Multi-source Perturbations in the Evolution of a Low-latitudinal Equatorial Plasma Bubble Event Occurred over China

In this paper, multi-ground-based instruments, including an all-sky airglow imager (ASAI), a very high frequency (VHF) radar, and eight digisondes, were combined to investigate multi-source perturbations in the evolution of an EPB event that occurred over low latitudes in China. We found this EPB event initially evolved from the bottom-type structures, most likely seeded by the atmospheric gravity wave (AGW) and the collisional shear-type instability (CSI)-inducing perturbations. Once formed, those bottom-type structures further evolved into bifurcated/plume-like structures at the ionospheric topside by the generalized Rayleigh-Taylor instability (RTI). Observed and analyzed are two diﬀerent perturbation mechanisms of RTIs: one is the prereversal enhancement of the zonal electric ﬁeld (PRE) inducing-RTI; another is the equatorward wind-inducing RTI around midnight. Accompanied by the PRE-inducing RTI are bifurcated/plume-like structures with a larger poleward (upward) velocity. The PRE could directly elevate the bottom-type structures to the ionospheric topside where the bifurcated/plume-like structures were generated by the RTI process. The near-midnight RTI was trigged by a vertical upward plasma jet caused by a seasonal equatorward wind in a region far away 10 ° N (20 ° N) from the geomagnetic (geographic) equator. This equatorward wind-inducing RTI persistently forced topside structures of those developed depletions to form secondary bifurcated/plume-like structures near midnight. Poleward developments of two cluster-type depletions of the EPB event were modulated by a large-scale wave-like structure (LSWS) occurring on the bottomside of the ionosphere. An eastward/westward polarization electric ﬁeld inside the upwelling/trough region of the LSWS could accelerate/suppress the development of cluster-type depletions.


Introduction
There are equatorial plasma bubbles (EPBs), a nighttime phenomenon that frequently occurs in the low-latitudinal ionosphere.They manifest themselves as the field-aligned depleted regions of airglow intensity in optical observations (e.g., Kelley et al., 2002), bite-out structures in satellite observations (e.g., Weber et al., 1982), and plume-like structures in radar observations (e.g., Tsunoda et al., 1982).Since these EPBs can cause severe outages in satellite-based communication and navigation systems, understanding the day-to-day variability of this phenomenon is one of the important topics of space weather interest.
Observations (e.g., Huang, Burke et al. 2001;Burke et al. 2004) indicate that EPBs frequently occur at post-sunset when the ionosphere is significantly uplifted by a prereversal enhancement of the zonal electric field (PRE; Fejer et al., 1991;1999).
The sunset PRE can trigger the generalized Rayleigh-Taylor instability (RTI), which is the dominant driving mechanism of EPBs (Kelley, 1989).However, this mechanism cannot explain the wave-like depletions with a typical wavelength of 400-1000 km, the cluster-type depletions with a smaller spacing (~100 km), and the freshly-generated depletions of EPBs near midnight.Thus, rather than the PRE, other seeding perturbations on the ionosphere bottomside can also initialize EPBs by the RTI process.
One of the most invoked seeding perturbations for EPBs is the atmospheric gravity wave (AGW), which well explains the periodic characteristic of the successive EPB depletions (e.g., Makela et al., 2010;Takahashi et al., 2009;2010).Freely propagating secondary AGWs with wavelengths higher than 150 km in the thermosphere (Vadas, 2007) can seed EPBs.Pieces of observation evidence (e.g., Tsunoda et al., 1982;Makela & Miller, 2008;Thampi et al., 2009;Narayanan et al., 2012) indicates that a large-scale wave-like structure (LSWS) can also initialize EPBs from its upwelling regions.An LSWS can appear well before E region sunset (Tsunoda et al., 2010;Thampi et al., 2009), playing a more dominant role in the development of EPBs than the post-sunset rise (PSSR) of the F-layer, and causing the day-to-day variation of EPBs (Tsunoda, 2005).However, observations (e.g., Makela & Miller, 2008;Narayanan et al., 2012) indicated that the LSWS is not sufficient to explain those cluster-type depletions (typical scale ~100 km) inside the upwelling regions of an LSWS.It is promising that either the gradient drifting-type instability (Kelley, 1989) or the collisional shear-type instability (Hysell & Kudeki., 2004) occurring on the ionospheric bottomside explains those cluster-type depletions.However, evidence for perturbations caused by such ionospheric instability is lacking.
