Correlations between plasmapause evolutions and auroral signatures during substorms observed by Chang'e‐3 EUV Camera

The plasmapause locations determined from the Chang'e‐3 (CE‐3) Extreme Ultraviolet Camera (EUVC) images and the auroral boundaries determined from the Defense Meteorological Satellite Program (DMSP) Special Sensor Ultraviolet Spectrographic Imager (SSUSI) images are used to investigate the plasmaspheric evolutions during substorms. The most important finding is a nightside pointing plasmaspheric plume observed at 23:05 UT on 21 April 2014 under quiet solar wind and geomagnetic conditions, which drifted from the dusk sector. High correlations between the plasmapause evolutions and the auroral signatures exist during substorms. After substorm onset, the plasmapause erosion and the equatorward expansion of the auroral oval occur almost simultaneously in both MLT and UT, and then both the erosion and the expansion propagate westward and eastward. It is suggested that the plasmaspheric erosion and its MLT propagations are induced by the enhanced earthward plasma convection during substorm period, and the substorm dipolarization causes pitch‐angle scattering of plasma sheet electrons and the resulting precipitation excites aurora emissions at the same time.


Introduction
The ionosphere-originated cold and dense plasma region in the plasmasphere is the core region of inner magnetospheric couplings (Lemaire et al., 1998). The plasmaspheric responses to storms/ substorms that are triggered by the changes of solar wind and interplanetary magnetic field (SW-IMF) have always been a hot topic in the inner magnetospheric research. Numerous investigations of IMAGE Extreme Ultraviolet (EUV) images Burch, 2003) have revealed that changes in SW-IMF parameters can cause the erosion or refilling of the plasmasphere (Sandel et al., 2001Spasojević et al., 2003;Goldstein et al., 2003aGoldstein et al., , 2005He F et al., 2012aHe F et al., , 2013. Plume is a typical global structure of the plasmasphere under disturbed SW-IMF conditions and in storm times, which can be generally formed in the post-noon to dusk sector as predicted by numerical modeling (Grebowsky, 1970;Chen AJ and Wolf, 1972;Chen AJ and Grebowsky, 1974;Rasmussen et al., 1993;Ober et al., 1997a;Pierrard and Lemaire, 2004) and observed by IMAGE mission (Sandel et al., 2001;Spasojević et al., 2003Spasojević et al., , 2004Goldstein et al., 2003aGoldstein et al., , b, 2004He F et al., 2013), Cluster (Darrouzet et al., 2008(Darrouzet et al., , 2009, CRRES (Moldwin et al., 2004), LANL (Goldstein et al., 2014), and THEMIS (Walsh et al., 2014). Most of the previous works have focused on the study of the plasmaspheric evolutions under strongly disturbed SW-IMF or geomagnetic conditions. During disturbed times, however, the variations of the SW-IMF and the geomagnetic conditions are so complicated that one cannot clearly distinguish which physical process (external or internal) contributes more to the dynamics of the plasmasphere. It is also known that the plasmapause can move toward the Earth during substorm times and substorms may play an important role in the plasmaspheric dynamics (Kwon et al., 2015;. On the other hand, the evolutions of the plasmaspheric structures under quiet SW-IMF and/or geomagnetic conditions have not been fully understood yet. The observation of nightside pointing plume during quiet and less disturbed SW-IMF and geomagnetic conditions has been seldom reported in the literature (Grebowsky, 1970;Ober et al., 1997b).
In addition to the external trigging factors (e.g., SW-IMF), dynamics of the inner magnetosphere (the ring current and radiation belts) and the ionosphere can also impact the evolutions of the plasmasphere. One important signature of the inner magnetosphere is the auroral substorm due to the strong ion and electron precipitations in the ionosphere (Akasofu, 1964). In the substorm onset phase, the aurora intensifies suddenly and moves equatorward and poleward in the premidnight magnetic local time (MLT) (McPherron et al., 1973). The front of the dipolar magnetic field (the so-called "dipolarization front") moves towards the Earth during the substorms, allowing more magnetotail plasma energized and transported into the polar ionosphere and the auroral oval expanded more equatorward (Goldstein et al., 2007;Runov et al., 2009;Ge YS et al., 2012). The strong sunward compression by the substorm dipolarization front also causes the earthward moving of the nightside plasmasphere (Carpenter and Smith, 2001).
