A Statistical Analysis of the Morphology of Storm‐Enhanced Density Plumes Over the North American Sector

The storm‐enhanced density (SED) is a large‐scale midlatitude ionospheric electron density enhancement in the local afternoon sector, which exhibits substantial spatial gradients and thus can impose detrimental effects on modern navigation and communication systems, causing potential space weather hazards. This study has identified a comprehensive list of 49 SED events over the continental US and adjacent regions, by examining strong geomagnetic storms occurring between 2000 and 2023. The ground‐based Global Navigation Satellite System (GNSS) total electron content and data from a new TEC‐based ionospheric data assimilation system were used to analyze the characteristics of SED. For each derived SED events, we have quantified its morphology by employing a Gaussian function to parameterize key characteristics of the SED, such as the plume intensity, central longitude, and half‐width. A statistical analysis of SEDs was conducted for the first time to characterize their climatological features. We found that the SED distribution exhibits a higher peak intensity and a narrower width as geomagnetic activity strengthens. The peak intensity of SED has maximum values around the equinoxes in their seasonal distribution. Additionally, we observed a solar cycle dependence in the SED distribution, with more events occurring during the solar maximum and declining phases compared to the solar minimum. SED plumes exhibit a sub‐corotation feature with respect to the Earth, characterized by a westward drift speed between 50 and 400 m/s and a duration of 3–10 hr. These information advanced the current understanding of the spatial‐temporal variation of SED characteristics.


