Ingestion of GNSS‐Derived‐TEC Into NeQuick 2 Model Over South America

Ingestion of GNSS‐Derived‐TEC Into NeQuick 2 Model Over South America Taiwo Olusayo Osanyin 1,2 , Claudia Maria Nicoli Candido 1,3 , Fabio Becker‐Guedes 1 , Yenca Migoya‐Orue 4 , and John Bosco Habarulema 2,5 1 National Institute for Space Research, São José dos Campos, Brazil, 2 South African National Space Agency (SANSA), Hermanus, South Africa, 3 University of Vale do Paraiba, UNIVAP, São José dos Campos, Brazil, 4 STI Unit, The Abdus Salam International Centre for Theoretical Physics (ICTP), Strada Costiera 11, Trieste, Italy, 5 Department of Physics and Electronics, Rhodes University, Makhanda, South Africa Abstract This study examines the ingestion of total electron content (TEC) data from a network of ground‐ based Global Navigation Satellite System (GNSS) receivers into the NeQuick 2 model, providing a benchmark for evaluating the cost‐effectiveness of GNSS data ingestion for near‐real‐time ionospheric specification. A significant reduction in root mean square error (RMSE) between NeQuick 2 outputs and GNSS‐derived TEC was observed across various test stations, including those located at mid‐latitudes. During geomagnetically quiet periods, the performance difference between using 35 and 166 GNSS stations was less than 1%, while a similar trend, less than 2% difference was observed during disturbed periods with 34 and 177 stations, respectively. These findings highlight the importance of the spatial distribution of GNSS receivers in enhancing the model's accuracy, particularly over regions with complex ionospheric dynamics. Diurnal variations in vertical TEC (vTEC) over the test locations showed remarkable improvement during geomagnetic disturbances, with enhancements ranging from 37% to 68% in the low‐latitude region. Additionally, comparisons between NeQuick 2‐derived plasma frequencies and digisonde observations demonstrated strong agreement in the 100– 200 km altitude range, with average correlation coefficients of 0.97 and 0.95 at São Luís (2.58°S, 44.20°W), 0.93 and 0.55 at Boa Vista (2.83°N, 60.70°W), and 0.88 for both conditions at Campo Grande (20.40°S, 54.50°W) during quiet and disturbed periods, respectively. The lower correlation observed at Boa Vista during disturbed conditions may be attributed to limited data ingestion coverage in the Northern Hemisphere. Plain Language Summary Data ingestion can be described as an optimization technique where actual observations and model predictions, referred to as the background information, are combined to give the best estimate of the model state. It tends to drive an empirical model towards one specific data set and compute “effective parameters” associated with the model. Of particular application is the response of the ionosphere during geomagnetic storms. This is important to mitigate the adverse effects of space weather‐related disturbances on technology such as communication and navigation satellite systems, as their signals transmit through the ionosphere. The recent deployment of the GNSS receiver network globally allows for the homogeneous distribution of GNSS data, providing an alternative to ground‐based digisonde for applications relating to data ingestion. We developed a regional ionospheric model that incorporates GNSS TEC data into the NeQuick 2 model over South America. Specifically, a spatial resolution indexing method was used for the selection of GNSS stations, resulting in an evenly distributed data set and subsequently minimizing errors related to data ingestion. 1. Introduction The South American sector cuts across two distinct regions: middle‐ and low‐latitude regions. The phenomena driving the low‐latitude ionosphere contribute to critical ionospheric variability, significantly affecting communication and navigation systems (Correia et al., 2018; Rama Rao et al., 2006). These phenomena result from the complex interactions between space weather and atmospheric elements (Hargreaves, 1992; Marini‐ Pereira et al., 2021). Moreover, the ionosphere over low latitudes possesses features distinct from those of the mid‐latitudes owing to the low inclination of the geomagnetic field lines and a high fraction of the incident solar radiation characterizing this region (Abdu, 2016; Balan et al., 2013; De Abreu et al., 2017). The mid‐latitudes are characterized by high‐inclination magnetic field lines, which make ambipolar diffusion and meridional winds the dominant means of vertical transport for layer formation and structuring (Abdu, 2016; Mitra, 1946). The former RESEARCH ARTICLE 10.1029/2024SW004212 Key Points: •Utilizing spatial indexing for ground‐ based GNSS receivers' station selection is useful for producing fewer stations on the maps •Ingestion of the vTEC map greatly enhances the characterization of the low‐latitude ionosphere under quiet and disturbed geomagnetic conditions •Both standard and data‐driven NeQuick agree with digisonde measurements below the F2 layer peak height (hmF2) Supporting Information: Supporting Information may be found in the online version of this article. Correspondence to: T. O. Osanyin, [email protected] Citation: Osanyin, T. O., Maria Nicoli Candido, C., Becker‐Guedes, F., Migoya‐Orue, Y., & Habarulema, J. B. (2025). Ingestion of GNSS‐derived‐TEC into NeQuick 2 model over South America. Space Weather,23, e2024SW004212. https://doi.org/10.1029/ 2024SW004212 Received 8 OCT 2024 Accepted 31 JUL 2025 Author Contributions: Conceptualization: Taiwo Olusayo Osanyin, Claudia Maria Nicoli Candido, Fabio Becker‐Guedes, Yenca Migoya‐Orue, John Bosco Habarulema Data curation: Taiwo Olusayo Osanyin Formal analysis: Taiwo Olusayo Osanyin Investigation: Taiwo Olusayo Osanyin, Claudia Maria Nicoli Candido, Fabio Becker‐Guedes, Yenca Migoya‐ Orue, John Bosco Habarulema Methodology: Taiwo Olusayo Osanyin, Claudia Maria Nicoli Candido, Fabio Becker‐Guedes, Yenca Migoya‐ Orue, John Bosco Habarulema © 2025. The Author(s). This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. OSANYIN ET AL. 1 of 19 effect results in the redistribution of the electron density, forming a trough at the magnetic equator and crests at ±15–20° of the magnetic equator (Hanson & Moffett, 1966; Kelley, 2009; Nigussie et al., 2022). However, the positions of the EIA crests are strongly dependent on geographical longitude and solar activity. This plasma anomaly is known as the Equatorial Ionization Anomaly (EIA) (Appleton, 1946; Romero‐Hernandez et al., 2020). The principal drivers defining the characteristics of the EIA are the zonal electric field and meridional neutral winds. The eastward electric field drives a vertical plasma fountain at the magnetic equator during the daytime, creating an EIA. The neutral