The auroral brightening and expansion are the general form of substorm activity which reflects the dynamic of the inner magnetosphere and ionosphere, such as magnetotail dipolarization, particle injection, and particle precipitation. A big collection of global images obtained from the Defense Meteorological Satellite Program (DMSP) Special Sensor Ultraviolet Spectrographic Imager (SSUSI) (Paxton et al., 1992) can be used to study the auroral substorm activities. Additionally, the Chang'e-3 (CE-3) Extreme Ultraviolet Camera (EUVC) landed on the lunar surface is used to image the global plasmasphere at 30.4 nm from side perspectives with a field of view of 15°, an angular resolution of 0.1° and a temporal resolution of 10 min Feng JQ et al., 2014;Yan Y et al., 2016;, and provides us an opportunity to study the global plasmaspheric evolutions during substorms. In this investigation, EUVC and SSUSI data obtained under relatively weak and steady SW-IMF conditions on 21 February and 21 April 2014 will be used to study the relationships between global plasmaspheric evolution and the substorm activities.
Note that the evolutions of the plasmasphere during substorms were investigated with a focus on the variations of the plasmapause at noon and it was inferred that there may be a causal relationship between the substorms and the formation of the plasmaspheric bulges in a previous work . In this investigation, we will explore the global evolution of plasmasphere and its correlation with the auroral signature during substorms.

Overview of SW-IMF and Geomagnetic Conditions
The SW-IMF and geomagnetic conditions on 21 February 2014 (Case I, in red) and on 21 April 2014 (Case II, in blue) are shown in Figure 1, with time shift of 45 min from ACE's location to the dayside magnetopause (the propagation of solar wind is calculated with the method in (Zhang XX et al., 2005). Case I shows that the magnetosphere was recovering from the moderate geomagnetic storm happened on 19 February, associated with Dst around -50 nT and Kp below 3.0. Figure 1b indicates a southward turning of IMF happened at approximately 03:52 UT and lasted until the end of the day. After 04:00 UT, both SW-IMF and geomagnetic conditions were steady, with the amplitudes of B y and B z both below 5 nT, V SW around 500 km/s, and N SW below 2 cm -3 . It can be seen in Figures 1e and 1f that the auroral geomagnetic activity was disturbed after ~ 07:00 UT, as indicated by the AE index, during the recovery phase of the moderate storm in which the solar-winddriven convection (E SW < 2 mV/m) was weak.
Figures 1f-1h show that the magnetosphere had several substorms for Case II, in which B z turned to south at ~01:00 UT and then was back to north at ~03:00 UT until ~08:00 UT when it turned to south again and continued for ~8 hours as indicated in Figure 1b. While as shown in Figures 1c and 1d, there were significant increases in N SW and slow decreases in V SW during these two periods of southward IMF, respectively. The magnetosphere was in a weak solar-wind-driven convection (E SW < 3 mV/m) as indicated in Figure 1e.

Overview of Plasmaspheric Evolutions Seen by EUVC
The EUVC images and the corresponding plasmapause locations determined from the EUVC images using the Minimum L Algorithm (MLA) (He F et al., 2011(He F et al., , 2012b are shown in Figure 2 and Figure 3 in 2-hour intervals for Case I and Case II, respectively. The MLA is a line of sight (LOS) tracing algorithm first proposed to determine the plasmapause locations from the IMAGE EUV images (Wang CL et al., 2007). In optical imaging of the plasmasphere, each pixel in the image represents a specific LOS in space and the intensity is proportional to the column integrated He + density along the LOS. For the pixels at the sharp boundary in a plasmaspheric image, the corresponding LOS's are assumed to be tangential to the plasmapause. When the points in the LOS are traced on the magnetic equatorial plane along magnetic field lines, the one with minimum L-value is determined to be the plasmapause. The MLA has been validated in application to plasmaspheric images obtained in an Earth's polar-orbiting imager (e.g., IMAGE EUV) and the MLT dependence of the plasmapause locations can be determined since the images are approximately projected onto the magnetic equatorial plane. Then the MLA was further developed (He F et al., 2011 to be applied to the plasmaspheric images obtained in side perspectives (e.g., CE-3 EUVC). The plasmapause locations determined from the CE-3 EU-VC images via MLA has been validated with in situ observations from the Defense Meteorological Satellite Program (DMSP), the Time History of Events and Macroscale Interactions during Substorms (THEMIS), and the Van Allen Probes (VAP) at different MLTs . Excellent agreement between the determined plasmapauses seen by EUVC and other satellites indicates the reliability of the data as well as the MLA.
