Magnetospheric Multiscale Statistics of High Energy Electrons Trapped in Diamagnetic Cavities

High‐energy electrons observed in the magnetosheath must be accelerated by some mechanism that is as yet undetermined. We present observations of high‐energy electrons trapped in diamagnetic cavities as measured by Magnetospheric Multiscale from 2015 to 2018. The observations support the notion of local acceleration in the reconnection quasi‐potential as many of events show particles with pitch angles that are increasingly closer to 90° with increasing energy. It is suggested that these particles can end up in the loss cone and be transported to the magnetosheath. We also characterize each diamagnetic cavity as formed due to low‐ or high‐latitude reconnection based on prevailing solar wind conditions. The character of the ions in the diamagnetic cavity is only briefly mentioned as their properties warrant another stand‐alone investigation.

size and structure of the cusp DMC has been studied Dunlop et al., 2006;Nykyri et al., 2011) as well as its dynamics and response to solar wind driving (Cai et al., 2015;Fuselier et al., 2003;Nykyri et al., 2004;Pudovkin et al., 1992). The cusp DMC, which consists of old reconnected field lines, is characterized by depressed magnetic field and high plasma beta , while the cusp proper consists of recently reconnected field lines, and is characterized by openfield particle precipitation (Wing et al., 2001). The stagnant exterior cusp (SEC) (Lavraud et al., 2002;Zhang et al., 2005) is a region with a similar generation mechanism as the cusp DMC but bounded by a gradual transition from the surrounding magnetic field strength to the depressed region, as opposed to the relatively sharp boundaries of the cusp DMC. The SEC, which has been observed during a period of low solar wind dynamic pressure, also contains little or no low frequency fluctuations (Lavraud et al., 2002), in contrast to the higher levels of magnetic field fluctuation observed inside the cusp DMC.
This study reports MMS observations of DMCs filled with a population of trapped high energy (>40 keV) electrons. We focus on electrons for this study since their motion is more adiabatic than the ions and also because the ion population in the cavities can take on an entirely different character than the electrons. While the orbit of MMS is largely equatorial it can reach higher geomagnetic latitudes during periods of large dipole tilt and close to spring and fall equinoxes, which is where many of the events found in this paper occur.

MMS Instrumentation
Observations in this paper are presented from the instruments onboard MMS2 and MMS3 . The flux gate magnetometer (FGM) gives DC magnetic fields Torbert et al., 2016). The electric field double probes (EDP) provide 3-axis electric fields Lindqvist et al., 2016;Torbert et al., 2016). The fast plasma investigation (FPI) gives phase space density distributions (Pollock et al., 2016) from which numerical moments can be calculated. The energetic ion spectrometer (EIS) provides pitch angle distributions for energetic ions and electrons  in an energy range from 10s of keV to ∼1 MeV. We note that no FPI data is available for the period February 10, 2017 to May 2, 2017 during which time MMS encounters at least 3 events of interest, although we are still able to perform most of the analysis for these events using the hot plasma composition analyzer (Young et al., 2016).

Diamagnetic Cavity Observations
Earth's geomagnetic cusp can be characterized by a portion of field lines attached at one end to the solar wind with the other end in the ionosphere, and a second portion of roughly dipolar field lines which define the boundary of the cusp. Magnetic reconnection with the solar wind opens these dipolar field lines, which can occur at low latitudes, for instance when the interplanetary magnetic field (IMF) is strongly southward, or at high-latitudes, such as during northward IMF, or as a result of the Kelvin-Helmholtz instability (KHI). These different scenarios are depicted schematically in Figure 1, which shows reconnection occurring at high-latitude (a) and low-latitude (b) and the subsequent formation location of the cusp diamagnetic cavity shown with green highlighting. The cavity forms where old reconnected flux is advected away from the reconnection site (blue ×) in the direction of the reconnection outflow (blue arrows). As confirmed from the simulation results in Adamson et al. (2011);Adamson et al. (2012); Nykyri et al. (2011), the cavity forms sunward of the cusp for southward IMF (B z < 0) and anti-sunward of the cusp for northward IMF (B z > 0). A strong positive (negative) IMF B y would shift the DMC to the dusk (dawn) sector at the northern cusp and to the dawn (dusk) sector at the southern cusp.