Introduction
The storm-enhanced density (SED), first coined by Foster (1993), is a large-scale midlatitude ionospheric total electron content (TEC) and/or electron density (Ne) enhancement observed in the local afternoon and dusk sectors during early stages of a geomagnetic storm.SED is usually characterized by a distinct "plume" structure at midand subauroral latitudes, which denotes a spatially-narrow channel of enhanced plasma density streaming sunward and poleward from a high-density base region at the equatorward edge of the main ionospheric trough to the noon cusp region (e.g., Kelley et al., 2004;Foster et al., 2021;Coster et al., 2007;S.-R. Zhang & Aa, 2021).SED is observed to be associated with the elevated F-region peak altitude and scale height, low electron temperature, and upward supply of cold O+ rich plasma in the topside region (Foster et al., 2005;Walsh et al., 2014;Zou et al., 2014).The plasma within the SED plume can occasionally transport through the cusp into the polar cap region via large convection flow, forming the tongue of ionization (TOI) and/or polar cap patches (Foster et al., 2005; Q.-H.Zhang et al., 2013).The sharp density gradient of the SED plume, along with irregularities at its boundaries, can be the source of severe radio scintillation and degradation in navigation performance, potentially leading to severe space weather effects at midlatitudes.Therefore, the morphology and evolution of SED have been comprehensively investigated utilizing observations from both ground-based and space-based sources, such as incoherent scatter radar data (e.g., Aa et al., 2021;Foster et al., 2005;Yizengaw et al., 2006;Zou et al., 2013Zou et al., , 2014;;S.-R. Zhang et al., 2017), TEC data from Global Navigation Satellite Systems (GNSS) observations (e.g., Coster & Skone, 2009;David et al., 2016;Thomas et al., 2013;B. Li et al., 2022;Liu, Wang, Burns, Yue, et al., 2016;Sori et al., 2019), satellite measurements through in-situ or remote sensing (e.g., Horvath & Lovell, 2015;Kitanoya et al., 2011;Lin et al., 2022;K. Zhang et al., 2021), as well as numerical simulations and/ or data assimilation (e.g., Aa et al., 2022;Aa et al., 2023;Dang et al., 2019;David et al., 2011;Gardner et al., 2018;Huba & Sazykin, 2014;Liu, Wang, Burns, Solomon, et al., 2016;Lu et al., 2020;Sojka et al., 2012;Z. Wang et al., 2019;Yue et al., 2016;Zou et al., 2017;Zhai et al., 2020).simultaneous satellite measurements, the SED signature has evolved to encompass a broader array of the Geospace plume phenomenon and a variety of magnetosphere-ionosphere coupling processes, likely associated with the drainage plumes and erosion of the plasmasphere (Foster et al., 2002;Moldwin et al., 2016).The generation mechanism of SED is still under wide debate with various interpretations, such as: (a) Storm-time penetration and/or expansion of high-latitude convection electric fields.This increases upward and poleward E × B drift in the dayside midlatitude ionosphere, lifting up plasma to a higher altitude with a lower recombination rate that contributes to the density enhancement in the SED base region (Aa, Zhang, Zou, et al., 2024;Heelis et al., 2009;Huang et al., 2005;Liu, Wang, Burns, Solomon, et al., 2016;Zou et al., 2014).The equatorward part of the enlarged convection cell continually encounters these enhanced densities and transports them toward higher latitudes at noon, producing the observed latitudinally-narrow SED plume, known as the "snowplow" effect (Foster, 1993).(b) Flow stagnation mechanism.In the local afternoon sector, the westward convection flows are in the opposite direction of the corotation flows, leading to the formation of a zonal flow stagnation region that increases the plasma residence time in sunlight at a specific local time sector within the inertial reference frame.When this zonal plasma stagnation occurs in the presence of enhanced upward and poleward flows, under a favorable condition of aforementioned expanded storm-time convection, the electron density therein can dramatically increase through the combination of prolonged photo-production and the reduced ion recombination rate (Fuller-Rowell, 2011;Heelis et al., 2009).(c) Zonal flux transport driven by subauroral polarization stream (SAPS).SAPS is an intense sunward plasma flow channel located in the duskside subauroral ionosphere equatorward of the auroral precipitation zone (Aa et al., 2020;Foster & Burke, 2002;Erickson et al., 2011;H. Wang et al., 2008).The horizontal advection of SAPS could transport vast plasma sunward from later local times to the dayside, thereby contributing to higher densities in the SED plume (Aa et al., 2023;Foster et al., 2007;Vlasov et al., 2003).However, some theoretical studies suggest that SAPS-induced sunward plasma transportation is essential to amplify the SED plume in the noon sector, while the SED plume in the dusk sector experiences a reduction due to increased frictional heating and plasma loss rate because fast SAPS flow enhances plasma temperatures (S.Li et al., 2023;Liu, Wang, Burns, Solomon, et al., 2016).(d) Equatorward neutral wind and thermospheric composition effect.The storm-time equatorward neutral wind surge tends to raise the plasma along the geomagnetic field lines to a higher altitude with lower recombination rates, thus contributing to positive ionospheric storm with enhanced electron densities at mid and low-latitudes (but not limited to the SED region) (Anderson, 1976;Balan et al., 2010;Mendillo, 2006).Moreover, the thermospheric downwelling of neutral species through constant pressure levels at mid-and low-latitudes can lead to the O/N 2 increment, thereby enhancing the column electron density within and beyond the SED region (Immel et al., 2001;W. Wang et al., 2012).(e) Meridional plasma transportation associated with the poleward expansion of equatorial ionization anomaly (EIA).Some studies suggest that the super fountain effect during an intense storm can push the EIA crest to a higher magnetic latitude of 25-40°, thus strong ambipolar diffusion with poleward transport of equatorial plasma to higher latitudes may play a role to enhance electron densities at the SED base (Kelley et al., 2004;Mannucci et al., 2005;Tsurutani et al., 2004).Nevertheless, Rishbeth et al. (2010) suggest that the local production associated with plasma uplift should be considered more significant for contributing to higher plasma densities in the SED, rather than relying on remote latitudinal transportation.Taken together, it is worth noting that the formation of the SED is likely to involve more than just a single mechanism, and their relative importance varies with respect to location, time, and storm magnitudes.
While prior studies have made significant progress, our current understanding of the SED characteristics based on individual case studies remains incomplete.Specifically, certain important issues need to be further addressed, as outlined below.Rideout, 2005;Gardner et al., 2018;Lin et al., 2022;Mannucci et al., 2005), the St. Patrick's day storms on 17-18 March in 2013 and 2015 (e.g., Heine et al., 2017;Liu, Wang, Burns, Solomon, et al., 2016;S.-R. Zhang et al., 2017;Zhai et al., 2020), as well as intense storms on 26-27 February 2023 and 23-24 April 2023 (Aa et al., 2023).However, there is still a lack of a comprehensively organized SED list that clearly shows which storms have exhibited the SED phenomenon, let alone a statistical analysis of SED data set.For instance, what is the solar activity dependence of SED, and what does the seasonal distribution of SED look like?A comprehensive list and a statistical analysis of SED data set is essential not only for providing significant information regarding the climatological knowledge of SED, but also can offer new physical insights into the complicated dynamical processes that produce SED. 2. In-depth analysis of SED features over key regions of focus, such as North America: SED events have been observed more frequently in the North American sector than in other longitudes.The prevalence of SED events in North America can be attributed to the "close-to-dipole" feature of the geomagnetic configuration in this longitudinal sector, where the geomagnetic latitudes are ∼10°higher than corresponding geographic latitudes.This factor renders the North American sector more susceptible to SED development, owing to the relatively lower geographic latitude of magnetically subauroral regions.These regions exhibit smaller associated solar zenith angles, resulting in higher photo-ionization reservoirs compared to other longitudinal sectors.This creates a favorable condition for SED development during the stagnation and storm-time convection expansion when the North American longitude region containing the magnetic pole is tilted toward the dayside (Aa et al., 2023;Heelis et al., 2009).Moreover, the observed prevalence of SED events in North America can also be ascribed to the availability of large database of observations in this region, in particular dense networks of ground-based GNSS receivers and observations from the altitude-resolved wide field incoherent scatter radar at Millstone Hill.Given the severe space weather effects of SED-induced density gradients and the ongoing advancements of utilizing GNSS observations for ionospheric diagnostic, it is important to take advantage of large GNSS TEC database over the North American longitude sectors to conduct in-depth analysis of SED features, such as the SED location and intensity variation.
Therefore, in the current study, we have derived a comprehensive list of 49 SED events over the North American sector and have conducted a first-time statistical analysis of the SED data set.The SED events over the North American sector are carefully identified through the examination of all intense geomagnetic storms (Minimum Sym-H index < 100 nT) in the past two solar cycles.The ground-based GNSS TEC data and a newly-derived TEC-based ionospheric data assimilation system (TIDAS, Aa et al. (2022Aa et al. ( , 2023)) is used to analyze the SED characteristics.We aim to partially address the aforementioned issues and to advance current understanding of the spatial and temporal variation of SED characteristics.