Few studies (e.g., Yokoyama et al., 2011;Ajith et al., 2016;Dao et al., 2016Dao et al., , 2017;;Sun et al., 2021a) found an equatorward neutral wind (ENW) can also initialize EPBs, especially those near midnight.Those studies suggested that the uplift of the ionosphere by an ENW could cause a decreasing ion-neutral collision frequency, resulting in an increasing gravity-driven eastward electric current that can initialize EPBs above the geomagnetic equator (Nishioka et al., 2012;Huba & Krall, 2013).
However, this mechanism is inconsistent with the early knowledge that an ENW can enhance the field-line integrated Pedersen conductivity and then depress the appearance/development of EPBs (Maruyama, 1988;Krall et al., 2009).Moreover, the activated EPBs near midnight could be secondary structures evolving from the topside structures of those drift-type/developed EPBs under the local ionospheric and thermospheric conditions.It is thus particularly obscure for the role of an ENW in initializing EPBs.More studies are required to investigate the possible role of an ENW in developing EPBs near midnight.
In this paper, we investigate multi-source perturbations in the evolution of an EPB event observed by an all-sky airglow imager (ASAI) deployed in low latitude, China on a geomagnetically quiet night (Kp < 3; Sum (Kp) = 15 -).We found this event initially evolved from the bottom-type structures.The possible roles the AGW-and CSI played in the formation of these bottom-type structures were discussed.We also separately explained how the PRE and equatorward wind-inducing RTIs drove those bottom-type structures to further develop into bifurcated/plume-like structures that occurred at post-sunset and near midnight.In section 2, we briefly describe the instruments and the data.Section 3 presents the results and analyses of the event, followed by section 5 as summaries and conclusions.

Instruments and Data
In this study, data from multiple ground-based instruments are used, including one Among these instruments, the Fuk and MH digisondes, and the Fuk VHF radar belong to the Chinese Meridian Project (Wang C, 2010).Here note that the JJ, IC, ShY, BJ, LM, and MH stations are at mid-latitudes, and the Fuk and Dax stations are at low-latitude; the LM is close to the magnetic conjugation point of the BJ station.The color pentagrams in Figure 1 , where I and I are the intensity of an unwarped airglow image and a 1.0-h running mean of successive unwarped airglow images, respectively.Figure 2 presents the temporal evolution of an EPB event observed by the Dax ASAI on the night of 17 November 2015 (17-Nov-2015).
The Fuk VHF radar operates at a 47 MHz frequency with 54 kW peak power and a 2-MHz bandwidth to observe the 3.2 m scale FAIs by allocating the radar beams in directions perpendicular to the geomagnetic field lines.The height resolution of the h'f (hfp) for all stations is about 5 km.These ionospheric parameters [h'f (hfp) and foF2] were used to investigate the influences of the thermospheric meridional wind on the evolution of the EPB event.The simulated thermospheric wind from the horizontal wind mode (HWM-14; Drob et al., 2015) was also combined to explain the variations of the ionospheric parameters.

Evolution of the EPB Event
Presented in Figure 2 are the processed airglow images on the EPB night.Figure 2a shows time sequences (in near 25 min intervals) of some unwarped images during Here we first consider the temporal evolution of the EPB event.We suggest the reader skip Figure 2a and see Figure 2b since the sharper structures in Figure 2b make the reader well understandable the following description.
At 11:59:57 UT (the first image), the airglow depletion structure (d1) aligned with the geographic North-South (N-S) direction appeared in the southern FOV of the image.
Its tip was passing through the head of the Fuk station (see the filled black triangle).
Later, it proceeded to move poleward as it continuously drifted eastward, and gradually tilted westward.At 13:13:05 UT, it developed into a bifurcated structure when reaching a maximum poleward geographic latitude of about 27.0°N.The mean poleward velocity of d1 was about 295 m/s (1062 km/hr).Later, the d1 became blurry, and its phase elongation again became aligned with the geographic N-S direction.
After 15:39:32 UT, the d1 disappeared in the FOV of the images.
