First Look at a Geomagnetic Storm With Santa Maria Digisonde Data: F Region Responses and Comparisons Over the American Sector

Santa Maria Digisonde data are used for the first time to investigate the F region behavior during a geomagnetic storm. The August 25, 2018 storm is considered complex due to the incidence of two Interplanetary Coronal Mass Ejections and a High‐Speed Solar Wind Stream (HSS). The F 2 layer critical frequency (f o F 2) and its peak height (h m F 2) collected over Santa Maria, near the center of the South American Magnetic Anomaly (SAMA), are compared with data collected from Digisondes installed in the Northern (NH) and Southern (SH) Hemispheres in the American sector. The deviation of f o F 2 (Df o F 2) and h m F 2 (Dh m F 2) are used to quantify the ionospheric storm effects. Different F region responses were observed during the main phase (August 25–26), which is attributed to the traveling ionospheric disturbances and disturbed eastward electric field during nighttime. The F region responses became highly asymmetric between the NH and SH at the early recovery phase (RP, August 26) due to a combination of physical mechanisms. The observed asymmetries are interpreted as caused by modifications in the thermospheric composition and a rapid electrodynamic mechanism. The persistent enhanced thermospheric [O]/[N2] ratio observed from August 27 to 29 combined with the increased solar wind speed induced by the HSS and IMF B z fluctuations seem to be effective in causing the positive ionospheric storm effects and the shift of the Equatorial Ionization Anomaly crest to higher than typical latitudes. Consequently, the most dramatic positive ionospheric storm during the RP occurred over Santa Maria (∼120%).

of eastward polarity can uplift the equatorial ionosphere, and consequently, there are intensifications of the Equatorial Plasma Fountain (EPF) and the poleward expansion of the Equatorial Ionization Anomaly (EIA) crests (Balan et al., 2017;Balan & Bailey, 1995;Fagundes et al., 2016;Sobral et al., 2001;Tsurutani et al., 2004). The PPEF also causes the intensification in the Pre-Reversal Enhancement (PRE) vertical drift. The sudden increment in the layer height may give rise to the Rayleigh-Taylor instability and the development of spread-F irregularities (Abdu et al., 2008;Li et al., 2010). If the IMF B z component suddenly changes direction and turns northward, the "overshielding" condition takes place, and the region-2 field-aligned current over shields the region-1 current from now-weakened convection. As a result, the dawn-to-dusk electric field is oriented in dusk-to-dawn direction in the inner magnetosphere (Kelley et al., 1979). The overshielding westward electric field at nighttime can also be intense to cause a suppression of the PRE and spread-F irregularities (Abdu et al., 2009). The second type of disturbing electric field is called "disturbance wind dynamo electric field" (DDEF; Blanc & Richmond, 1980). During the dayside (nightside) of the Earth, the DDEF is westward (eastward), i.e., contrary to the ordinary ionospheric dynamo. The study of DDEF is also important because it may cause the (total or partial) suppression of the EIA in the daytime hours during the recovery phase (RP) of geomagnetic storms (Balan et al., 2013). On the other hand, the eastward polarity of DDEF during nighttime hours can uplift the F layer in the equatorial region and enhance the electron density in the EIA crests region (15º-20º in the north and the south geomagnetic latitude). Joule dissipation at high latitudes also modifies the thermospheric concentration. The electron density is governed by the photoionization of atomic oxygen density [O] and loss with molecular nitrogen [N 2 ] and oxygen [O 2 ]. Thus, the F region electron density is directly proportional to the thermospheric [O]/[N 2 ] ratio when the transport is not important (Danilov, 2013;Yu et al., 2020).
The previous competing drivers (ionospheric disturbed electric fields and the thermospheric composition) usually cause global modifications in the F 2 layer critical frequency (f o F 2 , representing the peak electron density N m F 2 ) and its peak height (h m F 2 ) relative to some quiet-time background level. The increase of the f o F 2 concerning the quiet days is widely known as positive ionospheric storm. It is explained in terms of several physical mechanisms, among others, the increase in the atomic oxygen, disturbed electric fields, and Traveling Atmospheric/Ionospheric Disturbances (TAD/TID) (Fuller-Rowell et al., 1996;Huang et al., 2005). Contrarily, a decrease of f o F 2 during storm-time is known as negative ionospheric storm and it may be caused by a decrease in the thermospheric [O]/[N 2 ] ratio (Danilov, 2013;Prölss, 1995;Zhang et al., 2003).