Instrument and Data Sets
Ground-based GNSS TEC data are routinely processed at the Massachusetts Institute of Technology's Haystack Observatory, utilizing data from extensive networks of 5,000+ GNSS receivers worldwide, with more than 2,000 receivers covering the North American sector.The binned TEC products, with a spatial-temporal resolution of 1°( longitude) × 1°(longitude) × 5 min, are distributed to the community through the Madrigal data system (Rideout & Coster, 2006;Vierinen et al., 2016).In this study, we primarily utilize TEC data over the continental US and adjacent regions to examine and analyze SED features therein.TIDAS is a new TEC-based regional ionospheric data assimilation system, which uses a hybrid Ensemble-Variational technique to assimilate multi-instrument ionospheric TEC/Ne measurements into the NeQuick model, including the line-of-sight GNSS TEC data, COSMIC-I/II radio occultation data, altimeter TEC measurements from Jason satellite above the ocean, as well as Millstone Hill incoherent scatter radar data (Aa et al., 2022).TIDAS can produce reanalyzed regional 3-D time-evolving maps of ionospheric electron density with a spatial-temporal resolution of 1°× 1°in latitude and longitude, 20 km in altitude, and a time cadence of 5 min.The root mean square error of TIDAS is about 0.5-2 TECU, depending on geomagnetic condition.TIDAS was initially developed for the continental US and adjacent regions (TIDAS-US, Aa et al. ( 2022)) with the primary goal of studying midlatitude ionospheric electron density gradients, particularly focusing on the 3-D morphology and evolution of SED.Recently, TIDAS has undergone upgrades to encompass the European sector (TIDAS-EU, Aa et al. (2023)) and the South American sector (TIDAS-SA, Aa, Zhang, Erickson, et al. (2024)).For a more detailed description of TIDAS algorithm, readers may refer to Aa et al. (2022).In this study, we employ the F2-layer peak density (NmF2) obtained from TIDAS outputs to aid in the analysis of SED features.For example, Figure 1d-1f show the reanalyzed NmF2 and hmF2 (F2-layer peak height) distribution for aforementioned three intense storms, respectively.The SED plume manifest as a ridge-like high-density structure, exhibiting considerable NmF2 enhancement and layer uplift compared to poleward/equatorward parts.