On the left of d1 is cluster-type depletions marked with the CDS1.Its scale is about 150 km.At 13:37:28 UT, the CDS1 appeared almost due south of the image.Like the d1, the CDS1 also continuously grew poleward as they drifted eastward.From 13:37:28 UT to 14:01:50 UT, the tip of the CDS1 hiked poleward about 1.2° (~140 km).The third depletion inside CDS1 (CDS1-3) extended to a maximum geographic latitude of about 21.5°N.The mean poleward velocity of CDS1 reached about 97 m/s (350 km/hr).Interestingly, the CDS1 did not continue to grow poleward.Later 14:01:50 UT, it moved equatorward with a mean velocity of about 40 m/s (144 km/hr).
After 16:52:49 UT, the CDS1 also fades out of the FOV of the images.Narayanan et al. (2016) first described this regressive characteristic of EPBs in latitudes as a shrinking phase of EPBs.
On the left of CDS1 is another cluster of depletions marked with CDS2.They had a slower growth that CDS1 when drifting eastward.There are three depletions marked with d2-d4 subsequently appearing in the western regions of CDS2.Their wavefronts present the form of a plane wave.Although not so drastic in evolution as depletions d1 and CDS1-CDS2 described above, these depletions were still evolving and became bifurcation structures later.
There are other ionospheric phenomena appearing in Figure 2a.Firstly, a horizontal band-like region that has an extensional latitudinal width of near 5° and a broad longitudinal region appeared in the airglow images.As time proceeded, this region moved equatorward.Based on previous studies (e.g., Narayanan et al., 2013Narayanan et al., , 2014)), this region is the EIA crest, which usually propagates equatorward at nighttime because of the so-called "antifountain effect."Secondly, two other patchy regions (marked with "PR1-PR2") with brighter airglow intensity appeared within the EIA region.Previous studies by Sun et al. (2017Sun et al. ( , 2021a) ) found this kind of brightness region is the true plasma phenomenon that has an enhanced plasma density than the surrounding ionosphere.These brightness structures were early called "blobs" by Pimenta et al. (2004Pimenta et al. ( , 2007)).appearing after 17 UT were classified into those so-called "fossilized bubbles" previously described by Chapagain (2015) and Sekar et al. (2007).
Worthy of that bottom-type layer structures (BLSs) were observed in Figure 3. From

Explanations of the EPB Evolution
Airglow and VHF radar observations described above present three kinds of depletions that occurred at sunset, near midnight, and post-midnight.Those depletions that occurred at sunset had a faster poleward/vertical velocity than those near-and post-midnight.To explain these differences, here we investigate the background ionospheric conditions during the evolution of all EPB depletions.Figures 5a2-5a3   et al. (1991, 1999), this post-sunset ionospheric uplift resulted from the so-called PRE.
Presented in Figure 6 is evidence more direct indicating the occurrence of such a PRE on this EPB night.Figure 6 shows the UT variations of the EIA crests as the geographic latitude along the Fuk longitude.One can see that between 12 and 13:50 UT the EIA crests moved poleward with a velocity of about 50 m/s.After 13:50 UT, the EIA reversed to equatorward with a velocity of about 75 m/s.The EIA's poleward/ equatorward movement was attributed to the inferred-PRE/antifountain-effect above.
Interestingly, during the passage of the CDS2-3 (near 16 UT) no elevation occurred on the ionospheric bottomside.However, the passage of the CDS2-3 caused a small uplift in hfp (about 25 km) between 16 and 17 UT.Depletions d2-d4 did not cause any perturbations in both h'f and hfp.Between 14 and 20 UT, the bottomside ionosphere around Fuk remained almost horizontal, and the topside ionosphere varied a few.This explains why the observed EPB depletions survived through the night.with the almost same velocity as the background eastward wind (Haerendel et al., 1992).Because of the curl-free of downward electric field (Eccles, 1998;Eccles et al., 2015), an eastward-enhanced polarization electric field (PEF) would be generated on the daytime side of the solar terminator.This eastward-enhanced PEF is the PRE that is the most basic source for inducing the post-sunset EPBs, especially at equinoxes (Tsunoda, 1985).