Some review articles on ionospheric storms (i.e., Prölss, 1995;Danilov, 2013, and reference therein) have reported examples of serious problems in satellite communication and navigation during severe positive storms effects. On the other hand, severe negative storms can cause serious radio block out in ground-based high-frequency communication. Therefore, more than ever, a robust understanding of the positive and negative ionospheric storms is important in the anomalously low geomagnetic field intensity region currently located in South America known as South American Magnetic Anomaly (SAMA; Abdu et al., 2005;Kurnosova et al., 1962). In addition to positive/negative phases observed in the electron density that may cause problems in satellites, the large high-energy particles that penetrate in the ionosphere may cause problems in satellite sensors (Schuch et al., 2019). It was also reported that the operator who controls satellites in low-Earth orbit may need to turn sensors off to reduce the detector saturation when passing through the SAMA (Jones et al., 2017). The energetic particles also enhance the ionization distribution and conductivities in the SAMA during geomagnetic storms (Moro et al., 2012;2013).
As can be noted, the ionosphere-thermosphere (I-T) response in the SAMA region is an important research topic in space weather, especially during storm-time. It can be understood better by studying individual cases since any geomagnetic storm has its specific characteristics and, therefore, each case needs an individual approach (Astafyeva et al., 2020;Blagoveshchensky & Sergeeva, 2020). Although many case studies, statistics, and model simulations have reported the effects of geomagnetic storms in the I-T system everywhere in the globe, the drivers acting in changing the electron density in the SAMA are a poorly explored research topic. The data collected from a Digisonde installed in Santa Maria, near the SAMA center, permit now these studies that are otherwise unachieved. The selected disturbed period is the first one in which Santa Maria Digisonde data are available in all phases of the storm. The August 25, 2018 geomagnetic storm is considered very complex by several authors (Abunin et al., 2020;Astafyeva et al., 2020) since it was caused by two consecutive Interplanetary Earth-directed Coronal Mass Ejections (ICMEs) and a followed High-Speed Solar Wind Stream (HSS).
Thus, this work aims to study the F region responses over Santa Maria during the August 25, 2018 geomagnetic storm, the third most intense of the Solar Cycle 24. The geographical and geomagnetical extension of the drivers responsible for the storm-time F region behavior is analyzed by comparing ionospheric data acquired from digital ionosondes installed outside the influence of SAMA in the Northern (NH) and Southern (SH) Hemispheres within the American sector. Figure 1 shows the Digisonde (DPS-4D) map with the stations used in this work with respect to their position to the SAMA. The red lines in the map are the iso-intensity lines of the total geomagnetic field provided by the World Magnetic Model (WMM; Chulliat et al., 2020). The line with 23,000 nT defines the position of the anomaly center. It should be noted that out of seven stations shown on the map, Santa Maria and Cachoeira Paulista are located inside the iso-line of 23,000 nT and very close to it, respectively. The shape of the SAMA is asymmetric (Zou et al., 2015) and covers most of South America, from West Africa to the Pacific Ocean, and from the equator to Antarctica. Its center is located nowadays in Paraguay.

MORO ET AL.
10.1029/2020JA028663 3 of 18 Figure 1. Global map with the geomagnetic field total intensity. The South American Magnetic Anomaly (SAMA) position is defined by the lowest isointensity line (23,000 nT) currently located in South America. The contour interval is 1,000 nT. The map also shows the different ionosonde stations used in this work. Source: Modified from Chulliat et al. (2020). In Santa Maria, ionospheric sounding using Digisonde began in April 2017. The variability of f o F 2 , h m F 2 , E region critical frequency (f o E), and the thickness parameter B0 was reported by Moro et al. (2019). In a recent work, Moro et al. (2020) compared the observed f o F 2 , h m F 2 , and f o E collected during geomagnetically quiet days with the International Reference Ionosphere (IRI) outputs. The Digisondes installed at São Luís and Cachoeira Paulista belong to the Embrace Digisonde Network (DigiNet; Denardini et al., 2016). All necessary information about the Digisondes relevant to this work is shown in Table 1.
