Uncovering the Drivers of Responsive Ionospheric Dynamics to Severe Space Weather Conditions: A Coordinated Multi‐Instrumental Approach

Space‐weather conditions can often have a detrimental impact on satellite communications and limited experimental data has made it challenging to understand the complex processes that occur in the upper atmosphere. To overcome this challenge, we utilized a coordinated multi‐instrumental dataset consisting of GNSS airglow remote sensing, ionosonde, magnetometer, and in‐situ satellite data to investigate plasma depletions. We present a case study focused on the geomagnetic storm that occurred on 27 February 2014. During the storm, GNSS positioning errors exceeded undisturbed levels by at least 2 times, and ionospheric corrections reached amplitudes of up to ±20 m at the Rabat station. We identified 3 large depletions that were most likely generated by sudden vertical ionospheric drifts that began at approximately 17:00 UTC at sunset in Morocco and the southern regions of Spain. These drifts reached ∼500 m/s and lasted until 22:00 UTC. The observed depletions propagated to the northeast, as seen through ionosonde echoes and ground‐based airglow images. Satellite limb‐images revealed an ionospheric uplift of about 100 km due to the storm, consistent with ionosondes in Spain. The observed local anomalies may be influenced by variations in equatorial electric current flows, which are correlated with fluctuations in ground‐based magnetometer data. These variations are likely a result of the effects of the inner radiation belt on the development of plasma bubbles in the African longitude sector. Sudden enhancements in upward E × B drift caused ionospheric uplift to higher altitudes, enhancing the “fountain effect” and shifting the Equatorial Ionospheric Anomaly crests to higher latitudes.


Introduction
The ionosphere is a region in the upper atmosphere extending from approximately 50 km to 2,000 km above the Earth's surface that is formed mainly through photoionization by solar radiation.Irregular variations in the ionosphere can significantly impact radio communications satellite operation and navigation, Positioning-Navigation-Timing (PNT), Global Navigation Satellite Systems (GNSS), Earth's Remote Sensing, and numerous other applications that rely on the propagation of radio waves.Equatorial Plasma Bubbles (EPBs) refer to regions of depleted plasma with respect to the background ionosphere, and are the main form of ionospheric irregularities in the equatorial F-region (Kelley, 1989).EPBs are transitory events whose dynamics are not yet fully understood, and existing models struggle to accurately represent the actual variations that occur in practical applications like GNSS and PNT.
EPBs can have a significant negative impact on radio-based technologies, like communications, navigation, and other space weather applications.EPBs can irregularly scatter and refract irregularly the radio wave signals traversing through them, causing rapid changes in signal amplitude and phase known as scintillations (Hargreaves, 1992), resulting in anomalous fading or distortion of electromagnetic signals.Scintillation can occur when the ionosphere is disturbed by solar and geomagnetic activity, particularly during geomagnetic storms, when the Earth's magnetic field is distorted by highly variable solar activity.Recently, several studies have provided valuable insights into the behavior of EPBs under different conditions and their interactions with various geophysical factors.The research conducted by Wu et al. (2021) focused on the unusual evolution of EPBs during a geomagnetically quiet night.They found that variations in ionospheric plasma vertical drift and zonal wind were key factors in this unusual evolution.Santos et al. (2016) investigated the zonal drift reversal of EPBs during storm time over the Brazilian region.Their findings highlighted the role of disturbance Hall electric fields in causing this reversal.Smith and Heelis (2017) studied the variations in the occurrence and spatial scale of EPBs in relation to local time, longitude, season, and solar activity.Their research revealed that EPBs occur late in local time, primarily after midnight in all longitude sectors during solar minimum conditions.
It is well known that the Sun's high-energy ultraviolet and X-ray radiation ionizes neutral molecules in the air, creating an electrically conductive atmosphere, particularly during midday hours.Additionally, the Sun's light heats up the atmosphere, which generates winds that power an ionospheric dynamo, creating electric currents.The Equatorial Electro-Jet (EEJ), an eastward electric field that appears along the magnetic equator at an altitude of approximately 100 km from dawn to dusk, results from this electric current flow (Chapman, 1951;Richmond, 1973).After the plasma became lifted up across the horizontal magnetic field lines over the magnetic dip equator by the vertical E × B plasma drift, gravitational forces parallel to Earth's magnetic field, along with plasma pressure gradients, cause plasma to move at approximately ±15-20°dip latitude.This process results in the formation of the Equatorial Ionospheric Anomaly (EIA).The EIA is characterized not only by a trough region over the dip equator (Duncan, 1960), but also by the two crests located at approximately ±15-20°dip latitude.The exact position of these crests can vary depending on magnetic activity.During geomagnetic storms, an equatorial eastward prompt penetration electric field (PPEF) subsequently enhances the vertical E × B plasma drift, creating strong variations in the system.While the Pre-Reversal Enhancement (PRE) is an everyday phenomenon, triggered by sunset conditions, it becomes particularly amplified during storm-time variations.This amplification is especially noticeable at the sunset equator, where the PRE deepens at the EIA trough and generates localized plasma depletions in low-latitude regions (Rishbeth, 1971).The eastward PPEF during storms further enhances this effect.
Despite the advancements in ionospheric modeling (Radicella, 2009;Bilitza et al., 2022;Qian et al., 2014;Huba et al., 2000), existing models used in practical applications are unable to represent small structures such as plasma depletions that may only span a few hundred kilometers (see Figure S1 in Supporting Information S1).However, modern measurement techniques are capable of sensing and estimating the dynamics of small ionospheric features (Martinis et al., 2018), such as ground-based all-sky cameras that measure electromagnetic radiation in the ultraviolet, visible, or infrared range.Other techniques include ionosonde (Gilli et al., 2018;Jerez et al., 2020), GNSS receivers (Socola & Rodrigues, 2022;Yu & Liu, 2021), radio telescopes (Mangla & Datta, 2023), and space-based sensors, such as Global-scale Observations of Limb and Disk (GOLD) of National Aeronautics and Space Administration (NASA) (e.g., Eastes et al. (2020)) that can measure EPBs.Ionospheric irregularities can manifest as spread F in ionograms, EPBs in radar maps, and traveling ionospheric disturbances (TIDs) in optical images and Total Electron Content (TEC) maps (Bowman, 1991;Kelley & Fukao, 1991;Ding et al., 2011).These irregularities can occur with spatial scales ranging from meters to several thousands of kilometers.Numerous studies have demonstrated that various instability mechanisms are responsible for the generation of plasma irregularities.The global climatology of ionospheric irregularities is primarily influenced by solar activity and magnetic field topology (Liu, Hernández-Pajares, et al., 2021;Liu, Zhou, et al., 2021).
Given the importance of understanding the impact of the ionosphere on countless applications, this work aims to contribute to a better understanding of space weather coupling phenomena in the upper atmosphere through the analysis and characterization of localized plasma depletions.We utilize data from a range of sources, including GNSS, airglow remote sensing, ionosonde, magnetometers, and in-situ satellite data, to investigate the possible drivers of responsive ionospheric dynamics to the geomagnetic storm of 27 February 2014.In particular, this work focuses on the following aspects: • Investigating ionospheric variability through multi-instrument data and analyzing the localized processes that result from the interaction between the solar wind-magnetosphere and the ionosphere, specifically the characterization of localized plasma depletions for this particular case.• Analyzing the potential of advanced sensing instruments to map the ionospheric disturbances during geomagnetic storms, with particular interest in EPBs.
• Characterizing the behavior of radio signals as they pass through the ionosphere and detecting anomalies is crucial to understanding the patterns and temporal variations associated with these EPBs and their contributions to practical applications.
In the following section, we introduce the data and methods used in our study.We then present the results of our analyses in Section 3 and discuss the implications of our findings in Section 4. In the final section, we will provide our conclusions and present hypotheses based on our experimental results.