What is the possible role the PRE played in the current case?As analyzed above, the uplift of the bottomside (peaked) ionosphere due to the PRE was about 30 ( 50 EIA before.This difference could be attributed to the height difference in digisonde and GPS-TEC observations.The digisonde data estimated a PRE at the height between 250 and 300 km (from the bottomside to the peaked height of the ionosphere), while the GPS-TEC data estimated a PRE at the height of 450 km (topside ionosphere).A possible explanation is that the ionosphere below the peaked height located nearby/within the westward shear flow region of the bottomside ionosphere associated with the PRE (Kudeki & Bhattacharyya, 1998).The topside ionosphere is more sensitive than the bottomside ionosphere for the PRE on this EPB night.This is very reasonable because the plume-like echo structures of d1 and CDS2-3 were indeed connected with a bottom-type layer structure (BLS) on this event.Can such a low velocity cause the observed EPBs?From Figure 7  structures is very similar to those of plume-like structures that evolved from bottom-type layer structures in Figure 1 by Takahashi et al. (2010).Also, Hysell & Burcham, (1998) and Hysell (2000) found that different types of irregularities observed at Jicamarca usually occur sequentially, preceded by the occurrence of a bottom-type irregularity layer.The bottom-type irregularity layer was generally thought to be a precursor of a fully developed equatorial plasma plume (Hysell & Burcham, 2002;Li et al., 2017).Therefore, here we suggest that those AGW-and CSI-seeding perturbations described above could cause the bottom-type structures in the form of the wave-like and cluster-type depletions, which further evolved into the bifurcated/plume-like structures at the ionospheric topside.
So far, we have considered those EPB depletions in Figure 2 evolved from the bottom-type structures.However, one question remains: why did d1 depletion that first entered the FOV of the airglow images had a faster poleward/upward velocity than d2-d4/CDS2-3?Because the rapid poleward growth of the d1 occurred in a post-sunset elevated ionospheric region, bottomside structures of the d1 depletion could have been directly uplifted by the PRE and rapidly developed into bifurcated/plume-like structures at the ionospheric topside by the PRE-driving RTI process.However, this PRE-driving RTI mechanism is not sufficient to explain the d2-d4/CDS2-3 around midnight.Returning to Figure 2, one can see that when the d2-d4/CDS2-3 were/was evolving when the d1 had stopped growth.Meanwhile, from Figures 5a-5b and Figure 6, one can see that poleward growth of the d2-d4/CDS2-3 initialized when the antifountain effect began (~14 UT).This means that the PRE had almost uncoupled with the background ionosphere later 14 UT when the d2-d4/CDS2-3 appeared in the FOV of airglow images; the d2-d4/CDS2-3 were/was developing when the PRE had reversed to a westward electric field (WEF).Since a WEF would confine EPBs to the ionospheric bottomside and depress the EPB development (Seker et al. 2007), it is thus impossible for an electric field-driving RTI process to initialize the d2-d4/CDS2-3 later 14 UT.If no other physical processes, the d2-d4/CDS2-3 later 14 UT should also stop growth as d1 did.An additional perturbation source is required to initialize the poleward growth of d2-d4/CDS2-3 again.This driving source is not as effective as the PRE-driving source.
Where did such a perturbation come from?Firstly, it is impossible for such a perturbation generated at a latitude lower than the Fuk station.If yes, when it propagated above the Fuk, the bottomside ionosphere around Fuk must be first elevated.However, the bottomside h'f around Fuk remained near 225 km between 14 and 16 UT while the CDS2 was persistently growing poleward.Therefore, it must be a perturbation that came from the latitudes higher than the Fuk station and persistently disturb the ionospheric topside.Here we can think of two possible perturbation sources: one is the polarization electric field generated by other ionospheric phenomena (e.g., Es) whose footprint connects with the topside region of Fuk station by the magnetic field line at the more northern latitude of Fuk; another is a nighttime neutral wind which blows equatorward from a higher latitude than the Fuk.We checked and excluded the strong Es activities (foEs> 5 MHz) that occurred in the ionograms from the more northern digisondes (e.g., ShY, IC, JJ, and BJ), since only very weak Es (foEs< 3 MHz) activities occasionally appeared on this night.However, an equatorward wind prevails at night, and can persistently elevate the ionosphere for a long time.It is possible that an equatorward wind as a sustaining source of instability to excite the generalized RTI at the topside ionosphere exceeding over several hours.Unfortunately, we have no usable neutral wind data on the EPB night.