The Digisonde stations are distributed from middle, low, and equatorial latitude regions in the NH (Millstone Hill and Ramey) and the SH (São Luís, Jicamarca, Cachoeira Paulista, Santa Maria, and Port Stanley). As mentioned by Denardini et al. (2016), there is no exact definition where the boundaries of these regions are located. Therefore, in this work, it means that 5°N ≤ geomagnetic latitude ≤ 5°S is an equatorial region (Jicamarca), 5° < geomagnetic latitude < 20° is low latitude (São Luís and Cachoeira Paulita), and 20° < geomagnetic latitude < 50º is middle latitude (Santa Maria, Port Stanley, Millstone Hill, and Ramey). It is seen that São Luís and Santa Maria stations are currently located in a transition region from equatorial to low, and low to middle latitude region, respectively, due to the northwestward movement of the geomagnetic equator (Moro et al., 2017;Resende et al., 2016).
The Digisonde data came from the GIRO Digital Ionogram DataBase (DIDBase; http://umlcar.uml.edu/, Reinisch & Galkin, 2011). The ionospheric effects in the F region are investigated using the f o F 2 and h m F 2 parameters. The relative deviations of these ionospheric parameters are used to quantify the electronic density and height variations over each station. The relative deviation of the parameter The day-to-day variability is removed with the mean quiet level (f o F 2 (q)) obtained using the data acquired during the five International Quiet Days (IQDs) of the month. The range of ±20% in Df o F 2 is used by several authors (Habarulema et al., 2020;Matamba et al., 2015, and reference therein) to represent the day-to-day variability, and the deviations outside this range are attributed to the ionospheric disturbances during the geomagnetic storms. The five IQDs used in this work are August 6, 10, 13, 14, and 23, 2018, except for Port Stanley which used September 1, 3, 8, 19, and 20, 2018 due to the lack of measurements. Positive Df o F 2 values correspond to the existence of positive ionospheric storms. A similar relation was considered for the relative deviation of the peak height h m F 2 , Dh m F 2 .
It is also used in this work the thermospheric column-integrated [O]/[N 2 ] from the Global Ultraviolet Imager (GUVI) spectrograph of the TIMED satellite (http://guvitimed.jhuapl.edu/; Christensen et al., 2003;Paxton et al., 2005). The effects of the solar wind plasma and the IMF in the Earth space environment from August 25 to 29, 2018 are analyzed with observations of the Advanced Composition Explorer (ACE) satellite in the Lagrange point L1. The data are downloaded from the OMNI Website all at a 1-minute resolution (http://omniweb.gsfc.nasa.gov/). The convective IEF east-west E y component (i.e., the dawn-to-dusk component) is derived using the following expression (in geocentric solar magnetospheric coordinates):

Description of the Complex Interplanetary and Geomagnetic Conditions from August 25 to 29, 2018
An unexpected and unusual strong geomagnetic storm (G3 class storm accordingly to the NOAA Space Weather Prediction Center) occurred on August 25, 2018. The analysis performed by Abunin et al. (2020) revealed that the storm was initiated by a two-step long filament eruption in the central sector of the solar disk on August 20. It was accompanied by two consecutive ICMEs of a rather small size and low speed and followed by the HSS. The interaction of this first solar wind structure created a strong compression of the Earth's magnetosphere, in which the magnetopause standoff distance (R mp ) calculated by an empirical model (Shue et al., 1998) reached ∼7 Earth radius (R E = 6,371 km). It means that the outer and inner magnetosphere was considerably perturbed during the influences of these ICMEs (Alves et al., 2016), and, consequently, the ionosphere may have been suffered influences of this magnetopause compression.
The solar wind parameters measured by the instruments onboard the ACE satellite are shown from top to bottom in Figures 2a-2d. The R mp obtained with the model developed by Shue et al. (1998) is shown in Figure 2a (olive line). The estimated dawn to dusk IEF E y component is shown in Figure 2e. The SYM-H and ASY-H (blue line) indexes are shown in Figure 2f. The vertical dotted lines identified as IP, MP, and RP mark the initial phase, the main phase, and the recovery phase, respectively.