Data, Products, and Models for the Analysis
To carry forward this investigation, we use a variety of data, products, models, and indices related to the Sun-Earth connection.Particularly, these data include solar wind (SW), Interplanetary Magnetic Field (IMF) components, Interplanetary Electric Field (IEF) zonal East-West component (Ey), and geomagnetic activity indices such as the amplitude planetary index (Am) of geomagnetic variation, the Northern Polar Cap index (PCN), the Auroral Electrojet (AE) index, and the symmetric disturbance in the horizontal intensity of the magnetic field vector (SYM-H, ASY-D, and ASY-H).
The solar wind parameters and PCN index are obtained from the NASA OMNI website (http://omniweb.gsfc.nasa.gov/), the Am index (Mayaud et al., 2023) is obtained from the EOST (École and Observatoire des Sciences de la Terre) (https://eost.unistra.fr/en/eost/eost),and the AE (Davis & Sugiura, 1966) and SYM-H (Zhao et al., 2021) indices are obtained from the World Data Center Kyoto (https://wdc.kugi.kyoto-u.ac.jp/).These websites provide information on the derivation and meaning of the different data and indices.For instance, while the SYM-H index measures the ring current intensity, the Am index is susceptible to any geophysical current system, including magnetopause currents, field-aligned currents, or auroral electrojet.The ring current is mainly generated by pressure gradient and magnetic curvature drift.The IEF East-West component (Ey = Vx × Bz; where Vx is the x component of the solar wind velocity and Bz is the z component of the IMF) moves the plasma in the magnetosphere from the tail to Earth, and helps to provide the seed population for the ring current.The Polar Cap index measures the transpolar convection of magnetospheric plasma and embedded magnetic fields driven by the interaction with the solar wind, and the merging electric field (Em) assumes that there is an equal magnitude of the electric field in the solar wind, the magnetosheath, and on the magnetospheric sides of the magnetopause (Kan & Lee, 1979): In this equation, By and Bz are the IMF components, V SW is the solar wind speed, and θ is the IMF clock angle in geocentric solar magnetospheric (GSM) coordinates.
We use predictive ionospheric models to investigate abrupt ionospheric anomalies during geomagnetic storms, the empirical model developed at Calabia andJin (2019, 2020), which is based on a lower-dimensional reduction of vertical Total Electron Content (vTEC) data from 2003 to 2018, and the International Reference Ionosphere (IRI-2020) model (Bilitza et al., 2022), which specifies monthly averages of electron density, ion composition, electron temperature, and ion temperature in the altitude range of 50-2,000 km.The model of Calabia and Jin provides estimates of vertical Total Electron Content (vTEC) for any specified epoch with a spatial resolution of 2.5°in geographical latitude and 5°in geographical longitude.Furthermore, it provides the flexibility for users to adjust parameters related to solar and magnetospheric forcing, as well as annual and Local Solar Time (LST) cycles.For our experiment, we set the model to a constant quiet geomagnetic contribution (Am = 6) and the results are compared with the IRI-2020.It's important to note that some attenuation may occur due to missing contributions from the full plasmasphere (Jin et al., 2021).
We also use the post-processed UQRG Global Ionospheric Maps (GIMs) of vertical Total Electron Content (vTEC) provided by the Universitat Politècnica de Catalunya (UPC) (Hernández-Pajares et al., 2009;Liu, Hernández-Pajares, et al., 2021;Liu, Zhou, et al., 2021).UQRG GIMs-vTEC are provided with a latitude range of 87.5°S to 87.5°N in steps of 2.5°, with a longitude range of 180°W to 180°E in steps of 5°, and with a temporal resolution of 15 min; the highest accuracy in high-time-resolution (Wielgosz et al., 2021).The GIMs of vTEC from the International GNSS Service (IGS) network (Hernández-Pajares et al., 1999)  We also analyze the observations of airglow emissions, which are crucial for monitoring ionospheric structures and irregularities, using all-sky wide-angle Visible Imaging Spectrometers (VIS).As it is known, the OI 630.0 nm nightglow emission can be used as a tool to study plasma structures and dynamics in the ionosphere.EPBs can be visualized as a dark band region due to a decrease in the OI 630 nm emission intensity, which indicates the depletion of electron density.However, plasma blobs or enhanced plasma density regions appear as quasi-ovalbright regions in the OI 630.0 nm emission (Adebayo et al., 2023).The images used in this study were obtained from Malki et al. (2018).These images were captured at approximately 15-minute intervals, starting from 21:00 Universal Time Coordinated (UTC) on 27 February 2014, and ending at 00:50 UTC on 28 February 2014.On the other hand, the Special Sensor Ultraviolet Spectrographic Imager (SSUSI) instrument onboard the Defense Meteorological Satellite Program (DMSP) F18 spacecraft (Paxton et al., 2002) is also employed to obtain imaging data.We use the 135.6 nm wavelength, the ionized nitrogen line, whose resulting nightglow intensities can be related to the square of the ionospheric electron density (Tinsley & Bittencourt, 1975;Meier, 1991).
To estimate the ionospheric maximum related to the ionospheric F2-layer electron density peak (NmF2), the SSUSI nighttime non-auroral F-region algorithm is used to convert airglow measurements from nighttime latitudes outside the aurora region.The data can be downloaded from Paxton et al. (2002), and the image strips are made along orbits.The rotation of the Earth samples one strip from each orbit leg, and each strip is temporally separated by approximately 90 min.The use of NmF2 is convenient because it can estimate vTEC using the ionospheric slab thickness ratio (Davies, 1990;Pignalberi et al., 2022).When estimating TEC from NmF2, slab thickness is subjected to diurnal, annual, latitudinal, and storm-time influences (Davies & Liu, 1991;Stankov & Warnant, 2009), and these must be considered.
We use also data recorded by ionosondes which allow to detect ionospheric plasma bubbles by monitoring the well-known range and frequency spread F phenomena in the ionograms.For this work, we use the DPS4D ionosonde data (Reinisch et al., 2009) provided by the Ebro Observatory (EB040), located at 40.80°N 0.50°E, and the DGS256 ionosonde data provided by the El Arenosillo Observatory (EA036), located at 37.27°N 6.94°W.These data were either obtained from the instrument operators directly or from the Global Ionosphere Radio Observatory repository (GIRO) (Reinisch & Galkin, 2011).The ionosonde ionograms allow to measure several ionospheric characteristics, such as the Maximum Useable Frequency (MUF), which is the highest frequency that can be used for radio communications between two locations at certain distance (typically 3,000 km), as well as to estimate the vertical electron density profile.
The GNSS Rate Of TEC Index (ROTI) (Pi et al., 1997) reflects the rate of change in TEC, which can be used to quantify the severity of ionospheric plasma depletions.Recent research has shown that GNSS ROTI can also detect ionospheric irregularities caused by small-scale plasma bubbles in the ionosphere (Liu et al., 2019;Yang & Liu, 2016).The ROTI is a standard deviation of the rate of TEC over a 5-min period.We retrieved ROTI maps through SIMuRG (https://simurg.iszf.irk.ru)(Yasyukevich et al., 2020) to detect ionospheric plasma bubbles.
To estimate the effect of ionospheric bubbles on GNSS performance, we calculate GNSS Precise Point Positioning (PPP) errors.We use the GAMP (GNSS Analysis software for multi-constellation and multi-frequency Precise positioning) open-source software (Zhou et al., 2018) to compute geocentric coordinates X, Y, Z (WGS84) of GNSS ground-based receivers.Receiver and satellite clock offsets are considered in the GAMP PPP solution by applying IGS precise satellite orbit and clock products and estimating the clock offset.Although we have selected a fixed station, we use the kinematic positioning mode for our analysis.The 22-hr (2-24 hr) median of X, Y, and Z geodetic coordinates (WGS84) for the static ground receiver is considered as a reference position.
The three-dimensional positioning error is then calculated as the root-mean-square error between the reference and the position at each epoch.
We also estimate prompt fluctuations in single-frequency GNSS positioning caused by ionospheric irregularities by calculating the contribution of the ionospheric-free combination (Hofmann-Wellenhof et al., 2007).This method effectively removes the first-order (up to 99.9%) ionospheric effect, which depends on the inverse square of the frequency.The receiver position is obtained from the observation file and estimated using broadcast orbits, pseudo-range, and carrier phase measurements (Mahooti, 2019).To isolate short-term scintillations caused by plasma depletions from global effects, we apply a 3-min running median-average filter.
We employ the method developed at Blanch et al. (2018) to identify EPBs over particular GNSS receivers.The algorithm makes use of GNSS data from the International GNSS Service (IGS) network, which is readily accessible via the CDDIS Data Center Website at ftp://cddis.gsfc.nasa.gov/pub/gps/data/daily/.This detection tool provides us valuable information about the location, duration, depth, and total disturbance of plasma bubbles sensed at the ionospheric pierce point of the line of sight from a given GNSS satellite to a ground receiver.
We use also the in-situ ion vertical velocity and ion density provided by the C/NOFS satellite (Comberiate & Paxton, 2010), short for Communication/Navigation Outage Forecasting System, whose data is available at Heelis (2023).The C/NOFS satellite was placed into a low Earth orbit with a perigee of 405 km and an apogee of 853 km.This measurement will help us to understand the behavior of the ionosphere, affecting radio wave propagation, satellite communications, and space weather.
We make use of SuperMAG magnetometer data, as presented in Gjerloev (2012), to estimate the variations of the EEJ in the area under investigation according to Equation 2 (Forbes, 1981).The data is readily available at Gjerloev (2012), and Table 1 lists the geomagnetic ground stations employed in this research.
In this equation, ΔH eq and ΔH off represent the differential horizontal magnetic field intensity (transient variation) for the stations located at the equatorial dip region and outside the equatorial region, respectively.
We estimate the PPEF using the empirical model developed by the Cooperative Institute for Research in Environmental Sciences at the University of Colorado Boulder (http://geomag.colorado.edu).This model offers reliable estimations of the variations due to solar wind, which are then mapped along interplanetary electric field data collected by various satellites (Manoj et al., 2013).
Finally, the Energetic Particle, Composition, and Thermal Plasma (ECT) investigation is a research effort supported by NASA, part of the broader Radiation Belt Storm Probes (RBSP) mission, which aims to improve our understanding of the Earth's radiation belts and the processes that control their behavior.The RBSP-ECT investigation focuses on measuring and characterizing the energetic particles and plasma that populate the Earth's radiation belts (Baker et al., 2012).One of the key datasets produced by this investigation is the Relativistic Electron-Proton Telescope (REPT) data.The REPT instrument is designed to measure and record the fluxes, energy spectra, and pitch angle distributions of electrons and protons in the radiation belts.The REPT data is particularly useful for studying the dynamics of the radiation belts and the relationship between the energetic particles and the complex electromagnetic fields that exist in this region of space.The data can also be used to improve models of the radiation belts and to develop strategies for mitigating the impact of space weather on spacecraft and humans in space.The RBSP-ECT data are publicly available at the RBSP website (https://rbsp-ect.newmexicoconsortium.org/rbsp_ect.php).