Methodology
We have examined more than 80 intense geomagnetic storms between 2000 and 2023 whose minimum longitudinally symmetric (Sym-H) index reached below 100 nT.For each storm event, we created animations of TEC maps over the continental US and adjacent regions throughout the storm day, visually examining the spatialtemporal evolution of the ionospheric TEC during the storm.If the storm exhibited a clear structure of SED plume, as those demonstrated in Figure 1a-1c, it was identified and labeled as containing SED.Out of these intense storms, we have identified 49 storms in which a clear SED plume was visible.For each of these 49 storms, we have approximately estimated the time when the SED plume begins to form and the time at which the plume dissipated at the end of the SED.Table 1 shows the compiled list of storms containing the SED and the corresponding start and ending times.It is worth noting that SED are not exclusive to the North American sector but can also occur in other longitudinal sectors, such as the European region (Coster et al., 2007).Additionally, SED may be occasionally observed during certain moderate geomagnetic storms, even if the Sym-H index did not reach 100 nT (Zou et al., 2014).Hence, in this study, our emphasis is on characterizing SED features over the continental US specifically during these strong geomagnetic storms.
To quantitatively characterize the SED statistics, it is essential to parameterize each SED event in this list.As an illustration, Figure 2a presents a TEC map over the continental US at 21:15 UT during the St. Patrick's Day storm on 17 March 2015, which reveals a distinctive SED plume structure at midlatitudes that streams poleward/ westward.The scattered points in Figure 2b shows a slice of TEC that cut through the SED plume along the 45°N latitude, which clearly demonstrates that the high-density SED feature in the 1-D curve can be approximated by a Gaussian-like function.Thus, we use the following Gaussian function to fit the zonal TEC slice within the SED: where h is the elevation of the Gaussian that represents the intensity of the SED plume, φ is longitude in degrees, φ 0 is the central longitude of the SED plume at that particular latitude, and σ is the half-width of the Gaussian or SED plume.The polynomial coefficients b, c, and d are used to describe the parabolic-like variation of the background TEC.For example, the red curve in Figure 2b represents the corresponding Gaussian-fitting results  Journal of Geophysical Research: Space Physics 10.1029/2024JA032750 using Equation 1, with the relevant key parameters (h, φ 0 , and σ) being calculated and annotated.As evident, this Gaussian fitting effectively reproduces the zonal variation of the TEC slice within the SED.Hence, the SED plume at this latitude can be appropriately parameterized by a central longitude of 94.4°, a half-width of 3.5°, and an intensity of 11.57TEC unit.We can also calculate the relative intensity of the SED plume with respect to the background (i.e., h/[b + cφ 0 +dφ 2 0 ]).Likewise, Figures 2c and 2d depict the TIDAS-reanalyzed NmF2 map containing the SED plume and the corresponding Gaussian-fitting curve with key parameters being annotated, which further underscores the efficacy of this technique.The effectiveness of utilizing such a Gaussian-fitting method to parameterize the SED plume has been validated through tests conducted on various SED events at different latitudes.Thus, utilizing this technique, for every 5-min map corresponding to each storm event featuring the SED, we extracted a TEC/NmF2 slice at each latitude where the SED was present and performed the Gaussian fitting to calculate those aforementioned key parameters.The characteristics of these SED parameters will be collectively examined in detail in the next section.

