Spatial Development of Strong Storm-Induced Ionospheric Perturbations during 25–27 August 2018

: The 25–27 August 2018 geomagnetic storm was the third largest storm in the 24th solar cycle. It was a surprising space event that originated from low-level solar activity. This study provides an overview of the temporal–spatial behaviors of plasma irregularities as functions of geographic longitude, latitude, and altitude using ground-based (GNSS receivers and ionosonde) instruments and space-borne Swarm satellites. The results not only reveal enhanced equatorial ionization anomaly (EIA) and hemispheric asymmetry over the Asian–Australian and American sectors at a particular time but also hemispheric asymmetric features of global ROT in the main and recovery phases. Additionally, this storm triggered positive plasma irregularities in altitudes of 100 to 150 km near the Auroral zone, and the changed ratio of bottom-side plasma irregularities exceeded 250%. This ﬁnding has been cross-validated by other instruments and models. Furthermore, the storm signiﬁcantly affected the thermospheric O/N 2 density ratio, equatorial electrojet, and vertical E × B drifts. The equatorial and mid-latitude plasma irregularities may be a combined action of thermospheric composition change, equatorial electrojet, and vertical E × B drifts. Finally, the storm induced positive Joule heating irregularities in the Auroral ionosphere in altitudes of 100 to 400 km with a maximum changed ratio of over 200%, as well as enhanced cross-Polar voltage to ~90 kv. The Polar ionospheric irregularities may be associated with additional energy input through particle precipitation, Joule heating, and ionospheric current intensiﬁcation.


Introduction
The ionosphere has a serious effect on absorbing, scattering, and refracting radio signals, which is a main source of error in navigation and positioning services.Now, the global navigation satellite system (GNSS) differential technique and empirical/theoretical models are usually used to correct the ionospheric delay [1][2][3].However, ionospheric plasma during severe geomagnetic storms can suddenly increase or decrease violently, which easily reduces the accuracy of positioning and navigation.Currently available techniques are not good at correcting severe storm-effect ionospheric perturbations [4].Therefore, it is necessary and valuable to investigate the ionospheric spatial-temporal behaviors under strong magnetic storms and to discuss the probable drivers.
Geomagnetic storms usually have a severe effect on the ionospheric system, and auroral particle precipitation, polar ionospheric currents, and convection are largely reinforced during a geomagnetic storm.The enhanced polar ionospheric ionization and electric fields penetrate to low-middle latitudes, and this process affects global electrodynamics and changes the structure of the thermospheric-ionospheric system remarkably.The Joule heating and auroral particle precipitation heat and expand the thermosphere, which further changes the composition and dynamics of the thermospheric-ionospheric system [5].With the rapid development of the global navigation satellite system (GNSS) and radio occultation, the storm-effect ionospheric behaviors have been paid more attention to using multiple ground-based and space-borne techniques.The "Halloween" storm that erupted on 29-30 October 2003 was one of strongest geomagnetic storms of this century.After a few hours, when the interplanetary magnetic field suddenly turned southward, the dayside ionospheric total electron content (TEC) increased approximately 40%, and the Challenging Minisatellite Payload (CHAMP) profiles indicated that the dayside TECs over mid-latitudes increased ~900% on 30 October [6].Significant increments in the TEC and peak density (NmF2) were also observed over the European and North African sectors during a following stronger geomagnetic storm that occurred on 30 November 2003 [7].Moreover, sudden, irregular ionospheric behavioral responses to the geomagnetic storms were also reported over Jicamarca [8]; Brazilian equatorial-low latitudes [9]; China and adjacent areas [10]; the Asian-Australian sector [11]; the Indian sector [12]; and the Arctic and Antarctic [13,14], as well as on a global scale [15].These reports reveal that strong geomagnetic storms easily triggered large-scale positive or negative traveling ionospheric disturbances (TIDs), and sometimes the TIDs had a significant latitudinal asymmetrical structure in the Northern-Southern Hemispheres that was caused by the displaced magnetic poles and seasonal asymmetries in the thermosphere-ionosphere system.Multi-instrument observations and theoretical model simulations were conducted to explain the disturbed ionospheric dynamic convections, and the results conclude that the negative TIDs were primarily attributed to a decrement in the thermospheric density ratio of O/N 2 [16,17].However, the drivers for positive TIDs were various: thermospheric neutral winds, disturbance dynamo electric fields (DDEFs), prompt penetration electric fields (PPEFs), and quasi-trapped energetic electrons from the inner radiation belt, as well as charged particle precipitation, which have been reported as potential factors that enhance plasma densities [16,[18][19][20].
Ionospheric storms primarily occur as a consequence of strong coronal mass ejections (CMEs), such as the magnetic storms on 22-23 June 2015 [21] and 7-8 September 2017 [22].However, the 25-27 August 2018 space event, which was the third largest magnetic storm in the 24th solar cycle, happened after a slowly moving CME on 20 August, and it was a huge surprise that the weak CME, which did not even show a sudden impulse, could trigger a strong ionosphere-thermosphere response.In terms of the positive and negative ionospheric perturbations over North America [23], Brazil [24], and middle latitudes [25], as well as globally [26], responses to this surprising space event were reported.This geomagnetic-storm-induced penetration of electric fields created favorable conditions for strong fountain effects that enhanced the equatorial ionization anomaly (EIA) and generated equatorial plasma bubbles (EPBs).The EPBs appeared over a larger latitudinal extent of EIA crests, while the plasma density along the western coast of North America depleted in the northwestward direction [23].The Defense Meteorological Satellite Program (DMSP) also detected mid-latitude plasma depletion in the Asian sector, and the local TIDs were responsible for the mid-latitude plasma depletion in Asia and the United States, rather than the absence of EPBs [25].Different from the mid-latitude plasma depletion in the Northern Hemisphere, Spogli et al. [24] used the situ measurements provided by the China Seismo-Electromagnetic Satellite and by the Swarm-A satellite with ground-based observations to reveal the ionospheric response at low-middle latitudes over Brazil, and they found that significant foF2 increments appeared over the ionosondes located at both the dip-equator and southern crest of the EIA.The decrease in the eastward electric field was the main driver for the equator station, while it resulted from the storm-induced equatorward thermospheric winds for the crest station.In addition, the unprecedented hemispheric asymmetries of the thermospheric-ionospheric responses were also observed during the main and recovery phases of the storm, which expressed that strong positive plasma storms occurred in the Northern Hemisphere at the beginning of the space event, while an extreme expansion of the thermospheric composition ratio of O/N 2 appeared in the opposite hemisphere during the recovery phase.The seasonal asymmetry in the high-latitude plasma and neutral mass density distributions along with the asymmetries in the geomagnetic field played a decisive role in the hemispheric asymmetric structure of disturbed plasma [26].
Most previous studies have focused on the planar ionospheric response (TEC) to the storm, but the altitudinal plasma behavior is still not clear.In addition, the stormeffect ionospheric response is controlled by multiple drivers at a particular moment in time and in a particular location.The present study's objectives were to examine the following: (1) development of hemispheric asymmetry of plasma irregularities in the main and recovery phases; (2) altitudinal behaviors of plasma irregularities in the lowmiddle latitudes and Auroral zone; and (3) potential drivers of the latitudinal plasma irregularities over the Asian-Australian and American sectors.To address these objectives, latitudinal ionospheric irregularities were detected using a set of ground-based (GNSS receiver and digital ionosonde) and space-borne (SWARM) instruments, and the potential drivers for the equatorial-auroral irregularities are discussed after examination using a set of magnetometers, SuperDARN, a global ultraviolet imager (GUVI), and thermosphereionosphere electrodynamics general circulation (TIE-GCM).

Materials
In order to fully analyze the planar-vertical behaviors of global plasma irregularities during the magnetic storm that occurred on 25-27 August 2018, this study utilized groundbased observations from GNSS receivers and ionosondes, as well as plasma profiles derived from the space-borne Swarm constellation.Additionally, the horizontal components of the magnetic field, the thermospheric O/N 2 density ratio, and simulations from the TIE-GCM model were also used to explain the drivers of the plasma irregularities.The geographical locations of the ground-based instruments used in the study are shown in Figure 1, and more detailed information concerning the various datasets used is presented below.
expressed that strong positive plasma storms occurred in the Northern Hemisphere at th beginning of the space event, while an extreme expansion of the thermospheri composition ratio of O/N2 appeared in the opposite hemisphere during the recover phase.The seasonal asymmetry in the high-latitude plasma and neutral mass densit distributions along with the asymmetries in the geomagnetic field played a decisive rol in the hemispheric asymmetric structure of disturbed plasma [26].
Most previous studies have focused on the planar ionospheric response (TEC) to th storm, but the altitudinal plasma behavior is still not clear.In addition, the storm-effec ionospheric response is controlled by multiple drivers at a particular moment in time and in a particular location.The present study's objectives were to examine the following: (1 development of hemispheric asymmetry of plasma irregularities in the main and recover phases; (2) altitudinal behaviors of plasma irregularities in the low-middle latitudes and Auroral zone; and (3) potential drivers of the latitudinal plasma irregularities over th Asian-Australian and American sectors.To address these objectives, latitudina ionospheric irregularities were detected using a set of ground-based (GNSS receiver and digital ionosonde) and space-borne (SWARM) instruments, and the potential drivers fo the equatorial-auroral irregularities are discussed after examination using a set o magnetometers, SuperDARN, a global ultraviolet imager (GUVI), and thermosphere ionosphere electrodynamics general circulation (TIE-GCM).