Results
Figure 1 presents the space weather conditions for the period from 24 February to 1 March 2014.We will focus on the moderate geomagnetic storm that occurred on 27-28 February, 2014. Figure 1a illustrates an enhanced proton flux (>10 MeV) due to an X4.9-class flare that occurred on 25 February 2014, along with a halo Coronal Mass Ejection (CME) (Yashiro et al., 2004).Notably, this surge in proton flux surpassed the 10 cm 2 s 1 sr 1 warning level established by the Space Weather Prediction Center of the National Oceanic and Atmospheric Administration (NOAA).At around 10:00 UTC, a change in the Bz component (Figure 1c) from North to South triggered an increase in geomagnetic activity at high latitudes, as evidenced by the AE, and Am indices in Figures 1d and 1f.Subsequently, the SYM-H and ASY-H indices (Figures 1f and 1h) demonstrate the overall energy content of the particles responsible for the fluctuation of an electric current carried by charged particles (10-200 keV), trapped in the magnetosphere at an altitude of approximately 20,000 to 50,000 km.
The shock associated with the CME, which reached a peak velocity of around 500 km/s (Figure 1e), arrived on Earth at approximately 16:50 UTC on 27 February 2014.This was reflected by an abrupt change in solar wind temperature, solar wind velocity, density, pressure, and IMF components.When the CME-associated shock wave reached the magnetopause, the Sudden Storm Commencement (SSC) occurred as a result of the compression of the magnetosphere due to the high pressure of the solar wind, marking the beginning of the geomagnetic storm.
During the initial phase of the storm, the plasma pressure reached a maximum value of approximately 12 nPa (Figure 1a), while the IMF Bz component reached a minimum value of approximately 10 nT.Note that Bx and By components of the IMF maintain small positive values of around 12 nT (Figure 1c).The Auroral Electrojet index AE exhibits peak values of 700 nT during the initial phase of the storm (Figure 1d).The storm-time behavior of the geomagnetic parameters, such as ASY-D, ASY-H, and Am, is shown in Figure 1f, where it is evident that the three indices reach the maximum values of 100 nT, 80 nT, and 60 nT during the initial phase of the storm.Figure 1f     0.2 to above 1 TECU/min.This gradient in ROTI indicates a significant irregularity potentially associated to a EPBs which will be discussed further.We can observe the evolution of the patch from 20:30 to 22:00 UTC, and then it returns to normal values.However, at regional scales defining the structure and dynamics of the plasma depletion may be challenging.For a global representation of ROTI, Figure S2 in Supporting Information S1 portrays high values and elongated shapes over Morocco and southern Spain between 21:00 and 22:00 UTC.It can be observed that over the southern areas of Spain, the ROTI index values exceed 0.5 TECU/min.Furthermore, note the limitations in covering the entire globe due to missing GNSS receivers in many parts of the world.
Figure 4 displays the GNSS corrections derived from the Ionospheric-Free combination at the RABT GNSS ground receiver between 12:00 and 24:00 UTC from 26 to 28 February, 2014.Notably, on February 27th, large corrections commenced at around 19:50 UTC with values ranging from ±20 m, and then doubled to ±40 m at approximately 21:00 UTC.It is important to note the time delay of roughly 4 hr from the beginning of the SSC, which is further discussed in the following section.Supporting this observation, Figure S3 in Supporting Information S1 shows 3D errors in the study area when estimating GNSS PPP on 27 February 2014, from 20:30 to 23:00 UTC at 30-min intervals, indicating errors up to 1 m during distinct phases of the storm in several receivers located in Morocco and southern regions of Spain.Each point marks a GNSS ground receiver.At 15:17 and 16:50 UTC, GNSS PPP errors are small and usually less than 20 cm (background level).At 18:30 UTC, PPP errors increased, showing the most significant amplitudes at the low latitude receivers.Then, using the algorithm developed by Blanch et al. (2018) to identify plasma depletions in the suspected area, we detected significant cases with plasma depletions larger than 5 TECU on the night of 26-28 February from the GNSS receivers in Rabat (RABT: 33.82°N, 6.85°W), Funchal (FUNC: 32.47°N, 16.91°W), La Palma (IPAL: 28.60°N, 17.89°W), and Mas Palomas (MAS1: 27.61°N, 15.63°W).Table 2 presents the results of the algorithm.The data in Table 2 corresponds well with the GNSS peak corrections in Figure 4, where the plasma depletion in our area of study is detected between approximately 19:50 and 21:00 UTC at the RABT GNSS ground receiver.
Figure 5 depicts all-sky images taken at the Oukaïmeden Observatory (31.2061°N, 7.8664°W) during the storm.
A plasma depletion can be unambiguously identified in the images as a black shadow at around 5°W.Additionally, Malki et al. (2018) found a dynamic transient of the plasma depletion moving to the East at approximately 50 m/s.The authors attributed this mechanism to thermospheric equatorward winds originating from high latitudes during the storm, which propagate westward due to Coriolis force (Buonsanto, 1990).These results,  provided by Malki et al. (2018), showed a sudden westward drift observed to occur for a couple of hours, starting at 20:00 UTC.Furthermore, at the same time, the SSUSI mapped the same area, identifying the depletion shown in Figure 5.The reddish strips in Figure 6 indicate high plasma density as seen by the SSUSI-derived NmF2.The gray regions aligned along the geomagnetic field lines over Morocco are identified as plasma depletions.It is noteworthy that these depletions occurred over the region of Spain during the main phase of the storm and became weaker during the recovery phase on February 28.Finally, it is also worth noting the good spatial agreement of the disturbance as observed with the ROTI data (Figure 3), all-sky images (Figure 5), and SSUSI (Figure 6), which exhibits a similar structure and occurs for the same period.
In Figure 7, we present the SSUSI night-side peak F-region electron density and limb images obtained around the Greenwich meridian at approximately 21:00 UTC 26, 27, and 28 February 2014.The top panels indicate the location of the top and bottom lines of the profiles displayed in the bottom panels.We observe an uplift of electron density peak (hmF2) of approximately 100 km on the day of the storm compared to the peak height observed in the days before and after the storm (see black arrows in the profiles of Figure 7).
Figure 8 shows examples of ionograms recorded at the El Arenosillo Observatory (EA036) and the Ebro Observatory (EB040) for storm day (27 February) and post-storm day (28 February).Ionogram data is available at the GIRO web portal (https://giro.uml.edu/didbase/).Differences in the ionograms of EA036 and EB040 are due to the different digisonde models operating at that time, a DGS 256 and a DPS 4D respectively (https://digisonde.com/digisonde.html).The latter, being a digital system, allows for a larger pulse and time-range resolution.
Although not shown here, we observe that during the storm's main phase at dawn on 27 February, the ionograms of EA036 show a large range spread-F and frequencies extending the ionograms scanning limits, starting at 20:00 UTC on 27 February until 01:00 UTC on 28 February 2014.The observed spread of echoes in the ionograms of EA036 suggests the presence of a plasma cavity above the measuring site.Moreover, the estimated peak height hmF2 density in this time interval for EA036 remains close to 400 km and above.The storm time ionograms recorded at EB040 show a rapid uplift of the electron density peak height hmF2 from about 350 km on 27 February at 19:45 UTC to about 500 km on 28 February at 00:30 UTC.Moreover, irregular ionogram echoes, indicating a tilted ionosphere and electron density gradients (e.g., Thampi et al., 2012), are observed from 27 February at 23:00 UTC to 28 February at 00:45 UTC.During the recovery phase on February 28 at dawn, the ionograms of EA036 remain within the scanning limits, with a critical frequency foF2 and peak height hmF2 values of 7 MHz and 3,502 km, respectively.The ionograms of EB040 for the recovery phase on February 28 at dawn indicate no irregular structures and a peak height below 325 km.Similar to Figure 7, we observe a clear uplift of approximately 100 km in both sites EA036 and EB040 during the main phase of the storm event compared to the recovery phase.