In determination of the plasmapause locations in the equatorial plane of the solar magnetic (SM) coordinate system, the MLA is realized through the following steps .
The first step is to calculate the transformation matrixes between the satellite coordinate system (SCS) and SM based on the unit vectors of the x, y, and z axes of SCS in SM coordinate system given in the header of EUVC data files.
The second step is to automatically extract the plasmaspheric outline pixels defined by the outermost sharp edge where the brightness of 30.4 nm He + emissions drops more than 1 order of magnitude in the radial directions in the EUVC images. Then the pixels are manually checked to ensure that the pixels are located at true plasmapause outline since in some times the noise comes from the contamination by sunlight might mislead the automatic processing procedure. This is necessary to ensure the determined plasmapause locations to be correct and reliable. So the contaminated parts of the plasmapause are not determined, causing gaps in the curves in Figures 2 and 3.
The third step is to calculate the LOS vector in SM. The vector in SCS is calculated based on the pixel size of 0.25 mm and focal length of 150.0 mm for the EUVC, and then the vector is transformed to SM coordinate system using the matrix calculated in the first step.
The forth step is to calculate the L values of all the magnetic field lines intersected by the LOS in SM using the IGRF 2010 internal magnetic field model combined with the Tsyganenko 2007 external magnetic field model (Tsyganenko and Sitnov, 2007;Sitnov et al., 2008). The field line with the minimum L value is identified and the corresponding point on the SM equator is saved for a plasmapause location.
Finally, after all the plasmaspheric outline pixels are processed, a set of discrete plasmapause locations on the SM equator are interpolated with B-spline to form a smooth curve representing the equatorial plane plasmapause, as shown in Figures 2 and 3.
The EUVC was located between ~4.0 h and ~5.0 h MLT for Case I, and between ~5.0 h and ~6.0 h MLT for Case II, respectively, and all the images are projected in the yz plane of the Satellite Coordinate System (SCS) . The EUVC data released by the National Astronomical Observatories are all projected onto the yz plane of the SCS. The x axis of SCS is pointed to the Earth (also the axis of the FOV of EUVC), the z axis of SCS is pointed to Vsw (km/s)  It is noted that the SCS varies with the lunar motion. For each EU-VC image, the unit vectors of the x, y, and z axes in SM coordinates system were calculated during the process of raw data. Then the vectors were included in the header of the released EUVC data files for users to calculate transformation matrix. Figure 2 indicates that the overall size of the plasmasphere was re-   latively small for the whole day during which the plasmasphere was recovering from the moderate geomagnetic storm on 19-20 February. Continuous plasmaspheric refilling started from 00:36 UT till 06:31 UT (Figures 2a-2d), and the plasmapause expanded by approximately 0.5 R E in 0-12 h MLT sector and by more than 1.0 R E in 12-18 h MLT sector. The nightside plasmasphere was then eroded around 07:30 UT while the dayside plasmapause moved sunward, especially in the noon to dusk sector, and an un-wrapped plume was finally formed at 22:30 UT in the dusk in Figure 2l.
It is clearly shown in Figure 2 that the plasmasphere shrank from 00:36 UT to 10:31 UT and expanded after 12:35 UT in the polar region (along the projected dipole axis), caused by the variations of the latitude of the EUVC in SM coordinate system and the generation of the unwrapped plume. Note that the distance of the EUVC to the SM equatorial plane changed from -21.85 R E to -5.65 R E   from 00:36 UT to 10:31 UT and changed from -5.65 R E to -26.26 R E after then. The closer the EUVC is to the SM equatorial plane, the narrower the plasmasphere is exhibited in the vertical direction of the images (e.g., Figure 3 in (He F et al. 2011)). Figure 2l indicates that the plume extended approximately along the vertical direction in the images when the longitudinal direction of the plume on the equatorial plane was approximately parallel to the axis of the FOV of EUVC (He F et al., 2013), resulting in expansion of the plasmasphere after 16:36 UT.