Materials
In order to fully analyze the planar-vertical behaviors of global plasma irregularitie during the magnetic storm that occurred on 25-27 August 2018, this study utilized ground-based observations from GNSS receivers and ionosondes, as well as plasm profiles derived from the space-borne Swarm constellation.Additionally, the horizonta components of the magnetic field, the thermospheric O/N2 density ratio, and simulation from the TIE-GCM model were also used to explain the drivers of the plasm irregularities.The geographical locations of the ground-based instruments used in th study are shown in Figure 1, and more detailed information concerning the variou datasets used is presented below.The GNSS observations were obtained from the University NAVSTAR Consortium (UNAVCO), which provides access to geodetic GPS/GNSS data used for geoscience research and education.The UNAVCO provides approximately 2500 receiver-independent exchange format (RINEX) files daily through the following link: https://www.unavco.org/data/gps-gnss/gps-gnss.html, accessed on 20 May 2021.It should be noted that the vertical total electron content (VTEC) estimated by the carrier-phase smoothed pseudo-range method was used to investigate the ionospheric perturbations during the geomagnetic storm [27].For similarity, the TEC signifies the VTEC in the whole study.In addition, two chain digital ionosondes located in the Asian-Australian and American sectors were also utilized to investigate the vertical behaviors of the storm-induced plasma irregularities.Due to the influence of the geomagnetic storm, the digital ionosonde in the Asian sector failed to observe the auroral plasma irregularities.The ionosonde PQ052 (14.6 • E, 50 • N) located at Pruhonice in Europe, was selected.The sounder profiles of the two chains can be obtained from the Lowell DIDBase (https://giro.uml.edu/didbase/,accessed on 20 May 2021) via the SAO explorer.
Swarm consists of three microsatellites (Alpha, Bravo, and Charlie) that are placed in two different orbital planes, among them Swarm-A and Swarm-C fly at a mean altitude of 450 km, and the satellite Swarm-B is placed at a mean altitude of 530 km.In this study, the electron density profiles of Swarm-A and Swarm-B were selected to analyze the plasma irregularities, and the electron density was derived from the high gain ion current that is determined using the Langmuir probe.The Swarm profiles can be obtained from the following website: https://swarmdiss.eo.esa.int/#swarm%2FLevel2daily%2FEntire_mission_data%_2FTEC%2FTMS, accessed on 20 May 2021.In addition, the daily F10.7 index, 81-day mean F10.7, and Kp index were imported to the TIE-GCM model as input parameters, and the output parameters included the electron density, neutral winds, thermospheric composition, and electric field.For more details concerning the TIE-GCM, the readers can refer to https://www.hao.ucar.edu/modeling/tgcm/tie.php, accessed on 20 May 2021.To improve the simulated accuracy of the TIE-GCM products' response to the geomagnetic storms, the period for the TIE-GCM products was from 00:00 (Universal Time (UT)) 23 August to 29 August 2018.

Results
Solar geomagnetic change is an important factor that effects the process of ionospheric plasma irregularities during storms.The record, third largest geomagnetic storm of the 24th cycle was initiated from a slow CME on 20 August 2018, which arrived at the earth thermosphere-ionosphere system on 25 August (DOY 238).Therefore, it is necessary to evaluate the variation in the space conditions.

Solar-Terrestrial Environment
The solar wind speed (Vsw), interplanetary magnetic field, and geomagnetic indices obtained from the Goddard Space Flight Center (https://omniweb.gsfc.nasa.gov/form/dx1.html, accessed on 20 May 2021) was analyzed.As shown in Figure 2, the solarterrestrial indices were at a quiet level before 14 UT, DOY 237 (25 August), and a strong geomagnetic storm occurred from 14 UT, DOY 237 to DOY 239 (27 August). Figure 2a shows that the Vsw was low with a mean velocity of 400 to 450 km/s in the main phase (14 UT, DOY 237-07 UT, DOY 238) of the geomagnetic storm, while it abruptly enhanced during the recovery phase, with a maximum speed of 620 km/s.Figure 2b shows that the Bz component of the interplanetary magnetic field (IMF) had an abrupt southward excursion.From 14 UT, DOY 237, the Bz turned southward with a minimum value of −18 nT in the forenoon of DOY 238.The horizontal component of the longitudinally symmetric disturbances (SYM-H) was essentially the same as the Dst index in the description of the mid-latitude geomagnetic disturbances.Figure 2c,d express that both SYM-H and Dst had a significant negative excursion since the afternoon of DOY 237, and these indices reached minimum values of ~−200 nT and ~−180 nT at UT 06-08, DOY 238, respectively.At the same time, both the geomagnetic AE and Kp indices suddenly enhanced to a maximum at ~18 UT, 26 August, with values of ~1300 nT and 7, respectively.According to the classification of the geomagnetic storm released by the National Oceanic and Atmospheric Administration (NOAA), this storm was classed as "strong".

Ionospheric Irregularities over the Asian-Australian and American Sectors
The dual-frequency observations of approximately 2500 GNSS receivers provided by UNAVCO were used to estimate the global TEC map; then, the TEC grid maps were constructed using the kriging interpolation method.The north-south cross-sections (keograms) of the GNSS TEC maps along the 110°E and 70°W longitudes over 24-28 August were plotted to illustrate the temporal evolution of the storm-effect TEC changes over two sectors.These keograms were plotted as a function of the UT time and geographic latitude, as shown in Figure 3. From Figure 2d-f, the geomagnetic storm experienced three stages, including the initial phase, main phase, and recovery phase, which are signified by dashed black, red, and green lines, respectively.For example, the Dst value gradually had a slight disturbance when the geomagnetic storm occurred, which is defined as the initial phase.Several hours later, the geomagnetic storm dropped into the main phase when the Dst value suddenly turned negative.Then, the Dst reached a minimum value and gradually recovered to a normal condition; the moment is used to determine the recovery phase.Figure 3a demonstrates that in the Asian-Australian sector, the ionospheric TEC kept to a low level of 20 to 25 TECU.In the main phase of the storm, the equatorial ionospheric TEC enhanced significantly, with a double-peak structure that coincided with the Dst index, decreasing to a maximum value of −174 nT.The enhanced TEC primary occurred at the ending of the main phase and the beginning of the recovery phase (04-10 UT, 26 August), and the maximum TEC reached 40 TECU.