The R&C catalog reports that the ICME shock arrived at Earth at 2:00 UT on August 20, 2018, when the solar wind pressure P SW (Figure 2a August reflects the complex structure. New and strong intensifications were observed in the IMF B z at 9:57 UT (3.18 nT), 13:09 UT (16.07 nT), 18:12 UT (10.33 nT), and 20:28 UT (4.32 nT) identified as 1, 2, 3, and 4, respectively. Similar to the IMF B z , the peaks also appeared in IEF E y . The auroral electrojet became more intense as represented by several peaks in the ASY-H index. The SYM-H also exhibited two peaks (2, 3 in Figure 2f) marking the first changes in the recovery slope. These peaks coincide with the Peaks 2 and 3 of the IMF B z .
During the influences of the HSS, the V SW increased (from ∼380 km/s to ∼580 km/s) between 7:00 UT on August 26 and 6:00 UT on August 27, followed by a second enhancement in which V SW plateaued around 630 km/s at 17:00 UT on August 27. After that, V SW started to decrease until it achieved ∼450 km/s on August 29. IMF features observed from August 27 suggest the presence of the Alfvénic fluctuations, as B x , B y MORO ET AL.     Table 1 for specific sunrise and sunset hours on August 25, 2018). The red line represents the parameter during storm-time. The blue line is the averaged quiet days, in which error bars are plotted only hourly to avoid overcrowding. The comparison between the storm-time f o F 2 and the averaged quiet days (Figure 3a) is particularly useful to show the manifestation of positive/negative storms. On the other hand, the variation of h m F 2 with respect to the reference level ( Figure 3b) is used to identify the possible action of an electrodynamic source (disturbed electric fields) acting in uplift or decrease the F layer in relation to the ground. It is also noticeable that there is no evening peak in the reference h m F 2 over Jicamarca. This behavior was reported before by Lee et al. (2008). The presence of spread-F in the ionograms are marked by horizontal orange bars, whereas the black ones show no data collected by the Digisonde.

Results and Discussions
The f o F 2 and h m F 2 for the remaining stations are presented in Figures 4 and 5, respectively. These figures follow the same pattern described in Figure 3 and are arranged from top to bottom in decreasing order of geomagnetic latitude. A first look at these results reveals similarities and differences in the American sector. Their quantitative descriptions are presented in terms of the relative deviations Df o F 2 (orange line) and Dh m F 2 (green line) for Jicamarca in Figure 3c, and in remaining stations in Figure 6.
Regarding the experimental data (red lines), one can notice gaps in the curves.  (associated with the relatively low resolution-30 min-set in the Digisonde); over Jicamarca (−2.4º) and São Luís (5.8º), f o F 2 was difficult to determine in some nighttime hours due to the presence of spread-F; over Cachoeira Paulista (−14.0º) and Santa Maria (−20.5º) the gaps before sunrise are caused by the decrease in F region electron density when the Digisonde could not detect the signals; and the presence of intense Es layers observed (not shown here) in all stations during the storm blocked the signals from the upper ionosphere in some hours.
The ICME shock at 2:00 UT on August 25 did not cause a sudden geomagnetic impulse (Figure 2f)  (Df o F 2 ∼60% at 2:30 UT, Figure 4c), and caused a slight rise in h m F 2 (Figures 3b and 5c) in both stations. When the ICME material arrived and compressed the magnetosphere at 12:00 UT, there was an increase of ∼100 km in h m F 2 over Jicamarca, which increased ∼40% the Dh m F 2 (Figure 3c) concerning the reference. Interestingly, the response over São Luís was seen as a decrease of Dh m F 2 of ∼20% ( Figure 5c).
Further, at the MP (16:00 UT, dayside along 45°W to 76.8°W meridians) when the IMF B z turned southward, the immediate response of the h m F 2 over Jicamarca (Figure 3b) is noted as a fast oscillation. The F 2 layer height increased from 330 km (at 16:00 UT) to 360 km (16:15 UT). In the next ionogram collected at Jicamarca, h m F 2 decreased to 320 km (16:30 UT). The rapid increase at 16:15 UT is a consequence of an undershielding PPEF of eastward polarity. However, h m F 2 data from São Luís (Figure 5c) did not show enhancements at this time, which would be a characteristic of the PPEF during daytime hours. It means that the PPEF was weak, which may be caused by the unusual gradual beginning of this storm as seen by the smoothed variation of the SYM-H index (Figure 2e). This behavior is in line with the difference (dH) MORO ET AL.