DPS4D ionograms contain information about the angle of arrival of radio echoes that are reflected within the ionosphere (Reinisch, et al., 2009).The capability of estimating the angle of arrival of the signals reflected in the ionosphere allows us to speculate about the movement of the irregularity over the station.Figure 9, shows the ionograms recorded at EB040 on 28 February 2014, at 00:30 UTC and 00:45 UTC.We observe oblique echoes in the ionograms from 22:45 to 23:30 UTC on February 27 with a SSE (South-Southeast) direction turning to have a NNE (North-Northeast) direction on 28 February at 00:45 UTC, which subsequently turn into undisturbed ionograms (i.e., vertical echoes only).The latter observations indicate a gradient in the electron density, which can be attributed to the movement of an irregularity in the NE direction.These results are in agreement with those presented in Figure 5. Furthermore, from February 27 at 23:45 UTC to February 28 at 00:25 UTC, we observe the spread of echoes in the ionograms of EB040, although not as evident as in the ionograms of EA036, again suggesting the presence of a plasma cavity (Spread-F).
Determining the slab thickness of the ionosphere can be a valuable tool for converting SSUSI's NmF2 into vTEC (Davies, 1990).The equivalent slab thickness refers to the width of an idealized ionosphere with a constant electron density and equal to the electron density maximum of the real ionosphere, so it can be estimated as the ratio of the vTEC over NmF2 (Jakowski & Hoque, 2021).The linear fitting results for the region of interest are presented in Table 3 and Figure 10.To generate these results, TEC data derived from SSUSI's airglow was combined with IGS TEC GIMs, resulting in enhanced maps of the region for 27 February 2014 at 18:00 and 20:00 LST.When compared to the IGS TEC GIMs, the SSUSI data offers superior resolution for representing small features such as depletions quasi-perpendicular to the equatorial dip crossing the EIA, with symmetrical patterns mirroring latitudinally.These small features are indicated with dashed black lines, showing decreases of approximately 100 TECU in the form of latitudinal troughs of a few hundred kilometers in width.Figure 10 also  11d highlights an anomaly that is larger than the other three.The top panels show the absence of anomalies in the data before 18:00-19:00 UTC, while after it, only the ion density observations exhibit large anomalies, with a wider range than seen during 18:00-19:00 UTC.This suggests that the depletions develop over time, lasting several hours until reaching maximum depth, and are likely caused by a sudden and powerful vertical ionospheric drift in the area during the SSC event.On the other hand, the ion velocities have no anomalies after 20:00 UTC, indicating its short-term duration.The peak ion velocities during the SSC period reach up to about 500 m/s, while the peak densities drop to 10 4 ion/cm 3 , which is significantly lower than the quiet background range of 10 6 ion/cm 3 .Note the existence of additional events during the period, such on 26 February during 20:00-21:00 UTC, which also displayed anomalies in both ion density and The transient variation in the differential horizontal magnetic field intensity during the geomagnetic storm that occurred on 27 February 2014 in the three stations located at the equatorial dip region and outside the equatorial region are shown in Figure 13b.Figure 13c displays the variations in the strength of the EEJ.Following the SSC, the Cowling conductivity, which is a measure of the conductivity of the ionosphere, significantly increased due to the movement of charged particles through the ionosphere.This increase in conductivity led to an increase in the current in the EEJ, which caused more intense variations in the Earth's magnetic field.During this time, the weakening of the eastward electric field was coupled with the sudden increase in Cowling conductivity, which changed the ionospheric dynamo electric field by decreasing the upward vertical E × B drift.Changes in the Earth's magnetic field during the storm led to large variations in the ionospheric dynamo electric field, which were observed through changes in the SYM-H index (Figure 13e).Note also in Figure 13d the injection of PPEF after the storm reduced the strength of the EEJ.The injection of westward PPEF led to a reduction in plasma diffusion, which subsequently resulted in a decrease in upward plasma drift.These effects were primarily driven by the westward PPEF.The northward orientation of the IMF Bz during the recovery phase and changes in the ionospheric dynamo electric field due to storm-induced equatorward wind further contributed to these variations.
Following the sudden enhancement after the SSC, three distinct peaks can be observed in both the EEJ and SYM-H.We can observe a significant correlation between the sequence and magnitude of these peaks and the plasma depletions previously identified in our analysis (Figures 10-12).Therefore, we have marked with yellow ellipses the correlated events, as we have done in Figure 10.The coupling is clear, demonstrating a strong agreement to plasma depletions detected from SSUSI's TEC, and ionospheric density and upward drifts measured by the C/ NOFS satellite in the aforementioned experiments.The relationship between the sequence and magnitude of the three peaks and the three plasma depletions shown in Figures 10-12 suggests an interrelated sequence of events, with each event potentially triggering each depletion.
To further investigate the global vTEC variations and distribution during the geomagnetic storm that occurred on 27 February 2014, we study TEC at approximately L-shells of 1.5, 1.2, and 1 for both the northern and southern hemispheres.The L-shell parameter is a measure of the distance from the magnetic dipole axis in units of Earth radii.Figure 14 shows the differences between the GNSS GIMs of TEC and the model of Calabia andJin (2020, 2019), while the locations of the L-shells are depicted in Figure S6 in Supporting Information S1.Approximately 2 hr after the SSC, we note a night-  time enhanced TEC at L-shell 1.5, while lower latitudes (L-shell 1.2 and 1) show a decrease in TEC at the magnetic equator.These responses suggest a deposition of incoming ions at high latitudes and an ion outflow in the equator, which may be generated by upward E × B drifts revealed in our previous analyses.
Finally, we propose that the sudden ionospheric upward drifts at sunset, which are more pronounced during storm conditions, could be influenced by the inner radiation belt particles.This is suggested by the significant decay in proton flux observed in Van Allen Probes data (Figure 15).However, the extent of the coupling would depend on various factors, such as the strength and duration of the upward drifts, as well as the characteristics of the particles themselves.Coulomb interactions between charged particles can either accelerate or decelerate particles, while wave-particle interactions involve the transfer of energy between electromagnetic waves and particles in a plasma.In general, sudden and strong vertical ionospheric drifts can create plasma density gradients and can enhance wave-particle interactions in the ionosphere.However, more detailed research would be needed to fully understand the potential effects between the inner radiation belt and the sudden ionospheric upward drifts.