Ionospheric Irregularities over the Asian-Australian and American Sectors
The dual-frequency observations of approximately 2500 GNSS receivers provided by UNAVCO were used to estimate the global TEC map; then, the TEC grid maps were constructed using the kriging interpolation method.The north-south cross-sections (keograms) of the GNSS TEC maps along the 110 • E and 70 • W longitudes over 24-28 August were plotted to illustrate the temporal evolution of the storm-effect TEC changes over two sectors.These keograms were plotted as a function of the UT time and geographic latitude, as shown in Figure 3. From Figure 2d-f, the geomagnetic storm experienced three stages, including the initial phase, main phase, and recovery phase, which are signified by dashed black, red, and green lines, respectively.For example, the Dst value gradually had a slight disturbance when the geomagnetic storm occurred, which is defined as the initial phase.Several hours later, the geomagnetic storm dropped into the main phase when the Dst value suddenly turned negative.Then, the Dst reached a minimum value and gradually recovered to a normal condition; the moment is used to determine the recovery phase.Figure 3a demonstrates that in the Asian-Australian sector, the ionospheric TEC kept to a low level of 20 to 25 TECU.In the main phase of the storm, the equatorial ionospheric TEC enhanced significantly, with a double-peak structure that coincided with the Dst index, decreasing to a maximum value of −174 nT.The enhanced TEC primary occurred at the ending of the main phase and the beginning of the recovery phase (04-10 UT, 26 August), and the maximum TEC reached 40 TECU.
Different from the TEC change over the Asian-Australian sector, Figure 3b found a significant hemispheric asymmetrical structure of TEC irregularity in the American sector.For example, significant TEC depletion occurred over the American sector in the recovery phase.In the afternoon of 26 August, the TEC in the Northern Hemisphere depleted from 15-20 TECU to 10 TECU, while the equatorial and mid-latitude TEC in the opposite hemisphere enhanced approximately 5 TECU.The positive TEC storm was stronger in the afternoon of the following two days, and the maximum enhanced TEC exceeded 30 TECU. Figure 3 concludes that the geomagnetic storm triggered an asymmetric ionospheric storm, and the occurred phases of the ionospheric storm over different longitudes had a particular diversity.Figure 4 provides an overview of the daily development of storm-effect ionospheric electron density (Ne) as measured by the space-borne Swarm-A satellite that flies at an orbital altitude of ~460 km.Here, the electron density signifies the amount of plasma at the orbital altitude of Swarm-A, which is useful for revealing the change in the topside ionosphere during the storm.The result shown in Figure 3 demonstrates that the storminduced plasma irregularities over the Asian-Australian sector were larger in the main phase, but in the recovery phase, the American sector was larger.Therefore, five Ne profiles derived from Swarm-A over 24-28 August were plotted to describe the Asian-Australian ionospheric response during the main phase.These profiles not only have an adjacent longitude in a ~120 • E to 140 • E span but also have a nearly observed time within 05:10 UT to 06:30 UT, as shown in the Ne tracks in Figure 4a. Figure 4b shows the variation in the daily Ne profiles as a function of the geographic latitude; we can find that the plasma density was kept at a low level with a peak density of ~5 × 10 5 el/cm 3 over 24-25 August.In the main phase, the topside Ne over the Asian-Australian sector significantly enhanced.The red curve expresses a significant structure of EIA on 26 August, the double plasma crests are located at the 10 • N to 20 • N and 0 • to 10 • S latitudinal spans, respectively.Compared to the background values, the storm-enhanced Ne increased ~2 times with a maximum value of ~1.05 × 10 6 el/cm 3 .The enhanced EIA is believed to be caused by a daytime "super-fountain" effect that is driven by the PPEFs.During strong geomagnetic storms, PPEFs of eastward polarity can largely uplift the equatorial ionosphere over the sunlit and post-sunset sectors, which drive the equatorial plasma along the geomagnetic field line to higher altitudes and expand poleward latitudes with a significant enhancement of the EIA [23].In the following two days, the intensity of the EIA gradually decreased to a normal level.
enhanced.The red curve expresses a significant structure of EIA on 26 August, the double plasma crests are located at the 10°N to 20°N and 0° to 10°S latitudinal spans, respectively.Compared to the background values, the storm-enhanced Ne increased ~2 times with a maximum value of ~1.05 × 10 6 el/cm 3 .The enhanced EIA is believed to be caused by a daytime "super-fountain" effect that is driven by the PPEFs.During strong geomagnetic storms, PPEFs of eastward polarity can largely uplift the equatorial ionosphere over the sunlit and post-sunset sectors, which drive the equatorial plasma along the geomagnetic field line to higher altitudes and expand poleward latitudes with a significant enhancement of the EIA [23].In the following two days, the intensity of the EIA gradually decreased to a normal level.Different from the sudden enhanced TECs over the Asian-Australian sector, the TEC's change over the American sector in the main phase (25-26 August) was not significant.However, remarkable TEC enhancements were observed in the recovery phase, especially on 28 August, and the peak density occurred in North America with a value of ~8 × 10 5 el/cm 3 .The hemispheric asymmetrical structure of ionospheric TEC agrees well with the observations of ground-based radars and space-borne Swarm-A.Finally, the TEC's changes over both the Asian-Australian and American sectors reveal that the magnetic storm not only enhanced the equatorial plasma density but also triggered drastic polar ionospheric disturbances.The development of storm-induced polar ionospheric disturbances are investigated in the following section.Different from the sudden enhanced TECs over the Asian-Australian sector, the TEC's change over the American sector in the main phase (25-26 August) was not significant.However, remarkable TEC enhancements were observed in the recovery phase, especially on 28 August, and the peak density occurred in North America with a value of ~8 × 10 5 el/cm 3 .The hemispheric asymmetrical structure of ionospheric TEC agrees well with the observations of ground-based radars and space-borne Swarm-A.Finally, the TEC's changes over both the Asian-Australian and American sectors reveal that the magnetic storm not only enhanced the equatorial plasma density but also triggered drastic polar ionospheric disturbances.The development of storm-induced polar ionospheric disturbances are investigated in the following section.
An ionosonde sounder is a radar that sweeps the high-frequency (HF) band signals and receives the echoes to examine the ionosphere and monitor HF propagation conditions.Ionosondes primarily operate between 1.6 MHz and 12 MHz.With this advantage, the two ionosonde chains located in the Asian-Australian and American sectors can detect the vertical dynamic propagation of storm-induced plasma irregularities, as shown in Figure 5. From the left panels, the radars located in the Asian-Australian sector detected HF signals with frequencies between 5.5 and 7 MHz over DOY 235-240, in the midnight dawn (for local time, it was around noon), and the HF signals accumulated below 200 km.The equatorial radar detected HF plasma signals accumulated at a higher altitude.For example, Figure 5c shows that the ionosonde GU513, located at Guam, measured a maximum frequency of 7 MHz around the altitude of 250 km during a minor solar-geomagnetic activity.After the onset of the geomagnetic storm, the plasma density during the midnight-dawn in the next day (26 August) suddenly enhanced and uplifted.As shown by the black arrows, enhanced HF plasma densities were detected over the equatorial and mid-latitude radars (IC437, GU513, and LM42B) during the main phase.Especially over the radar IC437, the peak height of the ionosphere was uplifted above 350 km, with a peak frequency of 7 MHz, and the intensity of Ne profiles in the Northern Hemisphere was significantly stronger than that over the dip-equator and Southern Hemisphere (GU513 and LM42B), which agreed well with the hemispheric asymmetrical structure of plasma irregularities reported in [26].
conditions.Ionosondes primarily operate between 1.6 MHz and 12 MHz.With this advantage, the two ionosonde chains located in the Asian-Australian and American sectors can detect the vertical dynamic propagation of storm-induced plasma irregularities, as shown in Figure 5. From the left panels, the radars located in the Asian-Australian sector detected HF signals with frequencies between 5.5 and 7 MHz over DOY 235-240, in the midnight dawn (for local time, it was around noon), and the HF signals accumulated below 200 km.The equatorial radar detected HF plasma signals accumulated at a higher altitude.For example, Figure 5c shows that the ionosonde GU513, located at Guam, measured a maximum frequency of 7 MHz around the altitude of 250 km during a minor solar-geomagnetic activity.After the onset of the geomagnetic storm, the plasma density during the midnight-dawn in the next day (26 August) suddenly enhanced and uplifted.As shown by the black arrows, enhanced HF plasma densities were detected over the equatorial and mid-latitude radars (IC437, GU513, and LM42B) during the main phase.Especially over the radar IC437, the peak height of the ionosphere was uplifted above 350 km, with a peak frequency of 7 MHz, and the intensity of Ne profiles in the Northern Hemisphere was significantly stronger than that over the dipequator and Southern Hemisphere (GU513 and LM42B), which agreed well with the hemispheric asymmetrical structure of plasma irregularities reported in [26].The enhanced plasma was detected by the radars located in the American sector.It should be noted that the blank areas over PQ052, GU513, MHJ45, WP937, and JI91J signify that the ionosondes failed to receive the HF echoes, which are associated with various malfunctions, including instrument failure, communications failure, unavailable experimental observation, etc.It was found that storm-enhanced plasma irregularities were observed in both the main and recovery phases.Figure 5h has enough vertical profiles to describe the pattern.Additionally, this storm uplifted the peak height (hmF2) of the ionosphere but in the dawn-forenoon (around midnight for the local time).The largest intensity of the plasma irregularities was detected over the dip-equatorial radar JI91J, with a peak frequency (foF2) of 7 MHz, and the ionospheric peak height was uplifted approximately 50-80 km.Compared to the Asian-Australian sector, the hemispheric asymmetrical structure was not significant.Finally, the detection results of the radars in lowmiddle latitudes manifested that the positive plasma irregularities primarily accumulated between the altitudes of 200 and 300 km.Lastly, the radar PQ052 near the Arctic detected an interesting result, as shown by the red arrow in Figure 5a.During the recovery phase, significant positive plasma irregularities at the altitudes of 110 to 150 km were observed near noon, DOY 239 (27 August).The different altitudinal behaviors of plasma irregularities indicated that significant positive plasma irregularities may be triggered only in the bottomside of the auroral ionosphere, rather than in the equatorial and mid-latitude ionospheres.
Due to the fact that TEC maps and ionosondes are unable to reveal the altitudinal structures of plasma irregularities, the plasma densities over DOY 235-240 2018 were simulated by the TIE-GCM to solve this problem.The averaged plasma over DOY 235-236 was selected as the background value, and the altitudinal changing percentage of storminduced plasma irregularities compared to the background value is shown in Figure 6. Figure 6a-f show the temporal variations of the storm-induced plasma as a function of geographical latitude at the layers spanning from 150 to 500 km along the meridian 110 • .The vertical scale is proportional to the changing percentage, which is represented by a color bar for a better understanding of the storm-enhanced plasma behaviors.
Figure 6 successfully depicts a simulation of the development of double crests of the EIA at the altitudes of 150 to 500 km.In the layer at 150 km, since the main phase on DOY 237, two plasma increments appeared in the 60-65 • N and 40-45 • S geographical latitudes along the meridian at 110 • .On the next day, the amplitude of the plasma enhancements increased to 250%.At the same time, some tiny increments also occurred in the Antarctic.In the 250 km layer, the two plasma crests that were located at the 60-65 • N and 40-45 • S latitudes weakened, while the plasma densities in the Antarctic increased.With an increasing altitude, the plasma increments had an equatorward movement.For example, at the layer of 250 km, a plasma enhancement with a percentage of ~200% appeared in 30-40 • N latitude, and a weaker increment was located at the 40 • S latitude.The two low-latitudinal enhancements moved equatorward within ±20 • latitude at the 350 km layer.Above 350 km, the two plasma crests merged into one unit, and the EIA phenomenon disappeared.
The changes in the ionospheric plasma along the meridian −70 • were found to be consistent with those observed over the Asian-Australian sector.As shown in Figure 6g, two plasma increments appeared at the 150 km layer at the latitudes of 50-60 • N and 60-70 • S, with a maximum percent exceeding 250%.The EIA phenomenon was also observed within the layers of 250 to 350 km, and the EIA crests were enhanced by approximately 200% on DOY 238.In comparison to the Asian-Australian sector, the storm-induced plasma irregularities over the American sector were larger.Figure 6 successfully depicts a simulation of the development of double crests of the EIA at the altitudes of 150 to 500 km.In the layer at 150 km, since the main phase on DOY 237, two plasma increments appeared in the 60-65°N and 40-45°S geographical latitudes along the meridian at 110°.On the next day, the amplitude of the plasma enhancements increased to 250%.At the same time, some tiny increments also occurred in the Antarctic.In the 250 km layer, the two plasma crests that were located at the 60-65°N and 40-45°S latitudes weakened, while the plasma densities in the Antarctic increased.With an increasing altitude, the plasma increments had an equatorward movement.For example, at the layer of 250 km, a plasma enhancement with a percentage of ~200% appeared in 30-40°N latitude, and a weaker increment was located at the 40°S latitude.The two lowlatitudinal enhancements moved equatorward within ±20° latitude at the 350 km layer.Above 350 km, the two plasma crests merged into one unit, and the EIA phenomenon disappeared.
The changes in the ionospheric plasma along the meridian −70° were found to be consistent with those observed over the Asian-Australian sector.As shown in Figure 6g, two plasma increments appeared at the 150 km layer at the latitudes of 50-60°N and 60-70°S, with a maximum percent exceeding 250%.The EIA phenomenon was also observed within the layers of 250 to 350 km, and the EIA crests were enhanced by approximately