Results and Discussion
We first take the super storm on 20 November 2003 as an example, which was the strongest storm in Solar Cycle 23.Figures 3a-3d show the temporal variation of SED relative intensity, absolute intensity, central longitude, and half width from 17 to 02 UT.Each data point and associated error bar in the plots represent the mean value and standard deviation between all of the latitudes over which the SED plume spanned at that particular time.The SED intensity, as shown in Figures 3a and 3b, exhibited a gradual increase during the storm progression starting from 17 UT and peaked around 20:30 UT.The maximum absolute intensity reached approximately 110 TEC units, corresponding to a relative intensity 5 to 6 times higher than the background values.After 21 UT, the intensity rapidly decreased and remained at 20-30 TEC units between 23 and 01 UT.Moreover, Figure 3c shows that the central longitude of SED tends to move westward over the course of the event.The half-width of SED plume, as shown in Figure 3d, varies between 5 and 15°with intermittent fluctuation.These parameters quantitatively demonstrate that SED is a dynamic large-scale ionospheric density gradient structures.
Likewise, Figures 3e-3l show the temporal variation of SED parameters during the 17 March 2015 and 23 April 2023 storms, respectively.In a similar pattern, the absolute and relative intensity tend to increase during the main phase of the storm until they reached a peak at around 19-22 UT (13-16 LT).It is suggested that the SED enhancement should be most significant when the longitudes encompassing the magnetic pole is tilted toward the afternoon side.This configuration creates favorable conditions with magnetically subauroral regions situated at both relatively lower geographic latitude and under smaller solar zenith angles, collectively resulting in a higher photo-ionization reservoir for electron density enhancement (Aa et al., 2023;Heelis et al., 2009).This corresponds to the North American longitude sector around 19-22 UT (14-19 LT) in the northern hemisphere, which is proved by our SED statistics.Previous statistical studies on TOI and/or polar cap patches have also shown that patch occurrence is higher in later UT interval in Northern Hemisphere (David et al., 2016;Kagawa et al., 2021), which tends to peak during 14-19 MLT in North American sector (Ren et al., 2018).This preference for UT is consistent with our SED statistics, given that the intrusion of subauroral plasma into the polar cap through SED/ TOI serves as the primary source of polar cap patches.Moreover, the central longitude of SED exhibits a westward-moving tendency.Specifically, the averaged westward-moving speed of SED plume was estimated to be ∼78  To investigate the longer-term trends of the SED morphology and how they correlate with other factors, Figures 5a-5c show the comparison of the peak intensity of all SED events with respect to the geomagnetic activity (minimum Sym-H index), time of year, and the solar activity (F10.7 index), respectively.Figure 5a shows that the SED generally tends to have a higher peak intensity as the Sym-H index decreases.Specifically, those three SED events with the lowest minimum Sym-H index around 400 nT, which correspond to the October-November 2003 superstorms, have a significantly strong peak intensity that is 2-3 times higher than the other SEDs.Moreover, Figure 5b illustrates that the seasonal distribution of SED intensity tend to peak around the equinoxes.This phenomenon could potentially be linked to the semiannual variation of the ionospheric electron density climatology at midlatitudes that peaks in March and October (S.-R.Zhang et al., 2005;S.-R. Zhang et al., 2010).Given that the intrusion of subauroral plasma into the polar cap through SED/TOI serves as the primary source of polar cap patches, previous statistical studies on polar cap patches have also indicated that the occurrence of patches reaches its peak during equinoxes, reflecting the seasonal asymmetry in electron density in the global ionosphere (Coley & Heelis, 1998;Noja et al., 2013).Moreover, this seasonal dependence of SED could be related to the semiannual variation of the geomagnetic activity that generally attributed to the Russell-McPherron effect, which suggests that equinoctial periods are characterized by increased geomagnetic activity due to higher-than-average southward components of the interplanetary magnetic field (IMF) (Cliver et al., 2000;Russell & McPherron, 1973).Furthermore, Figure 5c shows that more SED events were observed during the solar maximum and declining phases compared to the solar minimum.Such a solar activity dependence could be partially attributed to the tendency for higher background source density and the heightened geomagnetic activity during these periods.Specifically, coronal mass ejection events tend to occur more frequently during years of high solar activity, while recurrent geomagnetic activity linked to high-speed solar wind streams is more prevalent during the declining phase of the solar cycle.In contrast, few SED events during solar minimum is potentially linked to the reduced occurrence of storm events during solar minimum, given the focus on storm periods in our research.
Similarly, Figure 6 illustrates the distribution of the mean half-width of all SED events with respect to the geomagnetic activity, time of year, and the solar activity, respectively.Contrary to the behavior of the peak intensity, Figure 6a shows that the SED tends to have a smaller width as the geomagnetic activity intensifies.In particular, SED events during those superstorms with the lowest minimum Sym-H index around 400 nT exhibit a much narrower half-width (∼7°) that were 20%-40% smaller than those less-intense storms, where the average half-width in longitude ranges from 9°to 11°.The correlation observed between Figures 5a and 6a indicates that as geomagnetic activity strengthens, the increased penetration and expansion of convection electric field result in a stronger SED plume with higher peak intensity.Consequently, the electron density gradients within the SED plume become steeper, corresponding to a narrowing of the SED channel and a reduction of the plume width.Additionally, Figures 6b and 6c show that there are no significant seasonal or solar cycle dependence of the SED width.
Figure 7 presents SED statistics based on plume duration and horizontal drift speed.Figures 7a and 7b show the histogram distribution of SED plume duration and the scatter-point distribution of storm intensity versus SED plume duration, respectively.It can be observed that the majority of SED plumes have duration between 3 and 10 hr over a wide range of storm intensities.The extended duration suggests that the SED plume is sub-corotating with respect to the Earth.Moreover, SED is actively driven in an electrodynamic sense, with a generally longer duration as geomagnetic activity strengthens.Figure 7c displays the distribution of SED plume horizontal drift speed.The result clearly demonstrates that SED plumes indeed exhibit sub-corotation with respect to the Earth, characterized by a westward drift speed ranging from 50 to 400 m/s (with the majority falling between 100 and 200 m/s).Additionally, Figure 7d shows scatter-point distribution of storm intensity versus SED plume drift speed, demonstrating that SED plumes tend to exhibit smaller westward drifting speeds (longer duration) as geomagnetic activity strengthens.This phenomenon can be attributed to the flow stagnation mechanism due to the enhanced counterbalancing effect between the westward convection flows, which strengthen as storm intensity increases, and the eastward corotation flows in the local afternoon sector.This creates a quasi-stagnation area in the midlatitude ionosphere with slower zonal flow and longer plasma residence time in local sunlight at a specific local time sector, leading to longer SED duration through the combination of prolonged photo-production and the reduced ion recombination rate (Aa et al., 2023;Heelis et al., 2009).