Astafyeva et al. (2020) from the vertical total electron content (VTEC) data estimated by the Global Position-
ing System (GPS) receiver onboard the Swarm satellite. The authors reported the occurrence of ∼5 TEC unit (TECU; 1 TECU = 10 16 electrons/m 2 ) positive storm around 60°N at ∼20:30 UT. Also, they observed similar effects during the next overflies of the satellites around 60ºS at midnight. Nevertheless, such effects were not observed over Port Stanley (Figure 4f).
The TIDs/TADs are one of the physical mechanisms that cause a positive ionospheric storm, as discussed by Prölls (1995) and Danilov (2013). Therefore, the TIDs generated during the storm may have propagated to equatorial latitudes promoting the positive effects in the other stations after ∼22:00 UT on August 25. However, the long-last positive effects over Ramey, São Luís, Cachoeira Paulista, and Santa Maria until ∼6:00 UT on August 26 is not expected to be caused by TIDs during nighttime, i.e., the TIDs may not be the main driver causing these positive effects over these stations. Astafyeva et al. (2020) showed signatures of negative effects at 5:00 UT. As can be seen in Figure 4 from Digisonde data, only positive effects were observed over Ramey, São Luís, Cachoeira Paulista, and Santa Maria. The results from the Port Stanley Digisonde between 1:00 UT and 2:00 UT also indicate the positive effects. The difference between the Digisonde and satellite results from Astafyeva et al. (2020) can be explained considering the variations of h m F 2 over each station. Note that the positive effects observed with Digisonde data were below 460 km altitude, i.e., below the Swarm altitude.
A tentative explanation of the physical mechanism that caused the positive ionospheric storm in nighttime hours on August 26 is given in terms of h m F 2 variations over the equatorial and low-latitude stations. The increase of h m F 2 to > 400 km (Dh m F 2 ∼50%) over Jicamarca (Figure 3b) after midnight indicates that a disturbing eastward electric field is the main driver acting. This rapid increase is a response to an overshielding condition, which caused the uplift of the It is also noticeable the large height increases/fluctuations over Santa Maria (Figure 5e) and Cachoeira Paulista ( Figure 5d) at 2:00 UT, 4:00 UT, and 6:00 UT compared to São Luís (Figure 5c). This response in h m F 2 seems to be caused by strong equatorward wind. Similar behavior was reported by Abdu et al. (2007) on 28 October 2013 geomagnetic storm, when the h m F 2 fluctuations/oscillations over Cachoeira Paulista were not observed over São Luís. In the same work, Abdu et al. (2007) observed the development of the EIA due to both conditions of PPEF in the Brazilian sector, similar to what is observed over Cachoeira Paulista and Santa Maria in this work as an enhancement in electron density during these nighttime hours.
The F region responses in the American sector became highly asymmetric between the NH and SH during the RP, which started at 7:00 UT on August 26. As seen from Figure 4a, f o F 2 presented a salient negative phase at Millstone Hill after sunrise. During daytime hours, f o F 2 remained below 4 MHz between 10:00 UT and 23:30 UT. The layer substantially decreased to 140 km at 16:30 UT on August 26 (Figure 5a). The lowest value of Dh m F 2 equal to −30% occurred at 18:30 UT (Figure 6a). The f o F 2 did recover to the undisturbed mean values after 10:00 UT on August 27. The negative effect was also observed over Ramey, and persisted for more than 20 h, from 10:30 UT on August 26 to 6:00 UT on August 27. During this period, f o F 2 stayed below 5 MHz most of the time, as shown in Figure 4b, despite the significant gaps caused by the strong Es layers occurrences.