Global Maps of GNSS-ROT in the Main and Recovery Phase
The results shown in Figures 3-5 reveal hemispheric asymmetrical structures of the plasma fluctuations over two sectors in different phases.As we know, the rate of total electron content (ROT) expresses the sharpness of the GNSS phase fluctuations caused by ionospheric irregularities and by strong spatial gradients of TEC.Therefore, the ROT is suitable for investigating the development of global storm-induced ionospheric irregularities, and the index is calculated as: where ROT t+1 signifies the change rate of the total electron content at time t + 1, TEC t+1 and TEC t signify the total electron contents at time t + 1 and t, respectively.Figure 8 provides an overview of the global ionospheric irregularities observed during the recovery phase of a geomagnetic storm.The study reveals that the storm induced strong GNSS ROTs, particularly in the American sector.On 10 UT, 26 August, strong plasma irregularities with a level of 0.06 TECU/min were observed over South America, the equator, and mid-high latitudes of North America.Interestingly, a narrow channel of positive ionospheric irregularities was detected in the northwestward direction over nighttime Alaska.Two hours later, a significant hemispheric asymmetry of plasma irregularities developed over the American sector, indicating that the storm-induced plasma over the latitudinal span of 20°S to 45°S was enhanced to larger than 0.08 TECU/min, whereas the ROTs over North America remained at the low level of 0.02 to 0.04 TECU/min.Apart from the American sector, plasma irregularities over other daytime or nighttime continents remained at a low level.However, from 14 UT on 26 August, the hemispheric asymmetry was reversed, with high-magnitude plasma irregularities concentrated in North America.These signatures of ionospheric irregularities persisted for many hours until midnight.Additionally, significant positive plasma irregularities were observed in the daytime Asian-Australian and European-African sectors, as shown in Figure 8g,j.In the following hours, the global ROTs gradually returned to a low level.After approximately 18 UT on 25 August, the AE index rapidly increased to above 500 nT, peaking at ~1300 nT during the main phase.Correspondingly, the daytime plasma irregularities suddenly intensified on 25 August, with significant hemispheric asymmetry observed in the American sector. Figure 7e depicts the enhancement of the equatorial and auroral plasma irregularities over North America, reaching a maximum magnitude of 0.08 TECU/min, as well as intensified mid-latitude plasma with a lower magnitude of 0.06 TECU/min.A narrow channel of positive ionospheric irregularities was observed along the northwestward direction of the western coast of North America, with EPBs reaching a latitudinal extent of 20-25 • N and a maximum value of 0.1 TECU/min.Additionally, the ROT observations over ground-based receivers located at several Pacific islands revealed the occurrence of positive storm-induced EPBs over a longitudinal span of 160-140 • W, with positive ionospheric irregularities observed on both sides of the magnetic equator and a latitudinal extent of up to 25-27 • N/S.The feature of the equatorial ionospheric rate agreed well with the results reported in [23].In the nighttime hemisphere, ionospheric irregularities were negative with a low intensity.From 26 August, the TEC rate over the western coast of North America and Greenland turned negative, although some negative irregularities were observed under the sunlit sector.However, the ROT over the European-African sector gradually turned positive, reaching its largest magnitude during 04-06 UT on 26 August.Positive irregularities with a maximum value of 0.1 TECU/min were observed in Europe, Africa, and Asia, but not in Australia, although it was also under the sunlit sector.The results conclude that the hemispheric asymmetry of the plasma irregularities was significant during the end of the main phase, with the ROTs over Africa being larger than that over Europe, while the opposite was observed in the Asian-Australian sector.
Figure 8 provides an overview of the global ionospheric irregularities observed during the recovery phase of a geomagnetic storm.The study reveals that the storm induced strong GNSS ROTs, particularly in the American sector.On 10 UT, 26 August, strong plasma irregularities with a level of 0.06 TECU/min were observed over South America, the equator, and mid-high latitudes of North America.Interestingly, a narrow channel of positive ionospheric irregularities was detected in the northwestward direction over nighttime Alaska.Two hours later, a significant hemispheric asymmetry of plasma irregularities developed over the American sector, indicating that the storm-induced plasma over the latitudinal span of 20 • S to 45 • S was enhanced to larger than 0.08 TECU/min, whereas the ROTs over North America remained at the low level of 0.02 to 0.04 TECU/min.Apart from the American sector, plasma irregularities over other daytime or nighttime continents remained at a low level.However, from 14 UT on 26 August, the hemispheric asymmetry was reversed, with high-magnitude plasma irregularities concentrated in North America.These signatures of ionospheric irregularities persisted for many hours until midnight.Additionally, significant positive plasma irregularities were observed in the daytime Asian-Australian and European-African sectors, as shown in Figure 8g,j.In the following hours, the global ROTs gradually returned to a low level.