It is shown in Figure 3 that the first small scale erosion began at 0 3:00 UT (Figure 3a) after the sudden enhancement of AE index at 02:00 UT (Figure 1f), and the plasmasphere recovered until 08:00 (e-f) the AE index; (g-h) the keogram-style plot of the PP; (i-j) the keogram-style plot of the L-shell of the EBAO in southern hemisphere (SH), respectively. The red vertical lines represent substorm onsets at UT I =07:00 and UT II =08:06, and the dashed lines marked with EF I and EF II represent the two erosion fronts (EFs) of the plasmasphere, with L I and L II representing the expansion fronts of the auroral oval in SH, respectively. Note that same color bar is adopted for both the PP and the EBAO.
UT (can be seen clearly in Figure 4h). Following the second erosion (Figure 3d) after the substorm began at ~08:00 UT, a wrapped plume was formed after ~09:00 UT and became the most apparent at 13:01 UT (Figure 3f). Then the plume started to corotate eastward around 17:00 UT as a result of the expansion of the corotation region due to the northward turning of the IMF shown in Figure 1. At about 23:05 UT, the plume had drifted to the midnight sector. This is consistent with Grebowsky's model (Grebowsky, 1970) that the plume just corotates and can move from dayside or dusk to nightside during weak convection or quiet time. It is worth noting that the nightside pointing plumes have been seldom reported and studied (Grebowsky, 1970;Ober et al., 1997b), and the generation of plumes at dusk sector and corotation of plumes to the midnight during less disturbed SW-IMF and/or geomagnetic periods needs further investigation.

Determination of Substorm Onset Times
To investigate the correlations between the plasmapause evolutions and the variations of auroral signatures during substorms, it is important to accurately determine the substorm onset times. The all sky imagers (ASI) and the magnetometers data from the Time History of Events and Macroscale Interactions during Substorms (THEMIS) Ground Based Observatories (GBO) (Wanliss, 2006;Mende et al., 2007;Zou S et al., 2010) were used to determine the timing of substorm onsets for these two cases, so as to investigate the correlations between the plasmaspheric evolutions and the substorm activities.
The keogram of the ASI auroral images with 3 seconds' cadence and the Pi1 pulsations (2-40s) calculated from the magnetometer data sampled at 2 Hz at Fort Simpson (FSIM) station (geographic latitude 61.8°, longitude 238.8°; magnetic latitude 67.3°) from 06:30 UT to 07:30 UT on 21 February 2014 are shown in Figures 4a  and 4c, respectively. In both panels, the vertical dashed line indicates the substorm onset time at 07:00 UT located at ~70° magnetic latitude (MLAT) and 22.4 h MLT. After onset, the aurora expanded equatorward and kept disturbed until the end of the day. The auroral brightening was associated with significant Pi1 pulsations as shown in Figure 4c.
The keogram of the auroral images and corresponding Pi1 pulsations at Fort Smith (FSMI) station (geographic latitude 60.0°, longitude 248.2°; magnetic latitude 67.4°) for the substorm happeded after 08:00 UT on 21 April 2014 are shown in Figures 4b  and 4d, respectively. The localized brightening of auroral from 07:16 to 08:00 in Figure 4b corresponded to the pseudobreakup event with no significant equatorward expansion of the aurora oval. After substorm onset at 08:06 UT located at ~65° MLAT and 0.0 h MLT, the aurora brightened significantly and expanded both equatorward and poleward associated with significant Pi1 pulsations in Figure 4d especially after 08:30 UT.

