another reviewer for their assistance in evaluating this paper.

References
Baker, D. N., S. G. Kanekal, J. B. Blake, and T. I. Pulkkinen (2001), The global efficiency of relativistic electron production in the Earths magnetosphere, J. Geophys. Res., 106(A9), 19,169 19,178. Bortnik, J., R. M. Thorne, T. P. OBrien, J. C. Green, R. J. Strangeway, Y. Y. Shprits, and D. N. Baker (2006), Observation of two distinct, rapid loss mechanisms during the 20 November 2003 radiation belt dropout event, J. Geophys. Res., 111, A12216, doi:10.1029/2006JA011802. Brizard, A. J., and A. A. Chan (1999), Nonlinear relativistic gyrokinetic Vlasov-Maxwell equations, Phys. Plasmas, 6, 4548. Desorgher, L., P. Buhler, A. Zehnder, and E. O. Fluckiger (2000), Simulation of the outer radiation belt electron flux decrease during the March 26, 1995, magnetic storm, J. Geophys. Res., 105(A9), 21,211 21,213. Fritz, T. A., M. Alothman, J. Bhattacharjya, D. L. Matthews, and J. Chen (2003), Butterfly pitch angle distributions observed by ISEE-1, Planet. Space Sci., 51(3), 205 219. Gannon, J. L., X. Li, and D. Heynderickx (2007), Pitch angle distribution analysis of radiation belt electrons based on combined release and radiation effects satellite medium electrons A data, J. Geophys. Res., 112, A05212, doi:10.1029/2005JA011565. Green, J. C., T. G. Onsager, T. P. OBrien, and D. N. Baker (2004), Testing loss mechanisms capable of rapidly depleting relativistic electron flux in the Earths outer radiation belt, J. Geophys. Res., 109, A12211, doi:10.1029/2004JA010579. Horne, R. B., and R. M. Thorne (2003), Relativistic electron acceleration and precipitation during resonant interactions with whistler-mode chorus, Geophys. Res. Lett., 30(10), 1527, doi:10.1029/2003GL016973. Imhof, W. L., J. Mobilia, D. W. Datlowe, H. D. Voss, and E. E. Gaines (1990), Longitudinal and temporal variations of energetic electron precipitation near the trapping boundary, J. Geophys. Res., 95(A4), 3829 3839. Kim, H.-J., and A. A. Chan (1997), Fully adiabatic changes in storm time relativistic electron fluxes, J. Geophys. Res., 102(A10), 22,107 22,116.