Potential Drivers of Equatorial and Mid-Latitude Ionospheric Irregularities
Changes in thermospheric composition are an important driver of positive or negative ionospheric irregularities.Therefore, the O/N2 density ratio measured by the Global UltraViolet Imager (GUVI) on board the space-borne TIMED satellite (~625 km) was analyzed.It is important to note that the O/N2 density ratio is a height-integrated quantity within the orbit altitudes of the GNSS constellation and the GUVI satellite.In addition, global topside TECs derived from Swarm-A and Swarm-B were investigated.TEC signifies the integrated electrons within the altitudes from the orbit of the Swarm microsatellite to the orbit of the GNSS constellation. Figure 9 provides an overview of  Changes in thermospheric composition are an important driver of positive or negative irregularities.Therefore, the O/N 2 density ratio measured by the Global UltraViolet Imager (GUVI) on board the space-borne TIMED satellite (~625 km) was analyzed.It is important to note that the O/N 2 density ratio is a height-integrated quantity within the orbit altitudes of the GNSS constellation and the GUVI satellite.In addition, global topside TECs derived from Swarm-A and Swarm-B were investigated.TEC signifies the integrated electrons within the altitudes from the orbit of the Swarm microsatellite to the orbit of the GNSS constellation. Figure 9 provides an overview of daily topside TEC and O/N 2 during the storm.It is also worth noting that the TEC distribution differs slightly among the data from the two satellites, most likely due to the ~80 km difference in altitude.Both the Swarm-A and Swarm-B profiles conclude that the topside TECs over the Asian-Australian and American sectors were quiet before 12 UT on 25 August, with an average TEC of under 6 TECU. Figure 9b shows that the TEC profile over the American sector suddenly increased, and the expanded profiles covered the eastern Pacific.This phenomenon was also validated by the observation of Swarm-B.During 00-12 UT on 26 August, the TEC over the Asia-Australia sector increased remarkably, with a maximum value exceeding 12 TECU.After, the enhanced TEC profiles gradually decreased and returned to a normal level.The profiles derived from Swarm-B agreed well with those of Swarm-A in that large-scale positive TEC irregularities appeared over the Asian-Australian and American sectors.Figure 9i illustrates that the density ratio O/N2 exhibited an inverse relationship with geographical latitude, with a range of 0.3 to 0.6 on quiet days.However, the O/N2 ratio underwent a sudden change with the onset of a storm.Specifically, the O/N2 in low to middle latitudes increased significantly, reaching a maximum value of 0.8, while in polar regions, it decreased to 0.2.Moreover, the enhanced O/N2 had a southward excursion in the American sector.As shown in Figure 9j, the O/N2 in North America was approximately 0.4, while in South America, it increased to 0.8.In the following two days, the storm-induced O/N2 gradually decreased as the geomagnetic field returned to normal levels.The change in the O/N2 density ratio was consistent with the TEC irregularities, indicating that it could be a crucial driver in generating plasma disturbances.O/N2 has a strong positive correlation with plasma density and is a reliable indicator of neutral composition disturbances for analyzing ionospheric storms [28].The loss rate of ionospheric ion density is proportional to the molecular concentration.An increase in the mean molecular mass results in a decrease in the electron density, whereas a decrease in the molecular concentration leads to a positive disturbance.
At altitudes of 90 to 130 km, many electrons move westward under the influence of the dayside electric field.This westward electron flow generates a dayside eastward electric current through the equatorial dynamo effect, which is known as the equatorial Figure 9i illustrates that the density ratio O/N 2 exhibited an inverse relationship with geographical latitude, with a range of 0.3 to 0.6 on quiet days.However, the O/N 2 ratio underwent a sudden change with the onset of a storm.Specifically, the O/N2 in low to middle latitudes increased significantly, reaching a maximum value of 0.8, while in polar regions, it decreased to 0.2.Moreover, the enhanced O/N 2 had a southward excursion in the American sector.As shown in Figure 9j, the O/N 2 in North America was approximately 0.4, while in South America, it increased to 0.8.In the following two days, the storm-induced O/N 2 gradually decreased as the geomagnetic field returned to normal levels.The change in the O/N 2 density ratio was consistent with the TEC irregularities, indicating that it could be a crucial driver in generating plasma disturbances.O/N 2 has a strong positive correlation with plasma density and is a reliable indicator of neutral composition disturbances for analyzing ionospheric storms [28].The loss rate of ionospheric ion density is proportional to the molecular concentration.An increase in the mean molecular mass results in a decrease in the electron density, whereas a decrease in the molecular concentration leads to a positive disturbance.
At altitudes of 90 to 130 km, many electrons move westward under the influence of the dayside electric field.This westward electron flow generates a dayside eastward electric current through the equatorial dynamo effect, which is known as the equatorial electrojet (EEJ).During strong geomagnetic storms, the EEJ can be severely disrupted by disturbed electric fields from the magnetosphere.The signature of the EEJ can be estimated by taking the difference between the horizontal components measured by a pair of magnetometers located off the equator and at the equator.In this study, the horizontal components of the PHU, DLT, SJG, and HUA magnetometers located in the Asian-Australian and American sectors were used to investigate the storm-induced changes in the EEJ.The observations were obtained from the International Real-Time Magnetic Observatory Network (https://intermagnet.org/index-eng.php,Accessed on 20 May 2021).
Figure 10a,b illustrate the changes in the EEJ signatures I, Accessed on 20 May 2021 the Asian-Australian and American sectors in response to the geomagnetic storm that occurred on 25 August 2018, as well as the corresponding variations in the horizontal components of the PHU, DLT, SJG, and HUA magnetometers.It was observed that the Dst index and magnetometers' horizontal components experienced a sudden decrease during the storm.In Figure 10a, the equatorial magnetometer GUA recorded a minimum value of approximately −270 nT during UT 06-08 on 26 August, and the differential component between GUA and KAK showed a negative perturbation with a minimum value of −100 nT.This was accompanied by a maximum TEC enhancement of 18 TECU observed at the PIMO station during the most severe moment.The changes in the EEJ, estimated by the difference between HUA and KOU, as shown in Figure 10b, were consistent with those observed in the Asian-Australian sector.The EEJ signature showed two distinct perturbations in the afternoons of 26-27 August, with an amplitude of approximately 50 nT.The differential TEC over the UNSA station also corresponded well with the changes in the EEJ signature.During the recovery phase of the geomagnetic storm, two TEC enhancements were observed, with the maximum delta TEC reaching 14 TECU.These results suggest that changes in the EEJ may be a significant driver in triggering storm-effect TEC disturbances.However, it should be noted that the slight fluctuations in the EEJ cannot fully account for the strong TEC enhancements, indicating the need for further analysis of other drivers responsible for the significant ionospheric disturbances.
The equatorial plasma fountain effect plays a dominant role in generating the equatorial ionization anomaly (EIA) by driving equatorial plasmas upward to higher altitudes via the vertical E×B drift.The accumulated plasmas then diffuse down to higher latitudes along the geomagnetic field lines, resulting in two highly concentrated plasma crests on either side of the magnetic equator.A stronger E×B drift in the fountain process can lift more plasmas to higher altitudes, resulting in stronger and more poleward EIA crests generated by the plasma diffusion process.Therefore, changes in the vertical E×B drift may be another driver of equatorial and mid-latitude plasma perturbations.The latitudinal changes of the vertical E×B drifts along the −70 • and 110 • longitudes were simulated using the TIE-GCM, as shown in Figure 10c,d.Figure 10c shows that the E×B drifts along the meridian −70 • began to increase from UT 14 on 25 August, with a slight enhancement of 3 to 5 m/s.However, the nighttime E×B drifts suddenly weakened, with a maximum decrement of −15 m/s at dawn on 26 August (local time).Subsequently, the daytime differential E×B drifts turned positive from UT 8 on 26 August, with a maximum increment of 15 m/s.On the forenoon of 27 August (DOY 239), the differential E×B had a hemispheric asymmetrical structure, with the differential E×B on DOY 239 enhancing by ~5 m/s in the Southern Hemisphere.In the Asian-Australian sector, the E×B drift enhanced from 20 UT on 25 August, with a magnitude of 5 to 10 m/s (local time).A slight positive E×B irregularity was also observed on the following day.
with the changes in the EEJ signature.During the recovery phase of the geomagnetic storm, two TEC enhancements were observed, with the maximum delta TEC reaching 14 TECU.These results suggest that changes in the EEJ may be a significant driver in triggering storm-effect TEC disturbances.However, it should be noted that the slight fluctuations in the EEJ cannot fully account for the strong TEC enhancements, indicating the need for further analysis of other drivers responsible for the significant ionospheric disturbances.The equatorial plasma fountain effect plays a dominant role in generating the equatorial ionization anomaly (EIA) by driving equatorial plasmas upward to higher altitudes via the vertical E×B drift.The accumulated plasmas then diffuse down to higher latitudes along the geomagnetic field lines, resulting in two highly concentrated plasma crests on either side of the magnetic equator.A stronger E×B drift in the fountain process can lift more plasmas to higher altitudes, resulting in stronger and more poleward EIA crests generated by the plasma diffusion process.Therefore, changes in the vertical E×B drift may be another driver of equatorial and mid-latitude plasma perturbations.The latitudinal changes of the vertical E×B drifts along the −70° and 110° longitudes were simulated using the TIE-GCM, as shown in Figure 10c,d.Figure 10c shows that the E×B drifts along the meridian −70° began to increase from UT 14 on 25 August, with a slight enhancement of 3 to 5 m/s.However, the nighttime E×B drifts suddenly weakened, with a maximum decrement of −15 m/s at dawn on 26 August (local time).Subsequently, the The results of this study not only confirm some previous papers but also conclude new findings.For instance, Astafyeva et al. [29] utilized ground-based instruments, such as GNSS, SuperDARN, and ionosondes, as well as space-borne tools, including Swarm, CSES, and DMSP, to detect significant storm-time enhancements in the Asian and American sectors on a large scale.The storm-enhanced equatorial E×B drift was the main driver for the most significant 100-300% large-scale enhancement occurred over the Asian region.In the American sector, the complex plasma irregularities were caused by the action of multiple drivers.This study confirms the enhanced vertical E×B drift successfully, which coincide with the Astafyeva et al.'s study.However, Figures 9 and 10 also indicate that the thermospheric density ratio of O/N 2 and equatorial electrojet were severely impacted by the strong geomagnetic storm.Specifically, the equatorial electrojet was activated by the disturbed electric field that penetrated from the magnetosphere, and the changes in vertical E×B drifts may be associated with PPEFs and DDEFs.Therefore, it is believed that the irregularities in the equatorial and mid-latitude ionosphere are the result of a combination of multiple physical-chemical processes.The enhanced density ratio O/N 2 , vertical E×B drift, and equatorial electrojet played a decisive role in inducing positive irregularities.In the recovery phase, the hemispheric asymmetrical O/N 2 and E×B drift on 27 August may be responsible for the asymmetrical TEC observed over the American sector, as shown in Figure 3.