However, alternatively, we can use the height variations of the ionosphere at mid-latitudes to reflect the influence of neutral wind from the higher latitudes.This is very reasonable because the F-region dynamo generated at midlatitude is too much higher than those generated by the E-region dynamo (Richmond et al. 1980).
Following is the simple deduction of such a wind.
To verify this possibility, Figures 7a1-7a6 and 7a8 present the ionospheric variations of h'f obtained from a digisonde chain that spanned from the SaY to the MH stations.
The result indicates a significant uplift of the ionospheric h'f occurred during 12-16 UT at IC, JJ, BJ, and MH stations.The uplifts that occurred at mid-latitudes can be directly attributed to a nighttime equatorward wind.Two uplifts occurred at ShY station.The first uplift that occurred between 12 and 13 UT resulted from the same PRE observed by the Fuk/SaY digisonde.The uplift that occurred between 14 and 18 UT was attributed to the same equatorward wind observed at IC, JJ, BJ, and MH stations.However, at the same time, the equatorward wind did not cause any ionospheric perturbations at Fuk and SaY stations.Such an equatorward wind appeared in the mean wind (black-solid line) at postmidnight but disappeared on the EPB night.As will be analyzed later, the disappearance of such an equatorward wind at postmidnight at Fuk/SaY station was most likely offset by a poleward wind associated with a passing-by nighttime temperature maximum (TM; Colerico & Mendillo, 2002) from the lower latitudes.Sun et al. (2021a) previously verified such a poleward wind by analyzing relative variations between the ionospheric heights observed by a chain of digisondes and the meridional winds measured by a Fabry-Perot interferometer in the Chinese sectors.Note that a significant uplift between 13 and 17 UT also occurred at LM station, which is located nearby the magnetic conjugation point of the BJ station.However, the uplift at BJ is higher than at LM.This suggests that the equatorward wind in the northern hemisphere was higher than in the opposite hemisphere during this time.Amplitudes of the equatorward-enhanced wind gradually decreased as it propagated to lower latitudes.A possible reason is that the equatorward enhanced wind would become damped when propagating to the EIA region where the significant increase of the ion drag effect (Shiokawa et al. 2002)  Based on the observations and the model results above, we found a vertical upward perturbation source of the plasma jet occurring in the northern region of the Fuk station.Such a perturbation source elevated the ionosphere and then resulted in those plume-like structures of d2-d4/CDS2-3 around midnight.
Previous studies (e.g., Nishioka et al., 2012;Huba & Krall, 2013) suggested that the uplift of the ionosphere by an equatorward-enhanced wind could cause a decreasing ion-neutral collision frequency, resulting in an increasing gravity-driven eastward electric current that can initialize EPBs above the geomagnetic equator.This mechanism explains that those bifurcated/plume-like structures near midnight were generated from the bottomside of the ionosphere by the gravity-driven RTI process occurring above the magnetic equator.However, as analyzed in section 3.1, those plume-like structures of d2-d4/CDS2-3 only evolved from the topside structures of the developed depletions that extended over poleward 24°N later 14 UT.No elevation in h'f occurred around the Fuk when the d2-d4/CDS2-3 were/was growing poleward of the Fuk latitude.Those plume-like structures of d2-d4/CDS2-3 did not need to develop from the geomagnetic equator as the d1 did.Those plume-like structures of d2-d4/CDS2-3 found here were only the secondary structures that evolved from tips of the developed depletions at a latitude away from 10° (20°) the geomagnetic (geographic) equator by the equatorward wind-inducing generalized RTI process.
Since the equatorward wind at low-mid latitudes usually reaches a maximum near midnight in our observations, it is expected that the maximum occurrence of the bifurcated/plume-like structures also appears near midnight.The equatorward wind-inducing RTI process described here should be the universal mechanism for triggering those activated EPB depletions/irregularities that occurred around midnight.