Summary
This study has compiled a comprehensive list of 49 SED events over the continental US by examining GNSS TEC and TIDAS NmF2 data for intense geomagnetic storms occurring between 2000 and 2023, during which the minimum Sym-H index dropped below 100 nT.For the first time, we have quantified the SED statistics for each identified event and have conducted a statistical analysis to characterize the climatological features of SED.The main results and findings are summarized as follows: 1.We have parameterized the spatial-temporal evolution of SED morphology through using a Gaussian function to quantify the essential characteristics of the SED plume, including the plume intensity, central longitude, and half-width.In those SED events during strongest storms, the intensity tends to increase as the storm progresses, reaching peak values around 19-22 UT before gradually decreasing.The central longitude tends to shift westward over the course of the event, and the SED plume appears to subcororate with respect to the Earth.The half-width of SED plume fluctuates between 5 and 15°throughout the lifespan of the SED. 2. For the first time, we have conducted a statistical analysis of SED events to characterize their climatological features.As geomagnetic activity strengthens, the distribution of SEDs shows a higher peak intensity and a narrower half-width.The seasonal distribution of SED intensity peaks around the equinoxes.The SED distribution has a solar cycle dependence, with more events being observed during the solar maximum and declining phases compared to the solar minimum.3. The statistical results demonstrate that SED plumes exhibit a sub-corotation feature with respect to the Earth, characterized by a westward drift speed between 50 and 400 m/s and a duration of 3-10 hr over North America.
To the best of our knowledge, the sub-corotating characteristics of the SED plume and its drifting speed have seldom been analyzed in prior case studies with merely vague descriptions.In contrast, our statistical analysis convincingly proves and demonstrates both the qualitative and quantitative aspects of this key feature of the SED plume.
1.A comprehensive SED list and a statistical analysis: SED mechanisms have been extensively investigated based on individual case studies during some strong geomagnetic storms in Solar Cycles 23-25, such as the Halloween storm on 29-30 October 2003 and the super storm on 20-21 November 2003 (e.g., Foster &