The cusp diamagnetic cavity is characterized by depressed magnetic field, enhanced plasma density and pressure, and stagnant plasma flows . The magnetic configuration is a magnetic Figure 1. Cartoon representation of the formation of the cusp diamagnetic cavity due to high-latitude (a) and low-latitude (b) reconnection. The magnetopause is shown with a light blue field line, while open and closed field lines are colored red and black, respectively. The reconnection outflows are depicted with blue arrows. Image reproduced and altered from Nykyri et al. (2011). bottle that can trap particles between regions of stronger field. Trapped particles whose drift paths coincide with the gradient of reconnection quasi-potential can be energized and the energy gain depends on how long the particles are trapped . In this study, we have created a database of diamagnetic cavity encounters by MMS from 2015 to 2018 where the cavity also contains a trapped population of high-energy (≥40 keV) electrons. Based on our survey MMS encounters DMCs more times than presented in this paper but during some of these encounters trapped high-energy ions but no trapped high-energy electrons were observed, which will be a topic for a future study. We also note that, MMS observed trapped high-energy ions during many of the events presented in this study, but we will focus the discussion on the electron observations as mentioned in Section 1.
Two examples of DMCs are given in Figure 2. During both events in Figure 2, MMS2 encounters a strongly depressed magnetic field region (second panels) that is coincident with slower moving plasma (third panels).
The plasma flow variability also shows more fluctuation inside the cavity than outside. The fourth panels show IMF conditions propagated from L1 to the bow shock as reported by OMNIWeb (https://omniweb.gsfc.nasa. gov/). The example on the top occurs when the IMF has a steady strong B z < 0 while for the bottom example the IMF is initially steady for the first two encounters and then begins to fluctuate and rotate from (B x , B y , B z ) = (−1,10,0) nT to (−5,4,−3) nT during a third foray of MMS into the cavity. The final panels show with a logarithmic colorbar the electron pitch angle distribution in the 39 keV channel from the EIS instrument. Distributions that are peaked at 90 degrees suggest a trapped population, when assuming adiabatic electron motion. A trapped high-energy (>40 keV) electron population is present inside each DMC in this study. We discuss these distributions more in Section 5.
It is straightforward to estimate a linear dimension of the diamagnetic cavity using the multi-spacecraft timing method (Paschmann & Schwartz, 2000). For the examples in Figure 2, the maximum spacecraft separations are 163 km for the top event and 55 km for the bottom event. For the September 9, 2015 event, we apply the timing method to the crossings into the cavity at ∼12: 49 and out of the cavity at ∼12: 59. The boundary normal unit vectors n 1 and n 2 , corresponding to the boundaries crossed at (1) 12: 49 and (2) 12: 59 have a dot product n 1 ⋅n 2 = 0.95, which indicates it is a reasonable approximation to take the distance of the path of MMS through the cavity as vt. We estimate the velocity of the cavity (v) as an average of the boundary normal velocities (v 1 and v 2 ) calculated with the timing method v = (v 1 + v 2 )/2 and t is the amount of time MMS spent in the cavity. The timing method gives v 1 = 35 km/s and v 2 = 23 km/s and taking t = 10 min we find the length of the path of MMS through the cavity to be ∼3 R E .
When we perform the timing analysis on the September 20, 2015 event, we find linear dimensions of ∼0.25 R E for the encounters occurring from 11:02-11:08 and 11:10-11:15. This suggests either that MMS goes through the cavity at a location where its dimension is short in the direction of the spacecraft motion or that the motion of the cavity sweeps MMS quickly into and out of it. In either case this would likely be an underestimation of the actual size of the cavity. The gyroradius of a 40 keV electron in a 15 nT magnetic field is ∼40 km which is in agreement with our interpretation of electron trapping even for cases where the cavity size is underestimated. In contrast, the proton gyroradius at an energy of 40 keV is ∼2,000 km which can be of the same order of magnitude BURKHOLDER ET AL.  as the size of the cavity. We will discuss briefly the presence of trapped high-energy ions in these DMCs in Section 5.
It is important to note that the length of the path, or the apparent linear dimension through the cavity, is highly dependent on the orientation of the orbit with respect to the cavity as well as due to motion of the cavity (e.g., due to dynamic pressure variations). Both simulations and observations have shown that the structure of the cusp DMC is elongated Nykyri et al., 2011). DMCs with dimensions of as large as 6 R E have been reported using data from Polar (Fritz et al., 2003), although the high-altitude cusp region and associated boundaries were at the apogee of the Polar orbit which could lead to an overestimation of the cavity size since the spacecraft spends a longer time in the cavity.