Since those secondary structures occurring around midnight did not directly evolve from the ionospheric bottomside, the equatorward wind-inducing RTI near midnight is not as effective as the PRE-driving RTI at sunset.This can reasonably explain why the d2-d4/CDS2-3 occurring near-midnight/at-postmidnight had slower growth than the d1 that occurred at post-sunset.suggest that the bottomside ionosphere provided a relatively stable condition occurring to prevent the erosion of the plasma because of the photochemical processes on this night.More plasma was accumulated on the bottomside of the ionosphere and caused the longtime plasma density enhancement.

Summaries and Conclusions
In this paper, we present the multi-source perturbations in the evolution of an equatorial plasma bubble (EPB) event observed by an all-sky airglow imager (ASAI) Observations indicated: 1.The primary structures of the EPB depletions were successively ordered with a spacing of about 400 km along the longitude direction; two cluster-type structures around the primary EPB depletions had a smaller spacing of ~150 km; 2. Three kinds of depletions that occurred near sunset, near midnight, and post-midnight were observed; those depletions occurring at sunset had a faster poleward/vertical velocity than those that occurred around midnight.Depletions that occurred at postmidnight became fossilized bubbles; 3. Accompanied by the EPB depletions are bottom-type layer structures that were connected to the roots of the plume-like irregularities observed by the VHF radar; 4. The poleward growth of those post-sunset depletions occurred when the ionosphere was elevated by a prereversal enhancement of the zonal electric field (PRE); the relatively slower poleward growth of those near-midnight depletions occurred when an equatorward seasonal wind elevated the ionosphere in regions with a latitude higher than the geographic (geomagnetic) equator ~20° (10°); Analyses indicated: 1.The EPB event evolved from the bottom-type structures that could be seeded by This paper thus highlights the significance of multi-source perturbations in the evolution of an EPB event that occurred over the low latitudes of China.
present the locations of these digisonde instruments; the red-filled dot represents the location of Dax ASAI, and the black-filled circle represents the nearly 160° field of view (FOV) of the Dax ASAI; the black-dotted line represents the magnetic equator (MQ).

Figure 1 .
Figure 1.(a) Geographic locations of the ground-based instruments.(b)-(c) indicate the altitude-zonal and latitude-longitude distributions of the seven beams of the VHF radar.(d) hapex-latitude distributions of magnetic field lines at seven latitudes.
Besides the ground-based data above, the global vertical total electron content (VTEC) map data provided by the Navigation Headquarter of the Chinese Academy of Sciences (CAS-TEC map data) were also used to investigate the evolution of the EPB event.The time resolution for each map is 15 min.The height of the TEC is assumed at 450 km.Based on the database, Sun et al. (2020) investigated a WSA-like plasma patch, which affected the propagation/evolution of an EMSTID event over midlatitude, China; Sun et al. (2021b) further investigated the possible influence of EIAs on the propagation/evolution of an airglow event occurred over Dax, China.
From 14:01:50 UT to 15:15:03 UT, the third depletion inside CDS2 (CDS2-3) hiked poleward from about 24.7°N to a maximum geographic latitude of 27.5°N, with a mean poleward velocity of 73 m/s (262 km/hr).During the poleward growth of CDS2-3, a bifurcated structure evolved from the left wall of CDS2-3.After 15:15:03 UT, the CDS2-3 remained at its maximum poleward latitude (27.5°N) until it finally drifted out of the FOV of the image.

Further
presented in Figures 3-4 are the VHF radar observations to investigate the vertical evolution of those EPB depletions above.Figures 3a-3b show the UT variations of 3.2-m scale irregularities separately represented by the SNR and the Doppler shift velocities as the altitude.Figure 4 further gives the altitude-zonal distributions of the corresponding radar echoes.Radar echo structures resulting from the d1, CDS2-3, and d2 airglow structures were identified.Following are descriptions of the vertical evolution of these echo structures.

Figures
Figures 3a2-3a3, there are indications that these BLSs were connected with the roots of the plume-like echo structures d1 and CDS2.The accompanied Doppler drift velocities in Figures 3b2-3b3 were near 0 m/s.
give the temporary variations of the ionospheric height (h'f, hfp, and hEs), and foF2 (foEs) from the Fuk digisonde on the EPB night.A time sequence of N-S cross-sections (keogram) of the successive airglow images in the Fuk longitude (109.1°E) was also shown in Figure 5a1.For a good comparison, the observed results on the night of 16/18-Nov-2015 were also presented in Figures 5b1-5b3 and 5c1-5c3.