Auroral Signature
The magnetic equatorial plasmapause (PP) locations are extracted from the EUVC images using the MLA. The auroral images at 135.6 nm obtained from the SSUSI onboard the DMSP satellites F16, F17 and F18 are mapped into a magnetic latitude-MLT grid in SM coordinate system. The equatorward boundary of auroral oval (EBAO) was extracted with the boundary recognition algorithm (Cao CG et al., 2009). Both the PPs and the EBAOs are extracted in 1 h MLT intervals when data are available. Since the time resolution for EUVC images is 10 min and the time intervals for DMSP passing the same hemisphere is 30 min to 1 h, the PPs are binned into 30 min intervals and the EBAOs are binned to 1 h intervals to ensure the continuity in UT. Figures 4e-4j show the evolutions of the AE index, the geocentric distances of the magnetic equatorial plasmapause (PP), the MLAT of the EBAO in the southern hemisphere for Case I (left) and Case II (right), respectively. Note that we are not going to focus on the white portion in Figures 4g-4j where no plasmapause location is determined from the EUVC images and/or no EBAO extracted from the DMSP SSUSI auroral images.  Figures 4g and 4h, respectively. It is noted that after substorm onsets, the dayside plasmapause began to move sunward and the dayside EBAO moved towards high latitudes. When the nightside plasmaspheric erosions propagated to the dayside, the dayside plasmapause began erosion too. It is also indicated that the plasmaspheric erosions were apparently modulated by the auroral activities during the substorms since the solar-wind-driven convections were very weak for these two cases as shown in Figure 1.

Discussion
In this paper, we have reported nightside pointing plume which drifted from the dusk sector observed by CE-3 EUVC on 21 April 2014. Several similar nightside pointing plumes were also found in IMAGE EUV images under quiet SW-IMF and geomagnetic activities. This substantial observation of nightside pointing plumes during quiet and less disturbed SW-IMF and geomagnetic conditions (|E SW | < 2 mV/s, AE < 50 nT, Kp < 2, and Dst > -10 nT) is different from the traditional large-scale convection theory of the plasmasphere (Chen AJ and Wolf, 1972;Rasmussen et al., 1993) and has seldom been reported in the literature.
High correlations between the plasmapause evolutions and the auroral signatures during substorms based on CE-3 EUVC and DM-SP SSUSI observations were also found. During the two substorms studied, the plasmapause erosions began at the substorm Earth and Planetary Physics doi: 10.26464/epp2017005 41 onsets around midnight, and then the erosion propagated towards morning side. The tendency of the plasmaspheric erosions was consistent with the equatorward expansion of the auroral oval. When the equatorial boundaries of the auroral oval are mapped onto the magnetic equatorial plane, the averaged L-values of the boundary around the line L I and L II are ~4.0 R E and ~4.3 R E with the corresponding averaged L-values of the plasmapause location around the line EF I and EF II to be ~3.5 R E and ~3.9 R E for Case I and Case II, respectively. This implies that the auroral equatorward boundary approximately matches the plasmaspheric erosion front location. A possible interpretation of such correlations is as follows.
Before substorm onsets, the geomagnetic field in the magnetotail is stretched and energy is stored in the magnetotail. It is commonly noted that southward turning of IMF triggered the dissipation of magnetotail energy which is injected into the auroral region and initiated the substorm growth phases. In dissipation of the magnetotail energy, geomagnetic field has changed to more dipolar configuration, and the dipolarization front is formed first in the premidnight-midnight sector, and then moves towards the Earth and spreads both eastward and westward (Runov et al., 2009;Ohtani et al., 1992). At the same time, substorm dipolarization causes pitch-angle scattering of plasma sheet electrons and the resulting precipitation excites auroral emissions. On the other hand, enhanced convection during substorms or dipolarization fronts produces plasmaspheric erosions. Therefore, we suggest that the substorm dipolarization is closely associated with plasmaspheric erosions, the propagation of EFs, and the corresponding electron pitch-angle scattering causing the auroral signatures during substorms.
Preliminary investigation in this work indicates that the dynamics of the plasmasphere and ionosphere are strongly dependent on internal processes of the magnetosphere when solar wind-driven convections are weak and steady. More works including simulations need to be done in future to further understand the mechanism of generating the nightside pointing plumes and the inner magnetospheric dynamics during substorms. For example, wave induced scattering and field line curvature scattering might also drive the DMSP observed auroral activity. More cases with global auroral images and global plasmaspheric images should be collected to systematically study the relationship between the auroral activity and the plasmaspheric dynamics during substorms in future.