wind effect results in the interhemispheric asymmetry of the EIA by modulating the plasma fountain and moving the ionosphere at the conjugate hemispheres to different altitudes (Balan & Bailey, 1995). However, the competing effects of the chemical process,  E X  Bdrift, and neutral winds during storms are quite dynamic. Therefore, predicting the evolution of the ionospheric EIA during storms remains a crucial challenge in sophisticated models of the upper atmosphere (Stole et al., 2008; Zhang & Paxton, 2021). Because of the dangerous effects of space weather events on both ground‐ and space‐based technological systems, continuous ionosphere monitoring is important for reliable forecasts and specifications of the near‐Earth space environment. For this purpose, there has been a high demand for highly accurate and efficient ionospheric correction methods over the past few decades (e.g., Blanch et al., 2013). Different approaches have been adopted to assess and minimize ionospheric effects, including ionospheric empirical and theoretical models such as the NeQuick 2 model (Aa et al., 2018; Nava et al., 2006; Radicella & Nava, 2020), the International Reference Ionosphere (Bilitza et al., 2022; Bilitza & Reinisch, 2008; Galkin et al., 2012), and theoretical models (Huba et al., 2008; Ridley et al., 2006). Relative to the intrinsic limitations following the development of empirical models, they cannot predict future behavior or provide accurate descriptions of ionospheric dynamics in areas with limited databases (e.g., Habarulema et al., 2010; Migoya‐Orué et al., 2017; Radicella & Nava, 2020). NeQuick 2 is a three‐dimensional ionospheric electron density empirical model that calculates the total electron content along any line‐of‐sight (LOS) using numerical integration (Nava et al., 2008). It uses either the smoothed sunspot number (R12) or an equivalent of the solar radio flux index (F10.7) to describe the average state and regular variations in the ionosphere. Owing to the strong correlation between the two indices, one can be converted into the other using a simple formula (Zhao & Han, 2008). Often, difficulties can occur when using solar‐ based indices because they are far from the ideal proxies required for solar activity in the EUV part of the solar radiation spectrum (Nava et al., 2006). This has led to the development of several “effective” indices (Az) in a process known as data ingestion. Generally, data ingestion is a method of combining observations with a model to effectively express ionospheric morphology, facilitating a better understanding of space weather and correcting ionospheric errors in GNSS navigation and positioning. When the solar radio flux is adjusted in NeQuick, it is called the Effective Ionization Level; otherwise, it is referred to as the Effective Sunspot Number (Nigussie et al., 2012). Globally, Migoya‐Orué et al. (2015) ingested vTEC data from the Global Ionospheric Maps (GIM) into IRI (International Reference Ionosphere) model to reproduce the critical frequency of the F2 layer during both high and low solar activity. Recently, the reconstruction of electron density distribution using data‐driven empirical models has been more directed toward local/regional concerns (Hajra et al., 2016; Rao, 2007). This is because local models are superior to global models. The latter may smooth out the unique features of a particular location. Hence, developing local/regional models is necessary for satellite communication and navigation systems applications. Habarulema and Ssessanga (2017) adopted a similar concept in Africa and performed regional GNSS data ingestion using the IRI 2012 model. They validated their findings with neighboring vTEC from GNSS, digisondes, and electron density profiles obtained from COSMIC Radio Occultation data. Likewise, Migoya‐ Orué et al. (2017) conducted a characterization study by incorporating regional GNSS TEC data from the network of Global Ionospheric Maps (GIM) into the NeQuick 2 model to reproduce a series of critical frequency maps over Africa's EIA region. The ingestion method considers all the available GIM/GNSS‐derived data over a defined grid. Recently, the first application of data ingestion in the NeQuick 2 model over Brazil was conducted by Osanyin et al. (2023). They investigated the performance of the NeQuick 2 model using a single‐station ingestion technique during four geomagnetic storms ranging from weak to moderate storms in 2014. Their findings showed variability in the improvement of the model during the storm periods, with the highest improvement (∼83%) reported during the June event. However, the study suggested a further investigation of the data ingestion technique employing multiple GNSS stations to better understand ionospheric variability over this complex region, especially during hours of development of ionospheric irregularities such as spread F and plasma bubbles. Project administration: Taiwo Olusayo Osanyin Resources: Taiwo Olusayo Osanyin, Yenca Migoya‐Orue, John Bosco Habarulema Software: Taiwo Olusayo Osanyin, Yenca Migoya‐Orue, John Bosco Habarulema Supervision: Claudia Maria Nicoli Candido, Fabio Becker‐Guedes, Yenca Migoya‐Orue, John Bosco Habarulema Validation: Taiwo Olusayo Osanyin, Claudia Maria Nicoli Candido, Fabio Becker‐Guedes, Yenca Migoya‐ Orue, John Bosco Habarulema Visualization: Taiwo Olusayo Osanyin Writing – original draft: Taiwo Olusayo Osanyin Writing – review & editing: Taiwo Olusayo Osanyin, Claudia Maria Nicoli Candido, Fabio Becker‐Guedes, Yenca Migoya‐Orue, John Bosco Habarulema Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 2 of 19 Although remarkable improvements in climatological models by the ingestion of ground‐based GNSS TEC data have been presented in the literature mentioned above, our study focuses on the validation of optimal GNSS receiver stations whose TEC data would be sufficient to improve the NeQuick 2 model over South America using the technique of spatial resolution indexing for data selection and multiple station technique for data ingestion. In this context, the objectives of this study aim to address the following scientific questions: (a) How the distribution of ground‐based GNSS receiver stations affect the performance of data ingestion in NeQuick model? (b) What is the minimal GNSS receiver number of stations required for an ingestion scheme? (c) To what extent do the optimal drivers determined by data ingestion improve the model predictions? and (d) How effective is the NeQuick ingestion mode in describing the development of the EIA for ionospheric variability during geomagnetically quiet and disturbed conditions over South America? Concerning the above questions, this paper presents an algorithm for adapting ground‐based GNSS data into NeQuick 2 to study the development of EIA over South America and reconstruct ionospheric parameters such as TEC and critical frequency maps. The algorithm applies to regional GNSS networks within several hundred kilometers. Therefore, the significance of the study lies in investigating the minimal ground‐based measurements able to modify the NeQuick 2 model over a complex region to improve regional ionospheric studies. 