Potential Drivers of Auroral Ionospheric Irregularities
The results from the GNSS ROT, sounder density profiles, and TIE-GCM simulations reveal significant auroral ionospheric irregularities induced by the storm.To further investigate the vertical structures of these irregularities, the TIE-GCM simulated plasma irregularities between altitudes of 96 and 400 km along the −70 • , 0 • , and 110 • longitudes, as shown in Figure 11a-c.In Figure 6, the TIE-GCM simulation finds that the largest plasma irregularities, with a changing of 250%, are located in the 70 • S-80 • S latitude span at a layer of 150 km.Therefore, the geographical latitude of 77.5 • S was selected, as shown in Figure 11a-c.It is important to note that the vertical scale of each panel is similar to Figure 7 but represents the amplitude of the plasma irregularities.The results show significant plasma enhancements in the topside and bottom-side of the Antarctic ionosphere along all three longitudes.During the main phase, the increment along the meridian −70 • was the largest, with a value up to 6 × 10 5 el/cm 3 , followed by the meridian 0 • , while the weakest increment of 4 × 10 5 el/cm 3 occurred along the meridian 110 • .It is noted that the difference between the −70 • and 110 • meridians is 180 • , and the occurrence times (LT) of largest increments over two sectors are opposite.In addition, plasma fluctuations were also observed in the bottom-side ionosphere along the meridians −70 Was the disturbance caused by the geomagnetic storm limited only to the bottomside ionosphere in western Antarctica?To investigate, we analyzed the temporal variations in the differential bottom-side plasma (150 km) as a function of geographical latitude along the meridians −70°, 0°, and 110°, as shown in Figure 11d-f.The results indicate that the storm-induced plasma increments were significant in the bottom-side ionosphere across all three longitudinal sectors, but the geographic latitudes of the increments varied.In Figure 11d, two plasma crests were located at latitudes 60°N and 80°S.These crests moved northward in the Eastern Hemisphere, with the plasma crests in the Asian-Australian sector shifting to latitudes 80°N and 60°S, with weaker values of 3 × 10 5 el/cm 3 .The latitudinal motion of bottom-side plasma enhancements in different longitudinal sectors was influenced by the asymmetrical structure of the geomagnetic field.
Figure 11g-i depict the longitudinal structures of the differential plasma (150 km) as a function of the day of year (DOY) for three geographic latitudes: 77.5°S, 67.5°S, and 57.5°S.In Figure 11g, positive plasma irregularities occurred in the −120°~0° longitudinal span on 26 August, with a value of 4 × 10 5 el/cm 3 .Along the 67.5°S latitude, the plasma irregularities had a double-peak structure in the −180°~60° longitudinal span.Along the 57.5°S latitude, double-peak plasma increments were observed in the 0°~90° and 150°~−120° longitudinal spans.The intensity of the bottom-side plasma irregularities decreased in the middle geographical latitudes compared to the auroral plasma irregularities.The results suggest that the strong storm not only induced topside plasma Was the disturbance caused by the geomagnetic storm limited only to the bottom-side ionosphere in western Antarctica?To investigate, we analyzed the temporal variations in the differential bottom-side plasma (150 km) as a function of geographical latitude along the meridians −70 • , 0 • , and 110 • , as shown in Figure 11d-f.The results indicate that the storm-induced plasma increments were significant in the bottom-side ionosphere across all three longitudinal sectors, but the geographic latitudes of the increments varied.In Figure 11d, two plasma crests were located at latitudes 60 • N and 80 • S.These crests moved northward in the Eastern Hemisphere, with the plasma crests in the Asian-Australian sector shifting to latitudes 80 • N and 60 • S, with weaker values of 3 × 10 5 el/cm 3 .The latitudinal motion of bottom-side plasma enhancements in different longitudinal sectors was influenced by the asymmetrical structure of the geomagnetic field.
Figure 11g-i depict the longitudinal structures of the differential plasma (150 km) as a function of the day of year (DOY) for three geographic latitudes: 77.5 • S, 67.5 • S, and 57.5 • S. In Figure 11g, positive plasma irregularities occurred in the −120 • ~0• longitudinal span on 26 August, with a value of 4 × 10 5 el/cm 3 .Along the 67.5 • S latitude, the plasma irregularities had a double-peak structure in the −180 • ~60 • longitudinal span.Along the 57.5 • S latitude, double-peak plasma increments were observed in the 0 • ~90 • and 150 • ~−120 • longitudinal spans.The intensity of the bottom-side plasma irregularities decreased in the middle geographical latitudes compared to the auroral plasma irregularities.The results suggest that the strong storm not only induced topside plasma fluctuations but also triggered positive bottom-side plasma irregularities near the auroral zone (~>50 • N/S), which is consistent with the sounder profiles of the radar PQ052, as shown in Figure 5a.
To explain the development of the auroral ionospheric irregularities, the TIE-GCM was used to simulate the Joule heating and the O + and O 2 + ion densities within the altitudes of 100 to 400 km.The neutral mass density decreased exponentially with height, and the Joule heating per unit mass was much larger at higher altitudes than at lower altitudes.Neglecting the bottom-side change, the ratio of changed Joule heating during storms compared to quiet background values was also investigated.The study area in the Arctic was located at 110 • E, 67.5 • N. Figure 12a-d show the vertical changes in the Joule heating, the ratio of enhanced Joule heating, and the O + and O 2 + ion density over the Arctic.The results show that the Joule heating was enhanced from the main phase with a magnitude of 1 × 10 4 erg/g/s, reached a maximum in the recovery phase with a value of 3 × 10 4 erg/g/s, and gradually recovered to background values thereafter.In the Antarctic, the study area was selected at 70°W, 77.5°S.The changes in the Joule heating over the Antarctic were found to be in good agreement with those over the Arctic.However, positive Joule heating disturbances only appeared in the main phase, which is consistent with the variation in the storm-effect Antarctic ionospheric plasma, as shown in Figure 11a.The magnitude of the enhanced Joule heating over the Antarctic was several Similar to the variations in the topside Joule heating, the positive Joule heating disturbances also appeared below 200 km on DOY 237-238, with a slight value of 1 × 10 4 erg/g/s.Figure 12b shows that the Joule heating in the main and recovery phases enhanced more than 200 times compared to the background values, with the maximum Joule heating enhancements located at altitudes of 100-150 km.The changes in the O + ion density, as shown in Figure 12c, agree well with the Joule heating, with the O + ion density increasing from DOY 237 and growing stronger on DOY 238 above the 200 km layer, with the maximum value reaching 1.5 × 10 5 /cm 3 .An O 2 + increment generated from UT 14, DOY 237, grew to 1 × 10 5 el/cm 3 on DOY 238.Unlike the O + ions, the O 2 + increments mainly occurred below the 200 km layer, consistent with the behavior of the bottom-side enhanced Joule heating, as shown in Figure 12b.
In the Antarctic study area, located at 70 • W, 77.5 • S, the changes in the Joule heating were found to agree with those over the Arctic.However, the positive Joule heating disturbances only appeared in the main phase, which is consistent with the variation in the storm-effect on Antarctic ionospheric plasma, as shown in Figure 11a.The magnitude of the enhanced Joule heating over the Antarctic was several times larger than that over the Arctic.Figure 12e shows that the topside and bottom-side Joule heating were enhanced by approximately 8 × 10 4 erg/g/s and 2 × 10 4 erg/g/s, respectively.However, Figure 12f indicates that the changed ratio of the bottom-side Joule heating in the main phase was significantly larger than that in the topside, and the maximum enhanced ratio exceeded 1 × 10 4 .Similar to Figure 12c-d, in the main phase, the O + ion density over the Antarctic was enhanced by approximately 5 × 10 5 /cm 3 above the 200 km layer, and the bottom-side O 2 + ion density was enhanced by approximately 2 × 10 5 el/cm 3 .On DOY 237-238, a positive Joule heating disturbance also appeared under 200 km with a slight value of 1 × 10 4 erg/g/s, similar to the variation of the topside Joule heating.Figure 12b shows that in the main and recovery phases, the Joule heating enhancements were more than 200 times that of the background values, and the maximum enhancements were located at altitudes of 100-150 km.The changes in the O + ion density, as shown in Figure 12c, agreed well with the Joule heating, with the O + ion density enhancing from DOY 237 and growing stronger on DOY 238 above the 200 km layer, reaching a maximum value of 1.5 × 10 5 /cm 3 .On DOY 238, an O 2 + increment was generated at UT 14, growing to 1 × 10 5 el/cm 3 , and mainly occurring under the 200 km layer, consistent with the behaviors of the bottom-side enhanced Joule heating, as shown in Figure 12b, which is different from the behavior of the O + ions.
Similar to the variations in topside Joule heating, a positive Joule heating disturbance also appeared below 200 km on DOY 237-238, with a slight value of 1 × 10 4 erg/g/s.Figure 12b shows that, different from the absolute change in the differential Joule heating in Figure 12a, the Joule heating in the main and recovery phases was enhanced by more than 200 times that of the background values, with the maximum enhancements located at altitudes of 100-150 km.The changes in the O + ion density, as shown in Figure 12c, showed good agreement with the Joule heating, with the O + ion density increasing from DOY 237 and growing stronger on DOY 238 above the 200 km layer, with a maximum value of 1.5 × 10 5 /cm 3 .An O 2 + increment was generated from UT 14, DOY 237, and grew to 1 × 10 5 el/cm 3 on DOY 238.Different from the O + ions, the O 2 + increments mainly occurred under the 200 km layer, which is consistent with the behavior of the bottom-side enhanced Joule heating, as shown in Figure 12b.
In the Antarctic, the study area was selected at 70 • W, 77.5 • S. The changes in the Joule heating over the Antarctic were found to be in good agreement with those over the Arctic.However, positive Joule heating disturbances only appeared in the main phase, which is consistent with the variation in the storm-effect Antarctic ionospheric plasma, as shown in Figure 11a.The magnitude of the enhanced Joule heating over the Antarctic was several times greater than that over the Arctic.Figure 12e illustrates that the topside and bottomside Joule heating was enhanced by approximately 8 × 10 4 erg/g/s and 2 × 10 4 erg/g/s, respectively.However, Figure 12f indicates that the changed ratio of the bottom-side Joule heating in the main phase was significantly larger than that in the topside, with the maximum enhanced ratio exceeding 1 × 10 4 .Similar to Figure 12c,d, in the main phase, the O + ion density over the Antarctic was enhanced by approximately 5 × 10 5 /cm 3 above the 200 km layer, while the bottom-side O 2 + ion density was enhanced by approximately 2 × 10 5 el/cm 3 .
The amplitudes of the Joule heating and the O + and O 2 + ion density over the Antarctic were considerably stronger than those over the Arctic, which is consistent with the magnitude of the polar ionospheric plasma disturbance, as depicted in Figure 11.During a space weather event, sudden enhanced energy could ionize the primary neutral gases, O 2 and N 2 , resulting in an increase in the ion density [30].Our simulations have confirmed the theory that enhanced Joule heating can accelerate the Polar ionospheric ionization process, and the enhanced O + and O 2 + ion densities are responsible for the topside and bottom-side plasma increments, respectively.
Geomagnetic storms not only create storm-enhanced densities (SEDs) in low-middle latitudes and tongues-of-ionization at the polar cap but also alter the global magnetic field and strengthen the ionospheric-magnetospheric current systems, as noted in [31].Since the ionosphere acts as a conductor, electric fields in the polar ionosphere can generate ionospheric currents similar to that of close field-aligned currents flowing in the ionosphericmagnetospheric system.This phenomenon can lead to Joule heating in the upper atmosphere and even influence the circulation of ionospheric plasma, thereby altering the polar ionospheric electron density structure.The ionospheric electric potential contour maps, estimated using the Super Dual Auroral Radar Network's (SuperDARN) Assimilative Mapping procedure, accessible at http://vt.superdarn.org/tiki-index.php?page=ASCIIData, Accessed on 20 May 2021, were utilized to examine the spatial-temporal variations in the polar convection patterns during the main phase of the geomagnetic storm.
Figure 13 shows the red and blue contours indicating the positive and negative ionospheric electric potentials, respectively.Typically, the potential pattern reaches a maximum near dawn and a minimum near dusk, and the difference between these values is referred to as the cross-polar voltage.The spatial-temporal evolution of the electric potentials over the Arctic from UT 20, 25 August, to UT 18, 26 August, is presented in Figure 13a-f.At UT 20, 25 August, the negative electric potential with a minimum magnitude of −32 kv occurred in a −60 • ~120 • longitudinal span, whereas a positive electric potential with a maximum value of 41 kv was distributed in a −90 • ~−180 • longitudinal span.Starting from 26 August, the polar electric potential remarkably intensified with the decreasing Dst index, and the strongest electric potential was observed in UT 04-08, 26 August, coinciding with a peak AE index of 1500 to 2000 nT.For instance, in Figure 13d, the negative potential decreased to −53 kv, while the positive potential increased to 36 kv, resulting in a cross-polar voltage of 89 kv.During the main phase's end, the convection zone in the Arctic extended to approximately 50 • N, consistent with the geographical latitude of radar PQ052, as shown in Figure 2a, the station with the minimum latitude capable of detecting bottom-side ionospheric regularities.Similarly, the scale and scope of the electric potential over Antarctica significantly expanded and intensified.The negative potential decreased from −32 kv at UT 20, 25 August, to −61 kv at UT 04, 26 August, while the corresponding positive potential reduced from 40 kv to 33 kv.The maximum cross-polar voltage on UT 04, 26 August, reached 94 kv.The scale of the Antarctic storm-effect electric potential was stronger than that over the Arctic, although the scope was smaller.
During active space weather events, sudden energy and momentum are deposited in the high-latitude ionosphere and thermosphere, primarily in the form of particle precipitation and Joule heating.The incident particles gradually transfer energy to different ionospheric layers, ionizing more charged particles as the energy deposition increases.The accelerated ionization process enhances the ionospheric current flowing in the medium, while particle precipitation and Joule heating control the variations in the short-scale structures of the ionosphere-thermosphere, resulting in increased electrical conductivity and heating of the ionosphere-thermosphere system.TIE-GCM simulations of the 5 April 2010 geomagnetic storm concluded that additional particle precipitation not only significantly increases ionospheric conductivity but also causes remarkable Joule heating enhancements [32].The enhanced conductivity, electric field, or a combination of both could intensify the ionospheric electric currents.The current density is directly proportional to ionospheric conductivity, which is itself proportional to the plasma density [33].Thus, there is a close connection among magnetosphere energy deposition, particle precipitation, ionospheric current intensification, Joule heating, and SEDs generation.Figures 12 and 13 reveal that the plasma density, Joule heating, and ionospheric electric potential affected by the storm were all significantly enhanced, further confirming the charged particle diffusion process reported in the previous literature.Therefore, it is believed that the storm-induced polar plasma irregularities are associated with the additional energy input from the particle precipitation, Joule heating, and ionospheric current intensification.During active space weather events, sudden energy and momentum are deposited in the high-latitude ionosphere and thermosphere, primarily in the form of particle precipitation and Joule heating.The incident particles gradually transfer energy to