Discussion
On 27 February 2014, sudden vertical ion drifts of approximately 500 m/s were observed between 18 and 19 UTC, coinciding with the presence of equatorial plasma bubbles detected by the C/NOFS satellite.These drifts, which were more pronounced during storm conditions, elevated ionospheric plasma to higher altitudes.However, it seems difficult for these small bubbles to drift up to the apex height of ∼3,000 km for L = 1.5 R E (Martinis et al., 2015) to interact directly with the inner radiation belt.Instead, the close development of plasma bubbles, auroral features, and stable auroral red (SAR) arcs in geographic latitudes (Martinis et al., 2015) suggests an indirect influence.As it is well known (Huang et al., 2001;Burke, Gentile, et al., 2004, Burke, Huang, et al., 2004), equatorial plasma bubbles prefer to develop in the Atlantic longitude sector and also in the African longitude sector, where the large plasma depletion under study is located.This spatially close development occurred due to the unusually strong equatorward movement of the plasmapause.During the main phase of the storm, characterized by a long-duration SYM-H minimum, high-energy protons in the inner radiation belt can become depleted but then regain their original state immediately in the recovery phase (Xu et al., 2019;He et al., 2023).The exact mechanisms behind this quick drop in high-energy proton flux are still unknown, but nonadiabatic processes are suspected.
Revisiting the whole process, it is known that the dayside neutral winds, in their natural configuration, generate an eastward electric field of approximately 1 mV/m.At the magnetic dip equator, this field produces a vertical E × B drift, which moves negative ions and electrons up to the F-layer and positive ions down to the E-layer.The resultant vertical polarization electric field, created by charge separation, is approximately one order of magnitude larger than the eastward electric field that produced it (Anderson et al., 2002).The vertical polarization electric field, when combined with the northward geomagnetic field, creates an East-West drift and increases the East-West electrical conductivity along the geomagnetic dip-equator.Therefore, at the evening terminator, the eastward electric field strengthens before reversing to a westward direction post-sunset due to the Haerendel-Eccles mechanism, which involves partial closure of the EEJ (Kelley et al., 2009).This process creates an upward current to meet the current continuity requirement of the F-region dynamo in the pre-zonal electric field reversal region (Prakash et al., 2009).The occurrence of plasma depletions during geomagnetic storms seems to be associated with sudden upward E × B drifts generated by strong eastward electric field perturbations.But the E × B drift developed within the plasma bubble is due to the interaction of the polarization E field developed due to gravitational Rayleigh-Taylor (R-T) plasma instability and the magnetic B field underlying the plasma bubble (Woodman & La Hoz, 1976;Fejer & Kelley, 1980;Li et al., 2021;Horvath & Lovell, 2021).The generalized R-T instability has been widely accepted as the physical mechanism responsible for the generation of EPBs (Li et al., 2021).But how the factors, which seed the development of R-T instability and control the dynamics of EPBs and resultant ionospheric scintillations, change on a short-term basis are not clear (Li et al., 2021).This E × B drift can be upward or downward defining which way the plasma bubbles drift.Note that the upward E × B developed over the dip equator and underlying the EIA is a different type of E × B drift.This E × B drift is the cross product of the equatorial eastward E field (a total E field including the E-F region dynamo E field and also, depending on the underlying geophysical conditions, the PPEF, DDEF, and the PRE E field) and the equatorial northward magnetic field producing and upward E × B drift driving the plasma fountain (Anderson et al., 2002).
Our explanation agrees with the findings of Tulasi Ram et al. (2008), where the LST dependence of the polarity and amplitude of electric fields (PPEF and DDEF) contributes to the development of spread-F irregularities.In a study conducted by Ghosh et al. (2020), it was demonstrated that the eastward electric field of pre-reversal enhancement can cause the F-region at the magnetic equator to be lifted to higher altitudes through upward E × B drift.However, when the up-flowing plasma loses momentum, it flows down along the inclined magnetic field lines to higher latitudes due to gravitational and pressure gradient forces.Therefore, the flow of charged particles from the equatorial region is contingent upon the orientation of magnetic field lines, which are generally closed within equatorial regions.