The list of events used for this study is given in Table 1. Each row represents a given time interval during which MMS passed through a diamagnetic cavity (characterized by depressed magnetic field, and trapped high-energy electrons). The eighth row corresponds to the event studied by Nykyri et al. (2019). During some intervals the cavity is encountered multiple times (column 4), which could be consecutive passes into and out of the same cavity due to large scale magnetospheric motions, or could be different DMCs corresponding to multiple, intermittent reconnection sites. The fifth column of the table lists the events from Cohen et al. (2017) which MMS observed near the same time frame as the cavity encounter. It is possible that these high-energy electrons have their origin in the diamagnetic cavity, having been accelerated by the gradient of reconnection quasi-potential and subsequently leaking into the magnetosheath. The escape of trapped particles from the diamagnetic cavity can occur as a result of a magnetospheric reconfiguration (change of IMF) or plasma instabilities (KHI) that can scatter particles into the loss cone. In fact, it is possible that this is exactly what MMS observed during the 9/9/15 event in Figure 2. The flux of 39 keV electrons during the cavity encounter from 11:02 to 11:08 is an order of magnitude higher than the flux during the encounter from 11:10 to 11:15, suggesting that particles began leaking rapidly once changes of the magnetic topology were initiated by variations of the IMF.
The final four columns of Table 1 present some evidence for the formation of the cavity as a result of magnetic reconnection. The "jets" column indicates that strongly jetting plasma flow channels are observed as MMS passes through the boundary of the DMC, which is the case for at least 1 of the cavity encounters when the table indicates "yes" (see ∼12: 59 for the September 9, 2015 event and ∼10: 59 for the September 20, 2015 event). These jets are indicative of the reconnection outflows predicted in the standard Sweet-Parker (Parker, 1957;Sweet, 1958) and Petschek (1964) models of magnetic reconnection. A sequence of data where the plasma velocity v and Alfvén velocity V A satisfy Δv = ±ΔV A , otherwise known as the Walén relation (Sonnerup et al., 1987(Sonnerup et al., , 1995, at the boundary of one of the cavities is given in the seventh column. The corresponding slope of the fit and correlation coefficient are given in the last two columns, similar to the test of the Walén relation for the diamagnetic cavity observed by Nykyri et al. (2019). The rows marked "no" lack an observation of changes in the plasma and magnetic field that satisfy the Walén relation, which does not rule out the possibility of a reconnection origin for the cavity as the boundaries could have been influenced by waves or other fluctuations. The final 3 rows which have italicized text are during the period February 10, 2017 to May 2, 2017 for which there is no FPI data available. For these cases, we have used the Hot Plasma Composition Analyzer (HPCA) instrument to examine qualitatively the plasma bulk flow and have chosen not to perform any test of the Walén relation since the measurement cadence is significantly slower than the FPI instrument. Fortunately, using the HPCA instrument we were able to identify jets in the H+ flow measurements similar to the rest of the cavities.

Magnetopause Reconnection Location During DMC Observations
While most past observations of the cusp DMC have explained its formation through the high-latitude reconnection scenario like Figure 1a, Nykyri et al. (2019) presented the first MMS observation of high-energy particles trapped in a cusp DMC that formed as a result of low-latitude reconnection, similar to Figure 1b. Therefore, we sort each individual cavity from Table 1 column 4 based on the solar wind conditions at the bow shock nose as predicted by the OMNI. The following simplified predictions for the expected magnetopause reconnection location do not take into account any fluctuations or other possible structure in the solar wind that is not measured by OMNI. Figure 3 illustrates the location of the cavities from rows 6 (top left) and 15 (top right) in Table 1. The black dots give the coordinates of the cavity encounters projected onto the GSM y-z plane and also provided is a reference grid with cell width 5 R E (Earth radii). The TS96 magnetic field lines, which are colored blue at low-latitude and red at high-latitude, provide magnetospheric context and have been calculated using the Orbit Visualization tool. Each DMC observation is also labeled with a vector which represents the IMF direction at Earth's bow shock as predicted by OMNI for the time when MMS was inside the cavity. The vector colors indicate whether the formation of the cavity can be associated with low-latitude (blue, example on 9-20-15) or high-latitude (red, example on 4-23-17) reconnection based on where the maximum magnetic shear occurs. The histograms below these examples show the geomagnetic latitude of all the events, showing that many occur at 20 degrees or greater. The different colored histograms correspond to the different predicted magnetopause reconnection locations, with the blue histogram showing the events where low-latitude reconnection is expected and the red histogram showing events where high-latitude reconnection is expected (based on the B y and B z components of the IMF as predicted by OMNI) . The green histogram shows a few events where the IMF orientation does not clearly correspond to low-or high-latitude reconnection but lies in between. The histograms show some events lie within 10 degrees of the magnetic equator, which would suggest a new type of DMC that is not associated with the cusp. These events are still important in the present study because they are filled with high-energy trapped electrons which may subsequently escape to the magnetosheath or magnetosphere.