Conclusions
The 25-27 August geomagnetic storm was a surprising space event that occurred over a period of very low solar activity.The prominent features of the global ionospheric response to this strong geomagnetic storm were analyzed using ground-based instruments, as well as space-borne constellations.The storm triggered several unusual ionospheric plasma irregularities that depended on geographical longitude, latitude, and altitude.The potential drivers for explaining these irregularities were discussed using observations from magnetometers, GUVI profiles, and TIE-GCM simulations.Some important conclusions are drawn as follows: (1) In the Asian-Australian sector, the observations of the GNSS receivers indicate that the storm enhanced equatorial and mid-latitude TEC to a maximum value of 40 TECU at the end of the main phase and beginning of the recovery phase.In the American sector, this storm triggered a remarkable hemispheric asymmetry in TEC during the recovery phase, with TEC depletion occurring in North America and low-level TEC enhancements occurring in the mid-latitudes of South America.This phenomenon was also validated by ionospheric topside profiles derived from Swarm-A, with space-borne observations confirming not only plasma density enhancements over the Asian-Australian and American sectors that occurred during the main and recovery phases, respectively, but also enhanced double-peak crests of EIA caused by a daytime "super-fountain" effect driven by PPEFs.
(2) The sounder profiles of ionosondes revealed that the storm induced positive plasma irregularities in low-middle latitudes, and the enhanced plasma irregularities primarily accumulated in altitudes of 200 to 300 km, with a maximum frequency of 7 MHz.In addition, the profiles of the ionosonde PQ052 near the Arctic zone revealed an interesting finding: the storm could trigger positive plasma irregularities in the bottom-side (<150 km) ionosphere near the auroral zone, which was confirmed by the simulation of TIE-GCM.Furthermore, the TIE-GCM also successfully captured the development of the double crests of EIA in the altitudes of 250 to 400 km.
(3) The global ROT maps reveal a remarkable hemispheric asymmetry of plasma irregularities at a particular time.During the beginning of the main phase, the ROT in the American sector showed a hemispheric asymmetrical structure, with larger plasma irregularities in North America than in South America.Towards the end of the main phase, plasma irregularities over Africa were larger than those in Europe, while the opposite was observed in the Asian-Australian sector.In the recovery phase, GNSS receivers not only detected large plasma irregularities in nighttime Alaska but also found a new hemispheric asymmetry in the American sector.
(4) The GUVI profiles indicated that the storm also induced significant changes in thermospheric composition over 26-27 August.The equatorial electric field (EEJ) was slightly affected by the enhanced field caused by the geomagnetic storm.Observations of the magnetosphere demonstrated that slight positive fluctuations in the EEJ occurred in the Asian-Australian and American sectors.The simulations of the TIE-GCM concluded that the daytime E×B drifts were enhanced during the storm, and the enhanced E×B drifts reinforced the equatorial fountain effect and strengthened the ionospheric double-peak structure.The equatorial and mid-latitude plasma irregularities are believed to be the result of a combined action of thermospheric composition changes, the equatorial electrojet, and vertical E×B drifts.
(5) The simulations of the TIE-GCM revealed that the geomagnetic storm had multiple effects on the ionosphere.Firstly, it enhanced the topside ionospheric plasma density and triggered positive plasma irregularities in the bottom-side ionosphere near the auroral zone.Secondly, positive Joule heating irregularities were observed in the altitudes of 100 to 400 km in both the Arctic and Antarctic, and the temporal-spatial changes were consistent with the behaviors of the polar plasma irregularities.The enhanced density of the O + ions was responsible for the topside plasma irregularities, while the increment in the O 2 + ion density may be a dominant driver for the positive bottom-side plasma irregularities.Finally, the polar ionospheric electric potential suffered severely during the storm, with the cross-polar voltage abruptly increasing to 89 kv and 94 kv in the Arctic and Antarctic, respectively.The polar ionospheric irregularities may be associated with additional energy input through particle precipitation, Joule heating, and ionospheric currents intensification.