Conclusions
In this study, we have presented an accurate description and characterization of localized plasma depletions using the most advanced observational techniques and analyzed the relationships between these depletions to physical parameters such as solar wind, geomagnetic field, and electric currents.The geomagnetic storm of 27 and 28 February, 2014, occurred as a result of an X4.9 solar flare and halo CME.The CME resulted in an enhanced proton flux and produced a change in the IMF Bz component from North to South, which triggered an increase in geomagnetic activity at high latitudes, with the consequent fluctuation in the SYM-H and ASY-H indices.The storm's main phase ended on the same day at about 23:00 UTC, with a minimum value of SYM-H at 101 nT, and the recovery phase required several days to return to magnetically quiet conditions.We have also explored the advantages and challenges in integrating data from different observation systems, and how to combine different products to improve vTEC resolution for a better detail of small structures, specifically localized plasma depletions at sunset generated during geomagnetic storms.Finally, we have studied the possibilities offered by each observation system and what types of variables can be used to enhance existing empirical models for a better understanding of the coupled processes.In summary, our findings can be described as follows: 1. Small-scale plasma depletions in the ionosphere have been identified in the area of study using the method of Blanch et al. (2018).The GNSS PPP errors estimated with GAMP exceed undisturbed level at least twice in several receivers (1 m), specifically from 20:30 to 23:00 UTC in Morocco and southern regions of Spain.Due to long PPP convergence, the effects lasted much longer than bubbles were observed.Our experiment with the RABT GNSS station has shown ionospheric corrections above ±20 m during the transient of a small-scale plasma depletion over Morocco, while during quiet conditions, these corrections are an order of magnitude smaller.Electron and Proton Fluxes Energetic particle flux in Van Allen Belts ground-cameras can only provide discrete observations at localized emplacements and are highly sensitive to atmospheric meteorology.The vertical profiles of SSUSI data have shown a clear ionospheric uplift of approximately 100 km in altitude, and a drift of 3°in latitude.The altitude values are in good agreement with the ionosonde data.4. The range-time display of ionograms from EA036 and EB040 revealed the spread-F and the subsequent recovery of the ionospheric layers.The ionograms recorded at EB040 on 28 February 2014 have provided an estimate of the 2D movement of the irregularity, and the presence of a plasma cavity, enabling us to better understand the behavior of the ionosphere during the storm.5.During this intense geomagnetic storm, ionospheric depletions with elongated shapes were observed at midlatitudes.The variations in the strength of the EEJ during the storm were closely related to changes in the Cowling conductivity and the ionospheric dynamo electric field.These changes were observed through variations in the SYM-H index and the injection of PPEFs following the storm.6.Our discussion of the potential relationship between equatorial electric current flow variations and groundbased magnetometer data suggests three distinct peaks correlate with three vertical ionospheric drift uplifts, three density depletions, and are also associated with the high-resolution SYM-H index.
A multi-instrumental dataset is essential to fully understand the complex processes that occur in the upper atmosphere of the near-Earth environment.Each instrument provides valuable information that complements the data collected by the other instruments, allowing us to create a more complete picture of what is happening in the atmosphere.For example, in this study, we utilized GNSS, airglow remote sensing, ionosonde, magnetometer, and satellite data to investigate plasma depletions and their potential coupling mechanisms with space weather processes.Table 4 provides a summary of the data used in this study and its usability.Each instrument provided unique insights into the nature of the storm and the resulting plasma depletions.Therefore, it is not advisable to exclude any of the instruments in the study of space weather conditions as each one provides a unique perspective, and additional instruments, such as radio interferometers, could be useful in further improving the accuracy of the dataset and the understanding of the phenomena observed.