Figure 1 .
Figure 1.(a-c) 2-D global navigation satellite system (GNSS) total electron content maps showing three examples of storm-enhanced density events over the continental US and adjacent regions on 20 November 2003, 17 March 2015, and 23 April 2023, respectively (d-f) The corresponding ionospheric hmF2 and NmF2 distributions given by TEC-based ionospheric data assimilation system data assimilation.

Figure 2 .
Figure 2. (a) A total electron content (TEC) map displaying the storm-enhanced density at 21:15 UT during the St. Patrick's Day storm on 17 March 2015.The 45°N latitude is indicated by a red line.(b) The longitudinal variation of the TEC slice at 45°N (blue dots), along with the Gaussian fitted curve (red) with relevant key parameters being marked.Panels (c and d) are the same as (a and b), but for the NmF2 results obtained from TEC-based ionospheric data assimilation system data assimilation.
m/s for the 20 November 2003 storm, ∼175 m/s for the 17 March 2015 storm, and ∼146 m/s for the 23 April 2023 storm, respectively.These values indicate that the SED plume appears to subcororate with respect to the Earth, consistent with the simulation results given by Lu et al. (2020).Furthermore, although the half-width (in longitude) does not consistently show an increasing or decreasing pattern, it tends to fluctuate within the range of 5-15°in longitude throughout the lifespan of the plume.Similar trends are also observed in other identified SED events.Moreover, Figure 4 shows the comparison between SED plume intensity in GNSS TEC data and NmF2 results from TIDAS data assimilation during the 20 November 2003, 17 March 2015, and 23 April 2023 geomagnetic storms, respectively.As observed, the SED intensity in TEC and NmF2 exhibits a generally similar variations.They both tend to increase and decrease at approximately the same pace and with comparable slopes, reaching peaks around the same time.They peak around 20:30 UT on 20 November 2003, 21 UT on 15 March 2015, and around 21:30 UT on 23 April 2023.This suggests a strong connection between the statistical characteristics of TEC values and NmF2 values in the SED plume during geomagnetic storms.

Figure 4 .
Figure 4. Comparison between storm-enhanced density plume intensity in (a-c) GNSS total electron content data and (d-f) TEC-based ionospheric data assimilation system NmF2 results during three strong geomagnetic storms.

Figure 5 .
Figure 5. Statistical characteristics of storm-enhanced density plume intensity from all events: (a) Geomagnetic activity dependence with respect to Sym-H index.(b) Seasonal distribution.(c) Solar activity dependence with respect to F10.7 index.

Figure 6 .
Figure 6.The same as Figure 5, but for the half-width of storm-enhanced density plume.

Figure 7 .
Figure 7. Storm-enhanced density (SED) statistics by plume duration and horizontal drift speed: (a) Histogram distribution of SED plume duration, (b) scatter point distribution of storm intensity versus SED plume duration (c and d) are the same as (a and b), but for SED plume drift speed.

Table 1
Intense Geomagnetic Storm List Containing the Storm-Enhanced Density Over the Continental US