2. Description of the Data Set and the Background Model's Improvement This study encloses observations from the GNSS‐derived TEC and digisonde data. The digisonde and GNSS data parameters were within 100–500 km and 0–20000 km. However, a comprehensive validation test was further limited to the F2 layer peak height (hmF2) because the topside profile of the digisonde measurement is modeled and not suitable for validation purposes. 2.1. Geomagnetic Data Figure 1a shows the variation in both the disturbance storm time (Dst) and planetary index (Kp) for the quiet (10 March 2015) and disturbed (17 March 2015) periods analyzed in this study. The intervals of interest were identified from the World Data Center for Geomagnetism, Kyoto. In terms of the Kp index, geomagnetic activity can be classified as quiet (Kp <4), active (Kp =4), minor storm (Kp =5), or major to severe storm (Kp >5) (Tan et al., 2018). The geomagnetic storm on 17 March 2015, was the strongest during solar cycle 24, with a minimum Figure 1. (a) Variation of Disturbance Storm Time (Dst) and planetary index (Kp) for geomagnetic quiet on March 10 (black) and storm time on March 17 (red), (b) Locations of GNSS stations collected over South America on 10 March 2015, and (c) Test stations used for validation, where stations in blue markers are representatives of GNSS receivers from the North to the South and those in blue markers are for stations of digisonde measurement. The red lines in (b) and (c) represent the geomagnetic equator. Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 3 of 19 Dst of −223 nT at 23 UT and a maximum Kp of 8 at 12, 15, and 21 UT (storm main phase), corresponding to strong disturbance geomagnetic conditions. 2.2. Ground‐Based GNSS Data GNSS TEC data over South America were collected during the study period from a wide database network of the International GNSS Service (IGS), Low‐Latitude Ionospheric Sensor Network (LISN), UNAVCO network, Argentinian (RAMSAC), Chilean network, Brazilian network (RBMC), and Uruguayan network (REGNA). This enables further investigation of the ionospheric variability in a wider spatial distribution. The data cover 10°N– 60°S latitude and 30°–80°W longitude, with approximately 312 receivers on 17 March 2015, and 274 receivers on 10 March 2015. The latitudinal and longitudinal distributions of the GNSS stations on 10 March 2015, are shown in Figure 1b, and the locations of the instruments used in the validation stage are displayed in Figure 1c. The geographic and geomagnetic coordinates of the instruments used for the validation are listed in Table 1. Batch processing of the GNSS TEC data was performed using the TEC calibration software developed at Boston College (Seemala & Valladares, 2011). The software uses a dual‐frequency technique to correct GPS observations with an estimation of the ionospheric delay obtained by a linear combination of dual‐frequency pseudorange measures. Owing to possible error sources associated with the thin‐shell approximation of the ionosphere and geometry during the conversion of slant to vertical TEC (Davies & Hartmann, 1997; Mannucci et al., 1998), a quality control check was performed on the processed GNSS data by removing outliers using the median and median average deviation (MAD) technique. This method has been widely employed in the literature (Dubazane & Habarulema, 2018; Huber & Ronchetti, 2009; Pignaberi et al., 2019) and is regarded as the most efficient and robust method. Further, only vTEC corresponding to satellite links with an elevation greater than 30° was considered to remove errors due to multipath (Takahashi et al., 2014) after which data was selected at a 1‐hr interval. The local time (LT) at each station was calculated by transforming the universal time (UT) using the expression LT =UT ±LONG/15°, where LONG is the longitude of each data point (Liu et al., 2020). Furthermore, TEC maps over South America were generated using the EMBRACE technique, and the details of the methodology can be found in Takahashi et al. (2016). 2.3. Observations From Digisonde Data The study includes analyses of ionospheric parameters, such as the F layer heights and plasma frequencies, taken from digisonde observations in Brazil. The data were obtained from the open‐access repository of the EMBRACE (Brazilian Study and Monitoring of Space Weather). Using the Sao Explorer software (Reinisch et al., 2005), the 10‐min cadence ionograms were manually scaled for four representative ionospheric locations, as shown by the blue circles in Figure 1c. Digisonde measurements were taken from stations at the northern crest of the EIA (Boa Vista), near the magnetic equator (SALU), and at the southern crest and border of the southern EIA crest (MSCG and CHPI), respectively. Table 1 Geographic and Geomagnetic Coordinates for GNSS Receivers and Digisonde Stations Used for Validation Receiver station Station code Observation Geographic Geomagnetic Local timeLat. ( o ) Lon. ( o ) Lat. ( o ) Lon. ( o ) Boa Vista BOAV GPS/Digisonde 02.83 −60.70 11.03 14.15 UT‐3 Belem BELE GPS −01.40 −48.45 01.68 25.78 UT‐3 Sao Luis SALU Digisonde −02.58 −44.20 −1.19 29.46 UT‐3 Riobamba RIOP GPS −01.65 −78.66 10.34 −6.74 UT‐3 Colider MTCO GPS −10.80 −55.45 −03.19 15.82 UT‐3 Rio Paranaíba MGRP GPS −19.20 −46.17 −14.42 21.88 UT‐3 Campo Grande MSCG Digisonde −20.40 −54.50 −11.98 14.19 UT‐3 Cachoeira Paulista CHPI Digisonde −22.68 −44.98 −17.86 21.88 UT‐3 Santa Rosa SRLP GPS −36.62 −64.29 −23.98 05.19 UT‐5 San Juan SUAN GPS −43.29 −65.11 −29.95 5.03 UT‐5 Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 4 of 19 The F2 layer frequency, f oF2is well known to be proportional to the ionospheric ionization peak value (NmF2) through the relation: NmF2=1.24 ×104(foF2)2(1) where NmF2is expressed in cm −3 and foF2in MHz. 2.4. Adaptation of the NeQuick 2 Model to Ground‐Based GNSS TEC Data The ingestion process involves calculating an effective ionization level based on vertical Total Electron Content (vTEC) measurements from GNSS receivers. The modified NeQuick 2 then incorporates this effective ionization level to better match outputs of interest, such as neighboring TEC, the plasma frequency, and F2 layer critical frequency. The subsequent steps are duly followed. Step 1: Grid‐Based Spatial Resolution Indexing for Stations density selection The study employs a grid‐based approach to select GNSS stations in South America. The method involves the following sub‐steps: 1. Load and structure GNSS station metadata (station ID, longitude, latitude, and ellipsoidal height. This data contains the overall stations over the region of study for a specific DOY. 2. Divide the region into grid cells: The study area is divided into fixed grid cells, starting with 2° in latitude ×2° in longitude spatial resolution. 