Figure 1 .
Figure 1.Spatial distribution of GNSS stations, ionosondes, and magnetometers.The blu pentagrams signify global GNSS stations, the red pentagrams signify the ionosondes located at th Asian-Australian and American sectors, the green triangles signify magnetometers, and the soli and dashed curves depict the location of the magnetic equator and the region of equatorial electroje

Figure 1 .
Figure 1.Spatial distribution of GNSS stations, ionosondes, and magnetometers.The blue pentagrams signify global GNSS stations, the red pentagrams signify the ionosondes located at the Asian-Australian and American sectors, the green triangles signify magnetometers, and the solid and dashed curves depict the location of the magnetic equator and the region of equatorial electrojet.
Remote Sens. 2023, 15, 2549 5 of 25 classification of the geomagnetic storm released by the National Oceanic and Atmospheric Administration (NOAA), this storm was classed as "strong".

Figure 3 .
Figure 3. Latitudinal TEC changes along the (a) 110 • E and (b) 70 • W longitudes over 24-28 August.The dashed, black line signifies the initial phase (IP) of the geomagnetic storm; the dashed, red line signifies the main phase (MP); and the dashed, green line signifies the recovery phase (RP).The white color depicts empty cells due to the lack of actual observations.

Figure 4 .
Figure 4. Daily changes in the ionospheric electron density detected using a Langmuir probe instrument onboard the Swarm-A satellite during (a,b) the main phase over the Asian-Australian sector and (c,d) the recovery phase over the American phase from 24 to 28 August.

Figure 4 .
Figure 4. Daily changes in the ionospheric electron density detected using a Langmuir probe instrument onboard the Swarm-A satellite during (a,b) the main phase over the Asian-Australian sector and (c,d) the recovery phase over the American phase from 24 to 28 August.

Figure 5 .
Figure 5. Vertical sounder profiles of two chain ionosondes located at the Asian-Australian (a-d) and American (e-h) sectors in a whole phase of geomagnetic storm.The dashed line signifies the onset of the main phase of the storm.

Figure 7 25 Figure 7 .
Figure 7. Global maps of GNSS ROTs from 10UT, 25 August (a) to 08UT, 26 August (l) with an interval of 2 Hours.The black line signifies the magnetic equator, and the shaded areas signify nighttime.

Figure 7 .
Figure 7. Global maps of GNSS ROTs from 10UT, 25 August (a) to 08UT, 26 August (l) with an interval of 2 Hours.The black line signifies the magnetic equator, and the shaded areas signify nighttime.

25 Figure 8 .
Figure 8. Global maps of GNSS ROTs from 10UT, 26 August (a) to 08UT, 27 August (l) with an interval of 2 Hours.The black line signifies the magnetic equator, and the shaded areas signify nighttime.

Figure 8 .
Figure 8. Global maps of GNSS ROTs from 10UT, 26 August (a) to 08UT, 27 August (l) with an interval of 2 Hours.The black line signifies the magnetic equator, and the shaded areas signify nighttime.

Figure 9 .
Figure 9. Topside TEC variations measured by GPS receivers on board (a-d) Swarm-A and (e-h) Swarm-B over 25-26 August; (i-l) thermospheric O/N2 density ratio as measured by the Global UltraViolet Imager over 25-28 August.The thin, black curve signifies the magnetic dip equator.

Figure 9 .
Figure 9. Topside TEC variations measured by GPS receivers on board (a-d) Swarm-A and (e-h) Swarm-B over 25-26 August; (i-l) thermospheric O/N2 density ratio as measured by the Global UltraViolet Imager over 25-28 August.The thin, black curve signifies the magnetic dip equator.

Figure 10 .
Figure 10.Variations in the horizontal intensity of the geomagnetic field, equatorial electrojet (EEJ), differential TECs, and Dst value in the (a) Asian-Australian and (b) American sectors, as well as the differential latitudinal vertical E×B drifts along the (c) −70° and (d) 110° longitudes.

Figure 10 .
Figure 10.Variations in the horizontal intensity of the geomagnetic field, equatorial electrojet (EEJ), differential TECs, and Dst value in the (a) Asian-Australian and (b) American sectors, as well as the differential latitudinal vertical E×B drifts along the (c) −70 • and (d) 110 • longitudes.

Figure 12 .
Figure 12.Vertical changes in the Joule heating, ratio of enhanced Joule heating, and O + and O2 + ion density as simulated by the TIE-GCM over the locations (a-d) 110°E, 67.5°N and (e-h) 70°W, 77.5°S over DOY 237-240.

Figure 12 .
Figure 12.Vertical changes in the Joule heating, ratio of enhanced Joule heating, and O + and O 2 + ion density as simulated by the TIE-GCM over the locations (a-d) 110 • E, 67.5 • N and (e-h) 70 • W, 77.5 • S over DOY 237-240.

Figure 13 .
Figure 13.Maps of the electric potentials from SuperDARN over the (a-f) Arctic and (g-l) Antarctic over 25-26 August 2018.The red and blue contours signify positive and negative potentials, respectively.

Figure 13 .
Figure 13.Maps of the electric potentials from SuperDARN over the (a-f) Arctic and (g-l) Antarctic over 25-26 August 2018.The red and blue contours signify positive and negative potentials, respectively.