Particle Acceleration and the Fate of Energized Particles
The population of trapped high-energy electrons observed in the DMCs could have been accelerated due to any of the different processes as mentioned in the Introduction. However, local acceleration in the cavity by the reconnection quasi-potential leaves a telltale signature in the accelerated population, where higher energies have pitch angles more strongly peaked at 90°, since the first two adiabatic invariants are roughly conserved for electrons . Under the assumption of adiabatic particle motion, electrons could not come to the cavity from a higher field region and have a 90° PAD without some re-processing that would break the adiabaticity of the electron motion. For instance, the particles could have been accelerated BURKHOLDER ET AL.

/A N/A N/A
Italics indicate time periods where FPI data is not available, so the HPCA instrument is used to qualitatively examine the plasma flows (no Walén relation test due to poor time resolution).

Table 1 Database of Time Periods Surrounding the Diamagnetic Cavity Encounters Compiled for This Study
close to the reconnection null-point, which is characterized by even smaller magnetic field, and eventually end up in the cavity. However, close to the X-line the parallel electric fields would increase the parallel component of the electron velocity making them more field-aligned as opposed to a 90° PAD. We identify this characteristic of local acceleration by comparing the PADs for the hierarchy of energy channels of the EIS instrument. Examples of this are given in Figure 4, which shows the same two time intervals as Figure 2. The top panels show the magnetic field strength and the next five panels give the electron PADs for first five energy channels from the EIS instrument. The last five panels show the ion PADs from the first five energy channels of the EIS instrument.
The example on the top of Figure 4 shows 39 keV electrons trapped in the diamagnetic cavity (12:50-13:00) and a population of electrons in the 68 keV channel that is more strongly peaked at 90° than the 39 keV channel. Just before MMS encounters the trapped population it sees strongly field-aligned counterstreaming electrons at the same energies. During the encounter from 12:50 to 13:00 there are very low electron counts in the 114 keV and above channels, which is different from the ions observed during this event. The ion panels also show a complex structure of PADs inside the cavity with significant counts at energies detectable up to the 188 keV channel. The ion flux enhancement can be found around 180 degrees at ∼12:40 UT, 0 degrees at ∼12:50, and ∼90 degrees at ∼13:00. The variations in the high-energy electron and ion PADs observed during a period of relatively stable IMF suggest the influence of KHI on the magnetopause or some other type of dynamics in the cavity.
The observation in the bottom panel of Figure 4 shows MMS encountering a diamagnetic cavity multiple times in succession. In the period 11:00-11:20 MMS falls into the DMC three different times as indicated by BURKHOLDER ET AL.
10.1029/2020JA028341 6 of 12  Table 1 showing the location of the cavities (black dots) in the GSM y-z plane with TS96 field lines. The unit vectors emanating from each black dot give the OMNI prediction of interplanetary magnetic field (IMF) direction at the time of the DMC observation. The color of the vectors indicates whether the cavity is consistent with having been formed by low-latitude reconnection (blue) or high-latitude reconnection (red). The reference grid spacing is 5 R E . The histograms in the bottom row show all of the events from column 4 of Table 1 binned by the magnetic latitude of the encounter. The different colors indicate the expected magnetopause reconnection location based on the OMNI IMF prediction and location of MMS (blue: low-latitude, red: high-latitude, green: unclear).