Figure 6 .
Figure 6.Geographic latitude-UT variations of the EIA crests reconstructed from the CAS-TEC map data on the night of 17-Nov-2015.
) km within 2 hrs.If ignoring the photochemistry, the vertical upward mean velocity of the background ionosphere was about 5.5 m/s (19.8 km/hr).According to the IGRF-13 model (https://ccmc.gsfc.nasa.gov/modelweb/models/igrf_vitmo.php), the northward and downward components of the geomagnetic field (B) at 300 km above the Fuk station are ~0.034 and ~0.017 mT, respectively.If the estimated vertical upward velocity above was caused by the PRE, the background plasma would drift poleward at a velocity of about 11 m/s [5.5×(0.036/0.017)m/s].This estimated poleward velocity for plasma is nearly a fifth of the inferred poleward velocity (~ 50 m/s) of the Figure2, one can see that the d1, the primary structures of CDS1-CDS2, and the d2-d4 are successively ordered with a spacing of about 400 km along the longitude direction.AGWs with a wavelength of 400 km likely seeded the successive EPB depletions in the current event.The AGW seeding perturbations would result in bottom-type structures whose horizontal wavelength has the same spacing as the successive EPB depletions.Vadas (2007) also pointed out that AGWs with a horizontal scale of less than 150 km would break and dissipate at the thermospheric height.Those cluster-like depletions inside CDS1/CDS2 on a smaller scale (~150 km) should evolve from other seeding perturbations.Hysell & Kudeki (2004) previously found that a collisional shear branch of the Kelvin Helmholtz instability (KHI) on the bottomside ionosphere can precondition the F region for initializing the bottom-type perturbations with an initial

Figure 8 .
Figure 8.The HWM-14 simulated neutral wind at 250 km.The first column gives the meridional winds at 10, 12, 14, and 16 UT.The second column gives the corresponding zonal winds.The third column gives the calculated vertical plasma drift from the simulated meridional and zonal winds.The filled black triangle represents the location of Fuk.
would slow down the passing-by neutral wind.Those dotted lines presented in Figures 7c1-7c8 further give the corresponding vertical drifts of plasma (positive upward) calculated by the HWM-14 simulated neutral wind.In the calculation, we included both the meridional and zonal winds and considered two factors of inclination and declination of the geomagnetic field lines at the height of 250 km.The same processes were conducted in calculating the latitude-longitude distributions of the vertical plasma drift as presented in the third column of Figure 8.The result indicates that a plasma jet with a vertical upward velocity of 10-30 m/s was passing nearby the Fuk station after 14 UT.Such an upward jet of plasma would result in the uplift of the bottomside ionosphere observed by those digisondes at mid-latitudes.However, as presented in Figures 7c1-7c2 (black dotted lines), the vertical drifts caused by the equatorward-enhanced neutral winds at Fuk and SaY stations are near 0 m/s.This well explains why the observed bottomside ionosphere remained almost horizontal between 14 and 16 UT at Fuk and SaY stations.
the AGW and CSI -inducing perturbations; the AGW with the same horizontal wavelength as the successive EPB depletions could explain the periodic characteristic of the EPB event; CSI occurring on the ionospheric bottomside could cause those cluster-type depletions on a smaller scale (~100-150 km); 2. The generalized RTI processes further drove those AGW and CSI -inducing bottom-type structures to form bifurcated/plume-like structures at the ionospheric topside: those plume-like depletions with faster growth at post-sunset evolved from the bottom-type structure most likely driven by the PRE-inducing RTI process; those near-midnight plume-like depletions with a slower poleward/upward velocity were the secondary structures evolved from topside structures of the developed depletions by a seasonal equatorward wind-inducing RTI process; 3.Those two cluster-type structures first appeared in the same upwelling region of a large-scale wave-like structure (LSWS), but later the right adjacent cluster-type structures were modulated into a rough/decreasing region of the LSWS.An eastward/westward polarization electric field inside the upwelling/trough region of the LSWS could accelerate/suppress the development of the left/right cluster-type depletions.