3. Compute the median ellipsoidal height in each grid cell in 2) above 4. Select closest‐to‐median height in each grid cell. 5. Repeat the process for different spatial resolutions: Steps 2 and 3 are repeated using different spatial resolutions (e.g., 2 ×4, 4 ×4, etc) to assess the impact of resolution on GNSS station selection. 6. Plot the resulting stations from (5) on a map (see Figure 2). Figure 2. Sequence of GNSS receiver stations generated at different resolutions with the corresponding number of stations over South America on 10 March 2015. The upper panel represents maps inferred from uniform resolution (n ×n), and the lower panels are those from non‐uniform resolution (n ×m). The legend in each plot indicates the spatial grid index sizes. Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 5 of 19 The method described in step 1 was implemented using GNSS stations on 10 and 17 March 2015 considering uniform (n x n) and non‐uniform (n x m) grid resolutions. In the uniform grid, latitude and longitude bin sizes remain constant, whereas in the non‐uniform grid, latitude separation (n) is fixed due to the sparse receiver distribution. As shown in Figure 2, spatial indexing technique results in evenly distributed and good spatial coverage of GNSS stations geographically. In the case of uniform grid, the number of GNSS stations decreases as the grid size increases. For instance, a coarser grid size of 6° ×6° results in 57 selected GNSS stations compared to finer grid sizes. Step 2: Collection of GNSS TEC data from each map obtained in Step 1 The individual maps shown in Figure 2are products of station maps based on spatial grid selection. The basic idea is to evaluate the map whose TEC data produces the least root mean square error in the NeQuick 2 model after data ingestion. The TEC data from each map were loaded as a single file, filtered for a specific epoch, and gridded with a resolution of 1° ×1° to achieve a finer TEC resolution before ingesting into the model. Step 3: Model's optimization NeQuick 2 is optimized as a function of the daily effective ionization level (Az) to be adapted to the measured vTEC values. Following the single station technique described in (Nava et al., 2006; Nigussie et al., 2012; Osanyin et al., 2023), a similar methodology has been employed for the multiple station technique that is, ingestion of vTEC maps (Nava et al., 2011). This technique searches for a range of values of ionization parameter (F 10.7index) to minimize the deviation of the model from measured values. We present a data ingestion model for the regional ionosphere based on a binary search algorithm optimization technique. The algorithm involves a FORTRAN iteration process (see Figure 3) that optimally updates the model's estimate by statistically minimizing the differences between the measured value and the model's prediction as given in Equation 2. ∆vTEC (t)=vTECmti −vTECoti (2) where vTECmt and vTECot are the model's predicted vTEC and GNSS observed vTEC at time t and grid point i. At the start, the NeQuick 2 calculates TEC at each epoch and corresponding grids (grid latitude and grid longitude) obtained in Step 2 using the input of the daily solar radio flux (F10.7index) as the solar activity proxy. At each grid point, the delta TEC (∆vTEC) is estimated. The iteration process, implemented in the flowchart optimizes the F10.7 index based on whether the modeled TEC is greater or less than the GNSS TEC, using a 0.1 ×F10.7 increment/decrement. A relative error threshold typically 5% (Adolfs et al., 2022; Ma et al., 2022; Mengistu Tsidu & Melaku Zegeye, 2020; Osanyin et al., 2023) ensures model accuracy, particularly in low‐latitude. Az (t)=F10.7(t)±0.1×F10.7(t)(3) where Az is the optimized flux at time t. The iteration process continues until the desired threshold is achieved, and the final output of the algorithm is the effective ionization parameter, regarded as the optimum flux (Az). The above‐described optimization technique is applied to all the GNSS gridded maps in Step 2. The Az values from each map are the regional Az over the defined grid, which is subsequently used in the model as the new solar activity proxy to reconstruct desired ionospheric parameters of interest. The above is summarized in Figure 3, similar to the concept adopted by Atiq et al. (2021). The model's accuracy is evaluated using the Root Mean Square Error (RMSE) RMSE =  (∑N i=1(TECGNSS −TECNeQuick(Az))2) N √ √ √ √(4) where N represents the number of individual observations during the current interval and TECGNSS and TECNeQuick are TEC values from GNSS measurement and NeQuick 2 model, respectively. Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 6 of 19 3. Results and Discussions 3.1. Statistical Analysis of Data Ingestion Using Different Spatial Resolutions for Station Selection Figure 4shows the statistical results of NeQuick's performance before and after data ingestion for geomagnetically quiet (DOY 069) and disturbed (DOY 076) conditions, using the RMSE and percentage improvement as metrics. Six GNSS receivers, Boa Vista (BOAV), Belem (BELE), Riobambe (RIOP), Colider (MTCO), Rio Paranaiba (MGRP), Santa Rosa (SRLP), and Sua Juan (SUAR) were used for internal validation. Figure 4A) and (b) represent the RMSE and percentage improvement obtained for the 0.5 TECu threshold, and the result is shown for uniform (n ×n, left panel) and non‐uniform (n ×m, right panel) resolution. At the top of each plot is the legend, which indicates the resolution and the corresponding number of GNSS stations whose TEC data were retrieved and used to improve the model. Among the various colors used in Figure 4, gray represents the standard NeQuick 2 model driven by the daily F10.7 index. The other colors correspond to results based on average Az values derived from the maps shown in Figure 2. As expected, the standard NeQuick prediction exhibits the highest RMSE compared to the adapted version of the model. Although one would anticipate greater improvement with an increasing number of ingested GNSS stations, this trend does not hold consistently, particularly for non‐uniform resolutions, where prediction accuracy fluctuates. While the results for non‐uniform grids are generally comparable to those of uniform grids, they are not discussed further here due to the significantly higher number of GNSS stations involved. Uniform resolution offers more consistent performance across latitudes with fewer stations. In panels (a) and (b), RMSE Figure 3. The flowchart for adapting the NeQuick 2 model to ground‐based GNSS TEC measurement. Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 7 of 19 Figure 4. Root‐mean‐square error (RMSE) and percentage improvement values for different spatial resolutions using threshold of (a) 0.5 TECu (b) 3 TECu and (c) threshold of 0.5 TECu and 3 TECu for 6 ×6 resolution only considering the equatorial, low‐latitude and mid‐latitude stations from the North to the South. Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 8 of 19 values depicted in (a) and (e) tend to decrease across the stations, with relatively lower improvement shown in (c) and (g), especially in the mid‐latitude region (SRLP) during quiet geomagnetic conditions. On average, the RMSE across latitudes shows nearly the same level of improvement for both the (8 ×8) and (2 ×2) grid resolutions. This reflects the core idea of the methodology in evaluating the optimal quantity of ground‐based GNSS data needed to enhance the NeQuick model. For DOY 067, 35 and 166 GNSS stations were used for the (8 ×8) and (2 ×2) grids, resulting in improvements of 25.11% and 25.78%, respectively. However, on the disturbed day (DOY 076), the (8 ×8) grid achieved a slightly higher improvement of 1.14% more than the (2 ×2) grid despite using fewer stations (34 compared to 177). The main finding here is that fewer stations may achieve similar desired accuracy in TEC and foF2 estimation as opposed to using all the available data, which slows down the computation process. This is useful in minimizing computational costs for real‐time applications such as nowcasting and forecasting based on this methodology. The results may be different for monitoring small‐ scale structures such as medium‐scale (Travelling Ionospheric Disturbances) TIDs. In similar studies conducted by Migoya‐Orué et al. (2017) and Habarulema and Ssessanga (2017), the authors used all available GNSS stations during the study period to reproduce the critical frequency maps over Africa. While the approach in this study strongly depends on data availability over a defined grid (e.g., South America), it may not be suitable over Africa, where there is data paucity. BOAV and MGRP are located in the northern and southern equatorial ionization anomaly (EIA) crest regions, respectively, while MTCO lies in the off‐equatorial zone. Despite its proximity to the equator alongside RIOP, BELE shows greater improvement for DOY 069, likely influenced by longitudinal differences. MGRP records the least improvement, highlighting the need for further refinement of climatological models over the Southern EIA crest. This region, located within the core of the South American Magnetic Anomaly (SAMA), experiences pronounced ionospheric variability (Abdu et al., 2005). In contrast to the quiet period, DOY 076 shows the highest improvement at the southern crest (MGRP), suggesting that data ingestion is highly sensitive to both latitude and geomagnetic activity (Osanyin et al., 2023). Notably, for DOY 076, the (10 ×10) grid resolution achieves comparable improvement to other resolutions, despite using only 27 stations. This underscores the critical role of GNSS station distribution in model performance. Therefore, a well‐distributed GNSS network with broad spatial coverage can significantly enhance climatological models. Based on the results in Figure 4, a minimum of 35 GNSS stations, corresponding to the 8 ×8 grid resolution, is recommended for ingestion, as it delivers superior performance across latitudes under geomagnetically disturbed conditions, which are of particular concern to the space weather community. Consequently, the NeQuick 2 results presented in the following sections for validation are based on the improvements achieved using the 8 ×8 grid resolution. 3.2. EIA Development and Variability Over South America The study of EIA development over South America has been investigated under both quiet (10 March 2015) and disturbed (17 March 2015) geomagnetic conditions using a TEC mapping technique. The GNSS maps were produced using (1° ×1°) spatial and 1‐hr temporal resolution, and the results were compared to the NeQuick 2 model. Both daytime and post‐sunset EIA (PS‐EIA) were observed. Diurnal variation at specific locations displays vTEC variation during the hours of the day with TEC enhancements between (13:00‐15:00 LT) and near local midnight at (20:00‐21:00 LT) in the low‐latitude stations during the disturbed period. 3.2.1. Quiet Time; 10 March 2015 A sequence of 2‐D TEC maps constructed from GNSS measurements over South America on 10 March 2015, is shown in Figure 5a at 2‐hr intervals from 11:00‐01:00 UT. The magnetic conditions during these observations were quiet (Kp =2, Dst = − 8 nT), and the F 10.7was approximately 120.9 ×10 −22 Wm −2 Hz −1 . For accuracy and better comparison with NeQuick, 239 stations were used to generate the maps after excluding those utilized (35 stations) for the model's improvement. Depletion and enhancement of TEC values can be seen in the color bar with maxima around red. It can be observed that the EIA starts to develop at 11:00 UT (∼8:00 LT at 30°W) on the eastern coast of Brazil during the day and expands westward with maximum enhancement between 17:00 and 19:00 UT (14:00 and 16:00 LT). Interhemispheric asymmetry can be observed, with a more pronounced northern crest at 19:00 UT. The EIA asymmetry has been explained by plasma recombination resulting from the effects of meridional winds (De Siqueira et al., 2011; Khadka et al., 2018; Loutfi et al., 2022), whereas the degree depends Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 9 of 19 As such, the model is best suited for long‐term, average ionospheric conditions rather than real‐time or event‐ specific applications. In this study, we investigated the minimum number of ground‐based GNSS receivers needed to significantly improve the performance of NeQuick 2 over a defined spatial grid across South America. This approach provides a benchmark for cost‐effective data ingestion in climatological models. Our results indicate that employing a uniform grid resolution (n ×n) not only reduces the number of required GNSS stations but also yields more consistent performance across latitudes. Notably, the RMSE between GNSS observations and NeQuick 2 predictions decreased substantially, particularly during geomagnetically disturbed conditions. Improvements ranged from 4% to 45% on quiet days and 33%–68% during disturbed