Figure 1 .
Figure 1.Physical parameters sensed within the Earth's environment during a larger period of the geomagnetic storm of 27-28 February, 2014.From top to bottom, the graph shows (a) Proton Flux, (b) Plasma Pressure, (c) IMF components, (d) Auroral indices, (e) Solar Wind Speed, (f) ASY-D, -H, and Am indices, (g) IEF Ey, Em, and Polar Cap index, and (h) SYM-H index.The vertical dashed red line at 16:50 UTC on 27 February 2014, marks the Storm Sudden Commencement (SSC).
depicts the storm-time response of the IEF Ey component, PCN, and Em, where these three variables reached a maximum value of 5 mV/m during the initial phase of the storm.The main phase of the storm started on the same day at about 19:30 UTC, with a minimum value of the SYM-H index of 101 nT at 23:25 UTC on 27 February.After 02:00 UTC on 28 February, the recovery phase started and required several days to return to the SYM-H index to quiet conditions.

Figure 2
Figure2illustrate the vTEC GIMs estimated from the IRI-2020 model (a and b), the model ofCalabia and  Jin (2020, 2019)  (c and d), and the IGS post-processed GIMs (e, and f) for 26 and 27 February 2014 at 20:00 UTC respectively.The IRI is a climatological model and its output shows no significant differences for similar epochs few days apart.Note the results of IRI are smaller than the GIMs and theCalabia and Jin (2020, 2019)  outputs.The IGS post-processed GIMs and theCalabia and Jin (2020, 2019)  model of vTEC provide better definition of the real spatial distribution.A more accurate representation of the real vTEC values is shown in FigureS1in Supporting Information S1.In FigureS1in Supporting Information S1, the UPC post-processed UQRG GIMs are presented at a 30-min interval starting from 20:30 UTC, demonstrating, in general, the incapacity to represent plasma depletions with structures of several kilometers in size.