the depressed magnetic fields coincident with fluxes peaked at 90°. During the first, very short encounter (10:59-11:01), a strong peak of trapped electrons is only observed in the 39 keV channel but the subsequent 2 encounters show electron populations that are more strongly trapped with increasing energy up to 114 keV. The different populations observed in successive encounters may be indicative of the time evolution of the trapped population, which, for this case we can understand as originating from a changing IMF. In this example, the ions exhibit similar behavior to the electrons, generally being trapped more strongly with higher energies up to the 188 keV energy channel, although, the ions appear to respect the boundaries of the cavity less, in accordance with their larger gyroradius. Both of these examples show electron PAD observations which are similar to the case study by Walsh et al. (2007Walsh et al. ( , 2010, who concluded the acceleration process occurred locally inside the cavity.   Figure 5 demonstrates the importance of the high-energy population versus the plasma in the FPI energy range (10 eV-30 keV) to an additional diamagnetic effect in the cavity. By additional diamagnetic effect we mean the difference in diamagnetism of the plasma inside the cavity versus outside the cavity. We have omitted the 16 events in the final 3 rows of Table 1 because HPCA does not provide electron data. To quantify the contribution of the high-energy population, we assume a total pressure balance between the plasma in the energy range of FPI, the plasma outside of this energy range, and the magnetic pressure: where FPI e P is the electron plasma pressure measured by FPI, FPI i P is the ion plasma pressure measured by FPI, P B is the magnetic pressure, and P* is the total contribution to the pressure from the plasma outside of the energy range of FPI, which, when inside the cavity, we assume is mostly contributed by particles with energies greater than 30 keV, and P* = 0 outside the cavity. This assumption of pressure balance ignores Reynold's stresses and the magnetic tension force from the curvature of the magnetic field. The horizontal axis of the right side of Figure 5 is the ratio of total particle pressure FPI FPI e i P P  inside versus outside the cavity measured in the energy range of FPI. The vertical axis is the ratio * FPI FPI / ( ) e i P P P  inside the cavity. Inside quantities are an average through the whole cavity and outside quantities are a finite length average as well. The cluster of events in the top left and bottom left quadrants have an additional diamagnetic effect that is contributed entirely by the high-energy population, since the thermal pressure is actually smaller inside the cavity than outside. Events in the top left quadrant have a particle pressure inside the cavity that is dominated by the high-energy distribution. In the bottom right quadrant both the high-energy and thermal plasmas are responsible for an additional diamagnetic effect with a cluster of events where the high energy population is nearly inconsequential. BURKHOLDER ET AL.  The top left panel of Figure 6 shows a histogram of the maximum energy channel for the EIS instrument in which trapped electrons were observed (E max ) inside the cavity, where E max also requires that all previous energy channels have PADs which become more strongly peaked at 90 degrees with increasing energy. The electron energy channels in increasing order are 39, 68, 114, 229, and 657 keV. Those DMC observations with 90° PAD electrons in 2 or more channels are strongly suggestive as having been accelerated locally. There are 10 cavity observations where trapped electrons were only observed in the first channel. These high-energy electrons cannot be ruled out as having been locally accelerated since the energy channels of the EIS instrument cover a range and the amount of acceleration in the DMC depends on how long the electrons are trapped (i.e., these particles may have only been trapped in the DMC for a short amount of time before being observed by MMS). The histogram on the top right gives the flux of particles trapped in the DMC observed by MMS at the energy channel E max . A scatter plot of E max vs the flux at E max is well fit by a linear regression, indicating the lower flux cases correspond to the higher values of E max , and vice versa for the higher flux cases. This suggests particle leakage is a regular and perhaps continuous occurrence during the acceleration process. The bottom left histogram bins observed electric field properties to show the contribution of the reconnection quasi-potential to the electron energization. The histogram bins the standard deviation (for all observations inside the cavity downsampled to the FPI fast mode data rate) of the absolute difference of the magnitudes of v × B and E, where v is the bulk ion velocity observed by FPI, B is the magnetic field vector observed by FGM, and E is the electric field vector as measured by the EDP instrument. The majority of events have a standard deviation of the difference that is ∼1 mV/m. The final histogram in Figure 6 shows the maximum value of |E| observed by EDP inside the cavity (with no downsampling from the EDP fast mode data rate). While the values of standard deviation are generally very small compared to the maximum value of electric field, this is not an apples to apples comparison given the much higher data rate from EDP. However the very small values of standard deviation suggest a dominance of the v × B electric field inside the cavity which is consistent with the electric field analyzed by Nykyri et al. (2012). We did not include the 16 events in the final 3 rows of Table 1 in this histogram because of the low measurement cadence of HPCA and very fine structure of electric fields observed inside the cavities.