Kim, K. C., D.-Y. Lee, E. S. Lee, C. R. Choi, K. H. Kim, Y. J. Moon, K. S. Cho, Y. D. Park, and W. Y. Han (2005), A new perspective on the role of the solar wind dynamic pressure in the ring current particle loss through the magnetopause, J. Geophys. Res., 110, A09223, doi:10.1029/ 2005JA011097. Lyons, L. R., D.-Y. Lee, R. M. Thorne, R. B. Horne, and A. J. Smith (2005), Solar wind-magnetosphere coupling leading to relativistic electron energization during high-speed streams, J. Geophys. Res., 110, A11202, doi:10.1029/2005JA011254. Lee, J. J., G. K. Parks, K. W. Min, M. P. McCarthy, E. S. Lee, H. J. Kim, J. H. Park, and J. A. Hwang (2006), Relativistic electron dropouts by pitch angle scattering in the geomagnetotail, Ann. Geophys., 24, 3151 3159. Li, X., D. N. Baker, M. Temerin, T. Cayton, E. Reeves, R. Christensen, J. Blake, M. Looper, R. Nakamura, and S. Kanekal (1997), Multisatellite observations of the outer zone electron variation during the November 3 4, 1993, magnetic storm, J. Geophys. Res., 102(A7), 14,123 14,140. Millan, R. M., R. P. Lin, D. M. Smith, and M. P. McCarthy (2007), Observation of relativistic electron precipitation during a rapid decrease of trapped relativistic electron flux, Geophys. Res. Lett., 34, L10101, doi:10.1029/2006GL028653. Northrop, T. G. (1963), The Adiabatic Motion of Charged Particles, Interscience Publishers, New York. OBrien, T. P., et al. (2001), Which magnetic storms produce relativistic electrons at geosynchronous orbit?, J. Geophys. Res., 106(A8), 15,533 15,544. Onsager, T. G., et al. (2002), Radiation belt electron flux dropouts: Local time, radial, and particle-energy dependence, J. Geophys. Res., 107(A11), 1382, doi:10.1029/2001JA000187. Ozturk, M. K. (2004), Bifurcation of drift shells near the dayside magnetopause, Ph.D. thesis, Rice Univ., Tex. Ozturk, M. K., and R. A. Wolf (2007), Bifurcation of drift shells near the dayside magnetopause, J. Geophys. Res., 112, A07207, doi:10.1029/ 2006JA012102. Press, W. H., S. A. Teukolsky, W. T. Wetterling, and B. P. Flannery (1994), Numerical Recipes in Fortran 77, 2nd ed., Cambridge Univ. Press, New York. Reeves, G. D., R. H. W. Friedel, M. G. Henderson, R. D. Belian, M. M. Meier, D. N. Baker, T. G. Onsager, and H. J. Singer (1998), The relativistic electron response at geosynchronous orbit during the January 1997 magnetic storm, J. Geophys. Res., 103(A8), 17,559 17,570. Reeves, G. D., K. L. McAdams, R. H. W. Friedel, and T. P. OBrien (2003), Acceleration and loss of relativistic electrons during geomagnetic storms, Geophys. Res. Lett., 30(10), 1529, doi:10.1029/2002GL016513. Roederer, J. G. (1970), Dynamics of Geomagnetically Trapped Radiation, vol. 2, Springer-Verlag, New York. Selesnick, R. S., and J. B. Blake (2002), Relativistic electron drift shell splitting, J. Geophys. Res., 107(A9), 1265, doi:10.1029/2001JA009179. Shabansky, V. P. (1971), Some processes in the magnetosphere, Space Sci. Rev., 12(3), 299 418. Shprits, Y. Y., R. M. Thorne, R. Friedel, G. D. Reeves, J. Fennell, D. N. Baker, and S. G. Kanekal (2006), Outward radial diffusion driven by losses at magnetopause, J. Geophys. Res., 111, A11214, doi:10.1029/2006JA011657. Tsyganenko, N. A. (2002a), A model of the near magnetosphere with a dawn-dusk asymmetry. 1: Mathematical structure, J. Geophys. Res., 107(A8), 1179, doi:10.1029/2001JA000219. Tsyganenko, N. A. (2002b), A model of the near magnetosphere with a dawn-dusk asymmetry. 2: Parameterization and fitting to observations, J. Geophys. Res., 107(A8), 1176, doi:10.1029/2001JA000220. Ukhorskiy, A. Y., B. J. Anderson, P. C. Brandt, and N. A. Tsyganenko (2006), Storm time evolution of the outer radiation belt: Transport and losses, J. Geophys. Res., 111, A11S03, doi10.1029/2006JA011690. Vampola, A. L. (1998), Outer zone energetic electron environment update, in Proc. of the Conf. on the High Energy Radiation Background in Space, volume IEEE Conference Proceedings, pp. 128 136, IEEE Nuclear and Plasma Sciences Society and IEEE Inc., Snowmass, Colo., 22 July 1997.

C. R. Choi, K. C. Kim, and D.-Y. Lee, Department of Astronomy and Space Science, Chungbuk National University, 410 Sungbong-ro, Heungduk-gu, Cheongju, Chungbuk 361-763, South Korea. ([email protected]; [email protected]; [email protected]) H.-J. Kim, Department of Astronomy and Space Science, Kyunghee University, Seocheon-dong, Giheung-gu, Yongin-si, Gyeonggi-do 446-701, South Korea. ([email protected]) E. S. Lee, Space Sciences Laboratory, University of California Berkeley, 7 Gauss Way, Berkeley, CA 94720-7450, USA. ([email protected]) L. R. Lyons, Department of Atmospheric Sciences, University of California Los Angeles, Los Angeles, CA 90095-1565, USA. (larry@ atmos.ucla.edu) M. K. Ozturk, Department of Information Technologies, Isk University, : Kumbaba Mevkii, I stanbul, Turkey. ([email protected])

14 of 14