Figure 3
Figure 3 depicts the GNSS ROTI over Morocco and the southern regions of Spain on 27 February 2014, from 20:30 to 23:00 UTC at 30-min intervals.A high ROTI value indicates a strong fluctuation in the electron density in the ionosphere.In this figure, several areas with high ROTI values are observed.For instance, the elongated patch in Figure 3d is marked with a green dashed line and shows a sharp gradient in ROTI values, ranging from

Figure 2 .
Figure 2. Ionosphere vTEC variations as predicted from the IRI-2020 model (a) and (b), the model of Calabia and Jin (2020, 2019) (c) and (d), and the IGS post-processed GIMs (e)-(f), for the 26 (a), (c) and (e) and 27 (b), (d) and (f) February 2014, at 22:00 UTC.A more accurate representation of the real vTEC values is available in the Figure S1 in Supporting Information S1 provided at the supporting information in this article.

Figure 3 .
Figure 3. Plasma variations as seen from GNSS ROTI maps for the geomagnetic storm of 27 February 2014 from (a) 20:30 to (f) 23:00 UTC, at intervals of 30 min.For a global representation of ROTI kindly refer to Figure S2 in Supporting Information S1.The Rabat GNSS station (RABT 33.82°N, 6.85°W) is indicated with a magenta dot and the alignment of the EPB is indicated with a green dashed line in (d).

Figure 4 .
Figure 4. Ionospheric corrections (3D distance to the 3-min median running-filter) estimated by Ionospheric-free combination at RABT GNSS ground receiver from 12:00 to 24:00 UTC on (a) 26, (b) 27 and (c) 28 February 2014.The SSC on 27 February 2014 is marked with a red dashed line.The corrections due to EPB effects from 20:00 to 21:00 UTC on 27 February 2014 are highlighted with a green dashed circle.

Figure 7 .
Figure 7.In top, we show the night-side peak F-region electron density measured by the satellite SSUSI F18 on (a) 26, (b) 27, and (c) 28 February 2014.The bottom panels show the corresponding limb images of the photon flux at 135.6 nm measured near the Greenwich meridian, at approximately 21:00 UTC.Panels (a)-(c) show the location of the bottom and top altitudes of the profiles (d)-(f).The black arrows show the ionospheric uplift of approximately 100 km.

Figure 9 .
Figure 9.The range-time display of ionograms from Ebro Observatory (EB040), Spain, on 28 February 2014 at (a) 00:30 UTC, and (b) 00:45 UTC.Note the sudden change of the echo from SSE to E in the data highlighted with dashed circles.

Figure 10 .
Figure 10.The (a) and (c) ion vertical velocity and the (b) and (d) ion density measured by the C/NOFS satellite on 27 February 2014 at (a) and (b) 18:00 and at (c) and (d) 20:00 LST.The IGS TEC GIMs improved with SSUSI airglow derived-TEC are shown in the background.The plasma depletions detected along the C/NOFS orbit path are marked with yellow ellipses and that from SSUSI with black dashed lines.

Figure 11 .
Figure 11.Ion density measured along the C/NOFS orbit path (see Figure 10) on 26-28 February 2014 between 18 and 22 LST and from 15°W to 15°E.From top to bottom, successive orbits within the ranges 16-17, 18-19, 20-22, and 22-23 UTC.The period when the Storm Sudden Commencement (SSC) occurs is marked with a red box.The red arrows mark the anomalies that occurred on 27 February.The black dashed-dot boxes show an aggregation of three anomalies.

Figure 12 .
Figure12.Ion vertical velocity measured along the C/NOFS orbit path (see Figure10) on 26-28 February 2014 between 18 and 22 LST and from 15°W to 15°E.From top to bottom, successive orbits within the ranges16-17, 18-19, 20-22, and 22- 23  UTC.The period when the Storm Sudden Commencement (SSC) occurs is marked with a red box.The red arrows mark the anomalies that occurred on 27 February.The black dashed-dot boxes show an aggregation of three anomalies.Figure13ashows the C/NOFS ion velocity and LST along UTC.

Figure 13 .
Figure 13.From top to bottom, we show (a) C/NOFS ion velocity and LST, (b) differential horizontal magnetic field intensity, (c) the strength of the EEJ, (d) the PPEF, and (e) the SYM-H during the geomagnetic storm that occurred on 27 February 2014.The Storm Sudden Commencement (SSC) is marked with a red dashed line.The possible correlation with the plasma depletions are marked with yellow circles.Similar to Figures 11 and 12, the group of 3 depletions is marked with green arrows.

Figure 14 .
Figure 14.High latitude TEC variations during the geomagnetic storm of 27 February 2014.Differences between GNSS GIMs of TEC and the model of Calabia and Jin at approximately McIlwain L-parameter (L-Shell) of (a) and (e) 1.5, (b) and (d) 1.2, and (c) 1, for (a) and (b) the northern and (d)-(e) the southern hemisphere.The model was set at constant quiet geomagnetic contribution (Am = 6).The red dashed line marks the SSC of the storm.We have marked with yellow ellipses the TEC enhancements at L-Shell 1.5.

Table 1
Geomagnetic Observatories Used in This Study for the Estimation of the EEJ

Table 4
Summary of the Data Used in This Study