When the energized electrons escape from the DMC, they can end up in the magnetosheath, where their high energy easily sets them apart from the ambient population, or in the magnetosphere, where typical energies are already 10s of keV. In the magnetosheath, Cohen et al. (2017) identified electrons with energies ≥ 40 keV, a number of which were observed in the vicinity of the DMCs in this investigation, as discussed in Section 3 (see Table 1 column 5). The list of events compiled by Cohen et al. (2017) Table 1. The left shows histograms of magnetic field depression as the ratio of magnetic field strength inside the cavity versus outside the cavity. The blue histogram takes the minimum magnetic field strength inside the cavity while the orange takes the average magnetic field strength inside the cavity. The scatter plot has on its x-axis the ratio of total particle pressure inside the cavity versus outside as measured by Fast Plasma Investigation (FPI). The y-axis gives the ratio of P* to the total particle pressure as measured by FPI inside the cavity. these events to understand whether they can be attributed to leakeage from the cusp DMC at higher latitudes. The left side of Figure 7 shows the three Cohen et al. (2017) events that were observed by MMS very near the DMC from row 9 of Table 1. The red × symbols show the locations of the Cohen et al. (2017) events projected in the GSM y-z plane and reference TS96 field lines are also given. The black dot gives the location of the DMC from row 9 of Table 1. The histogram in the right panel of Figure 7 bins all of the 238 events from Cohen et al. (2017) with respect to geomagnetic latitude. The distribution peaks around 25 degrees southern latitude, which is similar to the distribution of DMCs from Figure 3. The distribution also contains far more events at latitudes higher than 25 degrees as opposed to lower.

Discussion and Conclusions
A manual inspection of all MMS observations from 2015 to 2018 revealed a total of 44 encounters of a DMC filled with high-energy electrons. Many of these encounters occur at high-geomagnetic latitudes which is also the case for the high-energy electron observations reported by Cohen et al. (2017). While MMS does not often reach the high-latitude region where the cusp DMC forms it is interesting that we have found a few observations of low-latitude DMCs. All of the DMCs in this paper are characterized by a depressed magnetic field and the presence of a population of high-energy electrons with PADs strongly peaked at 90°. BURKHOLDER ET AL.
10.1029/2020JA028341 9 of 12 While Figure 6 shows strong evidence for local acceleration of these high-energy electrons inside the cavity, it is possible that the acceleration occurs in the vicinity of the electron diffusion region and their PAD is modified by electron-scale plasma waves so that they end up trapped in the cavity. Testing this hypothesis will be the topic of a future investigation. Additionally, in many cases the plasma populations observed inside the DMC evolve over a succession of encounters. More work is needed to understand the origin of these changes in the high-energy population that we must carry out on a case-by-case basis.
If the populations of energized electrons trapped inside the DMCs were accelerated from the background magnetosheath energy of 100s of eV, the fluxes observed by the EIS instrument (39-657 keV channels) indicate the gradient of the reconnection quasi-potential must be responsible for an acceleration of 10-100s of keV. The v ×B electric field required to drive this acceleration for a typically sized cavity is 5-10 mV/m . A smaller electric field may also be responsible for the same acceleration if particles are cycled through the cavity multiple times . While the observations in this paper show 5-10 mV/m can be just a fraction of the maximum electric field observed inside the diamagnetic cavity ( Figure 6), the high resolution measurements available with MMS reveal a highly structured electric field inside the cavity. The character of the ions, not fully explored in this paper, can show a variety of different PADs inside a DMC. In addition, other DMC observations not presented here show high-energy electrons with PADs not strongly peaked at 90°, such that the quasi-potential gradient may not be the only mechanism which can energize electrons near or inside the DMC. One possibility is foreshock transients, where Fermi-acceleration can lead to electron energization with isotropic pitch angles (Liu et al., 2017).
The long-standing mystery of high-energy electrons beyond the magnetopause has been considered by other authors through connections to many different magnetospheric phenomena. The nearly continuous occurrence of reconnection around the cusp, the subsequent formation of the cusp DMC, and the acceleration of particles trapped in the DMC, appears to be a likely candidate as the source for some of the mysterious high-energy electrons. The events from Cohen et al. (2017) found in the vicinity of the DMC events in this paper are interesting cases that provide a nearly complete picture of the source of the high-energy electrons. Furthermore, a simultaneous comparison of energetic electron phase-space densities as a function of magnetic moment in the DMCs and in the inner magnetosphere could help resolve the contribution of these DMC electrons to the inner-magnetospheric population.
BURKHOLDER ET AL.   Table 1. Coordinates are projected into the GSM y-z plane and TS96 field lines are given for reference.
The histogram on the right shows the magnetic latitudes of the Cohen et al. (2017) events.