periods, underscoring the sensitivity of the ingestion scheme to both geographic latitude and geomagnetic activity. Interestingly, using data from just 35 and 166 GNSS stations during quiet conditions resulted in less than 1% difference in performance, suggesting that a smaller, well‐distributed network can be sufficient for model enhancement, rather than relying on a large number of stations. For the first time, we examined the evolution of the Equatorial Ionization Anomaly (EIA) over South America using the NeQuick 2 model. The data‐ingested version of the model demonstrated improved capability in capturing the spatial and temporal development of the EIA. Additionally, we evaluated the model's accuracy in predicting plasma frequencies by comparing its outputs with digisonde observations. The results show strong agreement at lower altitudes (100–200 km), with decreasing correlation at the F‐region peak (∼250–350 km), likely due to limitations in vertical resolution and dynamic ionospheric processes. Overall, our findings highlight the importance of strategically distributed GNSS stations for enhancing ionospheric models through data ingestion. These insights are particularly relevant for ongoing efforts to deploy low‐ Table 2 Correlation Coefficient Between Digisonde Measurements and Ingested NeQuick 2 of the Plasma Frequency at Different Heights During Quiet (DOY 069) and Disturbed (DOY 076) Geomagnetic Conditions Before and After Data Ingestion SALU BOAV MSCG Altitude (km) Quiet Disturbed Quiet Disturbed Quiet Disturbed 100 0.97/0.97 0.97/0.96 0.93/0.95 0.57/0.55 0.89/0.91 0.91/0.87 150 0.99/0.99 0.99/0.99 0.96/0.98 0.67/0.65 0.95/0.96 0.96/0.95 200 0.94/0.96 0.90/0.91 0.80/0.87 0.30/0.40 0.72/0.77 0.75/0.82 250 0.81/0.77 0.68/0.71 0.71/0.58 0.16/0.54 0.42/0.27 0.18/0.27 300 0.64/0.49 0.50/0.57 0.65/0.58 0.21/0.21 0.82/0.92 0.50/0.84 350 0.64/0.77 0.70/0.77 0.82/0.86 0.64/0.62 0.94/0.93 0.80/0.53 400 0.68/0.85 0.82/0.81 0.93/0.97 0.87/0.84 0.85/0.91 0.74/0.91 Table 3 Root Mean Square Error (RMSE) Between Digisonde Measurement and Ingested NeQuick 2 of the Plasma Frequency at Different Heights During Quiet (DOY 069) and Disturbed (DOY 076) Geomagnetic Conditions Before and After Data Ingestion SALU BOAV MSCG Altitude (km) Quiet (TECu) Disturbed (TECu) Quiet (TECu) Disturbed (TECu) Quiet (TECu) Disturbed (TECu) 100 0.94/0.73 0.92/0.80 0.81/0.55 1.09/1.06 0.82/0.76 0.83/0.74 150 0.71/0.58 0.69/0.62 0.92/0.60 1.40/1.41 1.03/0.91 0.96/0.87 200 1.33/1.16 1.54/1.39 1.54/1.07 2.30/2.34 1.79/1.61 1.56/1.32 250 1.72/1.90 2.23/2.12 2.21/2.80 3.57/3.45 2.24/2.60 3.90/3.84 300 2.19/2.45 2.45/2.37 2.72/3.34 4.02/3.95 2.33/2.87 3.45/3.39 350 2.49/1.89 2.04/1.80 2.32/2.31 2.55/3.24 1.82/1.57 2.52/2.37 400 1.68/1.66 1.84/2.06 2.03/2.10 1.72/2.21 2.35/1.66 3.53/2.15 Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 16 of 19 cost GNSS receiver networks in under‐observed regions. Furthermore, this study represents the first comparison of NeQuick 2 plasma frequency estimates with digisonde data over South America, contributing to a better understanding of ionospheric behavior during space weather disturbances and supporting the development of region‐specific ionospheric models. Data Availability Statement We acknowledge the STI Unit of the ICTP, Trieste, Italy, for providing the NeQuick 2 source code (https://tict4d.ictp.it/nequick2/source-code). We thank Embrace/INPE for providing the digisonde data (https://www2. inpe.br/climaespacial/). The Dst index data (https://wdc.kugi.kyoto-u.ac.jp/wdc/) was obtained from the World Data Center for Geomagnetism, Kyoto. Kp index and solar radio flux data can be accessed from the NOAA (National Oceanic and Atmospheric Administration) website (https://spdf.gsfc.nasa.gov/pub/data/omni/). The GNSS ground‐based receiver data were collected from different GNSS networks in South America: RBMC from IBGE (https://www.ibge.gov.br), RAMSAC of Argentina (https://www.ign.gob.ar/NuestrasActividades/ Geodesia/Ramsac/DescargaRinex), IGS (https://cddis.nasa.gov/archive/gnss/data/hourly); and UNAVCO (https://www.unavco.org/data/gps-gnss/gps-gnss.html). The Python code for selecting GNSS data used in the NeQuick 2 model and samples of GNSS data during the study period are available in the Zenodo repository at T. Osanyin, (2025). References Aa, E., Ridley, A., Huang, W., Zou, S., Liu, S., Coster, A. J., & Zhang, S. (2018). An ionosphere specification technique based on data ingestion algorithm and empirical orthogonal function analysis method. Space Weather,16(9), 1410–1423. https://doi.org/10.1029/2018sw001987 Figure 9. Variation of F2 layer critical frequency (foF2) during (a) quiet (10 March 2015) and (b) disturbed (17 March 2015) geomagnetic conditions. The red solid lines depict digisonde measurement, blue solid and green dotted lines are NeQuick 2 predictions before and after data ingestion. Acknowledgments Taiwo Osanyin sincerely thanks the South African National Space Agency (SANSA), Hermanus, South Africa, for their financial support and hospitality through the Scientific Committee on Solar‐Terrestrial Physics (SCOSTEP) visiting graduate research program. The authors also appreciate the support of the Brazilian Ministry of Science, Technology, and Innovation (MCTI), Brazilian Space Agency (AEB), and Coordination for the Improvement of Higher Education Personnel (CAPES: Grant 88887.369369/ 2019‐00). We greatly appreciate the Editor (Dr. Shasha Zou) and two anonymous reviewers for their observations, comments, and suggestions, which are very helpful in clarifying the significance of the study. Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 17 of 19 Abdu, M. A. (1997). Major phenomena of the equatorial ionosphere‐thermosphere system under disturbed conditions. Journal of Atmospheric and Solar‐Terrestrial Physics,59(13), 1505–1519. https://doi.org/10.1016/s1364-6826(96)00152-6 Abdu, M. A. (2016). Electrodynamics of ionospheric weather over low latitudes. Geoscience Letters,3, 1–13. https://doi.org/10.1186/s40562016-0043-6 Abdu, M. A., Batista, I. S., Carrasco, A. J., & Brum, C. G. M. (2005). South Atlantic magnetic anomaly ionization: A review and a new focus on electrodynamic effects in the equatorial ionosphere. Journal of Atmospheric and Solar‐Terrestrial Physics,67(17–18), 1643–1657. https://doi. org/10.1016/j.jastp.2005.01.014 Adolfs, M., Hoque, M. M., & Shprits, Y. Y. (2022). Storm‐time relative total electron content modelling using machine learning techniques. Remote Sensing,14(23), 6155. https://doi.org/10.3390/rs14236155 Appleton, E. V. (1946). Two anomalies in the ionosphere. Nature,157(3995), 691. https://doi.org/10.1038/157691a0 Astafyeva, E., Zakharenkova, I., & Förster, M. (2015). Ionospheric response to the 2015 St. Patrick's day storm: A global multi‐instrumental overview. Journal of Geophysical Research: Space Physics,120(10), 9023–9037. https://doi.org/10.1002/2015ja021629 Atiq, M., Ameen, M. A., Sadiq, N., Khursheed, H., Ali, M. S., Yu, X., & Ameer, A. (2021). Estimating foF2 from GPS TEC over islamabad and Darwin using NeQuick2 during 2011–2014. Advances in Space Research,67(5), 1559–1569. https://doi.org/10.1016/j.asr.2020.12.003 Balan, N., & Bailey, G. J. (1995). Equatorial plasma fountain and its effects: Possibility of an additional layer. Journal of Geophysical Research, 100(A11), 21421–21432. https://doi.org/10.1029/95ja01555 Balan, N., Otsuka, Y., Nishioka, M., Liu, J. Y., & Bailey, G. J. (2013). Physical mechanisms of the ionospheric storms at equatorial and higher latitudes during the recovery phase of geomagnetic storms. Journal of Geophysical Research: Space Physics,118(5), 2660–2669. https://doi. org/10.1002/jgra.50275 Balan, N., Shiokawa, K., Otsuka, Y., Watanabe, S., & Bailey, G. J. (2009). Super plasma fountain and equatorial ionization anomaly during penetration electric field. Journal of Geophysical Research,114(A3). https://doi.org/10.1029/2008ja013768 Bilitza, D., Pezzopane, M., Truhlik, V., Altadill, D., Reinisch, B. W., & Pignalberi, A. (2022). The international reference ionosphere model: A review and description of an ionospheric benchmark. Reviews of Geophysics,60(4), e2022RG000792. https://doi.org/10.1029/2022rg000792 Bilitza, D., & Reinisch, B. W. (2008). International reference ionosphere 2007: Improvements and new parameters. Advances in Space Research, 42(4), 599–609. https://doi.org/10.1016/j.asr.2007.07.048 Blanc, M., & Richmond, A. D. (1980). The ionospheric disturbance dynamo. Journal of Geophysical Research,85(A4), 1669–1686. https://doi. org/10.1029/ja085ia04p01669 Blanch, E., Marsal, S., Segarra, A., Torta, J. M., Altadill, D., & Curto, J. J. (2013). Space weather effects on earth's environment associated to the 24–25 October 2011 geomagnetic storm. Space Weather,11(4), 153–168. https://doi.org/10.1002/swe.20035 Borries, C., Mahrous, A. M., Ellahouny, N. M., & Badeke, R. (2016). Multiple ionospheric perturbations during the Saint Patrick's day storm 2015 in the european‐african sector. Journal of Geophysical Research: Space Physics,121, 11–333. https://doi.org/10.1002/2016ja023178 Correia, E., Muella, M. T., Alfonsi, L., Prol, F. d., Camargo, P. O., & Ugur Sanli, D. (2018). GPS scintillations and total electron content climatology in the southern American sector. Accuracy of GNSS Methods. Davies, K., & Hartmann, G. K. (1997). Studying the ionosphere with the global positioning system. Radio Science,32(4), 1695–1703. https://doi. org/10.1029/97rs00451 De Abreu, A. J., Martin, I. M., Fagundes, P. R., Venkatesh, K., Batista, I. S., de Jesus, R., et al. (2017). Ionospheric F‐region observations over American sector during an intense space weather event using multi‐instruments. Journal of Atmospheric and Solar‐Terrestrial Physics,156, 1– 14. https://doi.org/10.1016/j.jastp.2017.02.009 De Siqueira, P. M., De Paula, E. R., Muella, M. T., Rezende, L. F., Abdu, M. A., & Gonzalez, W. D. (2011). Storm‐time total electron content and its response to penetration electric fields over South America. Annales Geophysicae,29(10), 1765–1778. https://doi.org/10.5194/angeo-291765-2011 Dubazane, M. B., & Habarulema, J. B. (2018). An empirical model of vertical plasma drift over the African sector. Space Weather,16(6), 619– 635. https://doi.org/10.1029/2018sw001820 Fagundes, P. R., Cardoso, F. A., Fejer, B. G., Venkatesh, K., Ribeiro, B. A., & Pillat, V. G. (2016). Positive and negative GPS‐TEC ionospheric storm effects during the extreme space weather event of March 2015 over the Brazilian sector. Journal of Geophysical Research: Space Physics,121(6), 5613–5625. https://doi.org/10.1002/2015ja022214 Galkin, I. A., Reinisch, B. W., Huang, X., & Bilitza, D. (2012). Assimilation of GIRO data into a real‐time IRI. Radio Science,47(4), 1–10. https:// doi.org/10.1029/2011rs004952 Habarulema, J. B., McKinnell, L.‐A., & Opperman, B. D. (2010). TEC measurements and modelling over Southern Africa during magnetic storms; a comparative analysis. Journal of Atmospheric and Solar‐Terrestrial Physics,72(5–6), 509–520. https://doi.org/10.1016/j.jastp.2010. 01.012 Habarulema, J. B., & Ssessanga, N. (2017). Adapting a climatology model to improve estimation of ionosphere parameters and subsequent validation with radio occultation and ionosonde data. Space Weather,15(1), 84–98. https://doi.org/10.1002/2016sw001549 Hajra, R., Chakraborty, S. K., Tsurutani, B. T., DasGupta, A., Echer, E., Brum, C. G., et al. (2016). An empirical model of ionospheric total electron content (TEC) near the crest of the equatorial ionization anomaly (EIA). Journal of Space Weather and Space Climate,6, A29. https:// doi.org/10.1051/swsc/2016023 Hanson, W. B., & Moffett, R. J. (1966). Lonization transport effects in the equatorial F region. Journal of Geophysical Research,71(23), 5559– 5572. https://doi.org/10.1029/jz071i023p05559 Hargreaves, J. K. (1992). The solar‐terrestrial environment: An introduction to geospace‐the science of the terrestrial upper atmosphere, ionosphere, and magnetosphere. Cambridge University Press. Huba, J. D., Joyce, G., & Krall, J. (2008). Three‐dimensional equatorial spread‐F modeling. Geophysical Research Letters,33(10), L19106. https://doi.org/10.1029/2009GL040284 Huber, P. J., & Ronchetti, E. M. (2009). Robust statistics (2nd ed.). Hoboken. NJ: Wiley. https://doi.org/10.1007/978-3-642-04898-2_594 Kelley, M. C. (2009). The earth's ionosphere: Plasma physics and electrodynamics. Academic Press. Khadka, S. M., Valladares, C. E., Sheehan, R., & Gerrard, A. J. (2018). Effects of electric field and neutral wind on the asymmetry of equatorial ionization anomaly. Radio Science,53(5), 683–697. https://doi.org/10.1029/2017rs006428 Liu, J., Zhang, D., Mo, X., Xiong, C., Hao, Y., & Xiao, Z. (2020). Morphological differences of the northern equatorial ionization anomaly between the eastern Asian and American sectors. Journal of Geophysical Research: Space Physics,125(3), e2019JA027506. https://doi.org/10. 1029/2019ja027506 Loutfi, A., Pitout, F., Bounhir, A., Benkhaldoun, Z., Makela, J. J., Abamni, S., et al. (2022). Interhemispheric asymmetry of the equatorial ionization anomaly (EIA) on the African sector over 3 years (2014–2016): Effects of thermospheric meridional winds. Journal of Geophysical Research: Space Physics,127(9), e2021JA029902. https://doi.org/10.1029/2021ja029902 Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 18 of 19 Ma, G., Fan, J., Wan, Q., & Li, J. (2022). Error characteristics of GNSS derived TEC. Atmosphere,13(2), 237. https://doi.org/10.3390/ atmos13020237 Mannucci, A. J., Wilson, B. D., Yuan, D. N., Ho, C. H., Lindqwister, U. J., & Runge, T. F. (1998). A global mapping technique for GPS‐Derived ionospheric total electron content measurements. Radio Science,33(3), 565–582. https://doi.org/10.1029/97rs02707 Marini‐Pereira, L., Pullen, S., de Oliveira Moraes, A., & Sousassantos, J. (2021). Ground‐based augmentation system operation in low latitudes‐ part 2: Space weather, ionospheric behavior and challenges. Journal of Aerospace Technology and Management,13, e4821. https://doi.org/10. 1590/jatm.v13.1237 Mengistu Tsidu, G., & Melaku Zegeye, M. (2020). Comparison of quiet‐time ionospheric total electron content from the IRI‐2016 model and from gridded and station‐level GPS observations. Annales Geophysicae,38(3), 725–748. https://doi.org/10.5194/angeo-38-725-2020 Migoya‐Orué, Y., Folarin‐Olufunmilayo, O., Radicella, S., Alazo‐Cuartas, K., & Rabiu, A. B. (2017). Evaluation of NeQuick as a model to characterize the equatorial ionization anomaly over Africa using data ingestion. Advances in Space Research,60(8), 1732–1738. https://doi. org/10.1016/j.asr.2017.01.013 Migoya‐Orué, Y., Nava, B., Radicella, S., & Alazo‐Cuartas, K. (2015). GNSS derived TEC data ingestion into IRI 2012. Advances in Space Research,55(8), 1994–2002. https://doi.org/10.1016/j.asr.2014.12.033 Mitra, S. K. (1946). Geomagnetic control of region F 2 of the ionosphere. Nature,158(4019), 668–669. https://doi.org/10.1038/158668a0 Nava, B., Coisson, P., & Radicella, S. M. (2008). A new version of the NeQuick ionosphere electron density model. Journal of Atmospheric and Solar‐Terrestrial Physics,70(15), 1856–1862. https://doi.org/10.1016/j.jastp.2008.01.015 Nava, B., Radicella, S. M., & Azpilicueta, F. (2011). Data ingestion into NeQuick 2. Radio Science,46(6), 1–8. https://doi.org/10.1029/ 2010rs004635 Nava, B., Radicella, S. M., Leitinger, R., & Coısson, P. (2006). A near‐real‐time model‐assisted ionosphere electron density retrieval method. Radio Science,41(6), 1–8. https://doi.org/10.1029/2005rs003386 Nava, B., Rodrıguez‐Zuluaga, J., Alazo‐Cuartas, K., Kashcheyev, A., Migoya‐Orué, Y., Radicella, S. M., et al. (2016). Middle‐and low‐latitude ionosphere response to 2015 st. Patrick's day geomagnetic storm. Journal of Geophysical Research: Space Physics,121(4), 3421–3438. https:// doi.org/10.1002/2015ja022299 Nigussie, M., Jakowski, N., & Hoque, M. (2022). Characterization and climatological modeling of equatorial ionization anomaly (EIA) crest position. Journal of Geophysical Research: Space Physics,127(12), e2022JA030798. https://doi.org/10.1029/2022ja030798 Nigussie, M., Radicella, S. M., Damtie, B., Nava, B., Yizengaw, E., & Ciraolo, L. (2012). TEC ingestion into NeQuick 2 to model the East African equatorial ionosphere. Radio Science,47(5), 1–11. https://doi.org/10.1029/2012rs004981 Osanyin, T. (2025). Python code for GNSS station selection [Dataset]. Zenodo.https://doi.org/10.5281/zenodo.16060313 Osanyin, T. O., Candido, C. M. N., Becker‐Guedes, F., Migoya‐Orue, Y., Bosco Habarulema, J., Obafaye, A. A., et al. (2023). Performance of a locally adapted NeQuick‐2 model during high solar activity over the Brazilian equatorial and low‐latitude region. Advances in Space Research, 72(12), 5520–5538. https://doi.org/10.1016/j.asr.2023.07.018 Pignalberi, A., Habarulema, J. B., Pezzopane, M., & Rizzi, R. (2019). On the development of a method for updating an empirical climatological ionospheric model by means of assimilated vTEC measurements from a GNSS receiver network. Space Weather,17(7), 1131–1164. https://doi. org/10.1029/2019sw002185 Pignalberi, A., Pezzopane, M., Nava, B., & Coısson, P. (2020). On the link between the topside ionospheric effective scale height and the plasma ambipolar diffusion, theory and preliminary results. Scientific Reports,10(1), 17541. https://doi.org/10.1038/s41598-020-73886-4 Radicella, S. M., & Nava, B. (2020). Empirical ionospheric models. In The dynamical ionosphere (pp. 39–53). Elsevier. Rama Rao, P. V., Gopi Krishna, S., Niranjan, K., & Prasad, D. S. (2006). Study of spatial and temporal characteristics of L‐band scintillations over the Indian low‐latitude region and their possible effects on GPS navigation. Annales Geophysicae,24(6), 1567–1580. https://doi.org/10.5194/ angeo-24-1567-2006 Rao, K. N. (2007). GAGAN‐the Indian satellite based augmentation system. 84.40. Ua. Reinisch, B. W., Huang, X., Galkin, I. A., Paznukhov, V., & Kozlov, A. (2005). Recent advances in real‐time analysis of ionograms and ionospheric drift measurements with digisondes. Journal of Atmospheric and Solar‐Terrestrial Physics,67(12), 1054–1062. https://doi.org/10. 1016/j.jastp.2005.01.009 Ridley, A. J., Deng, Y., & Toth, G. (2006). The global ionosphere–thermosphere model. Journal of Atmospheric and Solar‐Terrestrial Physics, 68(8), 839–864. https://doi.org/10.1016/j.jastp.2006.01.008 Romero‐Hernandez, E., Denardini, C. M., Jonah, O. F., Essien, P., Picanço, G. A., Nogueira, P., et al. (2020). Nighttime ionospheric TEC study over Latin America during moderate and high solar activity. Journal of Geophysical Research: Space Physics,125(10), e2020JA028210. https://doi.org/10.1029/2020JA028210 Seemala, G. K., & Valladares, C. E. (2011). Statistics of total electron content depletions observed over the South American continent for the year 2008. Radio Science,46(5), 1–14. https://doi.org/10.1029/2011rs004722 Stolle, C., Manoj, C., Lu  hr, H., Maus, S., & Alken, P. (2008). Estimating the daytime equatorial ionization anomalystrength from electric field proxies. Journal of Geophysical Research,113, A09310. https://doi.org/10.1029/2007JA012781 Takahashi, H., Costa, S., Otsuka, Y., Shiokawa, K., Monico, J. F., Paula, E., et al. (2014). Diagnostics of equatorial and low latitude ionosphere by TEC mapping over Brazil. Advances in Space Research,54(3), 385–394. https://doi.org/10.1016/j.asr.2014.01.032 Takahashi, H., Wrasse, C. M., Denardini, C. M., Pádua, M. B., de Paula, E. R., Costa, S. M., et al. (2016). Ionospheric TEC weather map over South America. Space Weather,14(11), 937–949. https://doi.org/10.1002/2016sw001474 Tan, Y., Hu, Q., Wang, Z., & Zhong, Q. (2018). Geomagnetic index Kp forecasting with LSTM. Space Weather,16(4), 406–416. https://doi.org/ 10.1002/2017sw001764 Zhang, Y., & Paxton, L. J. (2021). Space physics and aeronomy, ionosphere dynamics and applications, geophysical monograph 260 (1st ed.). John Wiley and Sons. Zhao, J., & Han, Y. B. (2008). Historical dataset reconstruction and a prediction method of Solar10. 7cm radio flux. Chinese Journal of Astronomy and Astrophysics,8(4), 472–476. https://doi.org/10.1088/1009-9271/8/4/11 Space Weather 10.1029/2024SW004212 OSANYIN ET AL. 19 of 19