According to the current understanding, the negative ionospheric storm arises from changes in the thermospheric composition generated during storm time at auroral latitudes that increase the N 2 with a reduction in O. The ionospheric and thermospheric effects on August 26 using GUVI/TIMED data were presented by Astafyeva et al. (2020). They show that the satellite flew over the American sector on August 26 between  Regarding the stations located in the SH on August 26, it is observed only positive ionospheric storm. The increase of the f o F 2 concerning the quiet days is explained in part by the enhanced thermospheric [O]/[N 2 ] ratio ( Figure 7c). However, the main feature observed over São Luís, Cachoeira Paulista, and Santa Maria during daytime is four peaks in f o F 2 that may be caused by a rapid mechanism such as an electrodynamic one. As shown in Figure 2d, these peaks appear in phase with successive intensifications in the IMF B z and IEF E y (marked with 1, 2, 3, and 4 in Figure 2d). These peaks might occur in f o F 2 over Port Stanley, but with substantially lower amplitudes (the peaks are difficult to detect due to the lack of measurements). Note also the several oscillations in the ASY-H index in Figure 2f. Huang et al. (2005) reported that fluctuations in the AE index under IMF B z south conditions cause PPEF (see also Abdu et al., 2007). Interestingly, h m F 2 values over Cachoeira Paulista and Santa Maria also exhibited similar peaks as observed in f o F 2 but ∼1 h before. Considering a competition between these two drivers (thermospheric composition shown in Figure 7 and disturbed electric fields), the decrease of [O]/[N 2 ] in the NH causing the negative effects was more effective than the disturbing electric fields (no peaks in f o F 2 are observed). To sum up, the positive storm effects observed over the stations in the SH on August 26 seem to be occurred by the combined effect of the enhanced thermosphere [O]/[N 2 ] ratio and the PPEF process (which dominated over the expected DDEF effects). It is also remarkable that in terms of Df o F 2 , the highest peaks occurred over Santa Maria, achieving 120% at 13:30 UT, 112% at 16:30 UT, and 95% at 20:15 UT.
On the following days of the RP, it would be expected a control of the thermospheric compositions and DDEF in the F region responses causing negative storm effects, as observed in the case presented by Abdu et al. (2007) or Balan et al. (2013). However, f o F 2 was higher than the quiet time pattern over all stations except Millstone Hill (which recovery faster), i.e., in general, the plasma density was larger in the EIA regions, and it was roughly hemispherically symmetric. Interestingly, Santa Maria and Cachoeira Paulista presented remarkably f o F 2 enhancements at 14:00-20:00 UT on August 27-29, whereas the f o F 2 over São Luís did not show significant changes. The Df o F 2 achieved 107% at 17:05 UT over Santa Maria (Figure 6e) and 87% at 18:30 UT over Cachoeira Paulista (Figure 6d) on August 27. On the same day, Df o F 2 achieved 70% at 19:25 UT over Ramey (Figure 6b) and 50% at 15:00 UT over Port Stanley (Figure 6f). In the next two days, August 28 and 29, positive effects are still clearly noticed over Santa Maria, Cachoeira Paulista, and less evident over Ramey and Port Stanley. Moreover, note that the enhanced thermospheric [O]/[N 2 ] ratio is still sustained over South America during the RP, as shown in the maps of Figures 7d-7f. As the storm progressed, the neutral composition changes extended to higher latitudes.
To illustrate the enhancement of the plasma density in South America during the geomagnetic storm, it is shown a set of TEC maps from 14:30 UT to 17:00 UT in Figure 8. The curved solid line in each map represents the geomagnetic equator. The methodology applied to generate these TEC maps can be found in detail in Takahashi et al. (2016). These maps have a spatial resolution in latitude and longitude of 0.5º × 0.5°. The TEC maps during the reference quiet day (August 6, 2018-08-06) are shown in the first row of Figure 8. During the reference day, the EIA developed at 14:30 UT, as seen on the eastern coast (around 30º-40ºW) of Brazil. In the following maps of the first row, the plasma density is enhanced in the equatorial and low latitude regions in Brazil (0º-15ºS), which correspond to the southern EIA crest. The most intense patches of TEC are identified around 15ºS at 16:00 UT. Then, the TEC decreased until around 22:30 UT (not shown here). Indeed, this reference quiet day represents well the average daily behavior of the EIA during winter in the SH, as discussed in Takahashi et al. (2016).
The EIA became stronger during the RP (2018-08-27, 2018-08-28, 2018-08-29), as observed by the second, third, and fourth row of Figure 8, respectively, as compared to the reference quiet day (first row). The largest enhancement of plasma density is observed all over the Brazilian territory in the maps correspondent to 27 August. Regions with TEC > 30 are very clear between 30º and 60ºW in the maps after 15:30 UT achieving a wider area. It is observed TEC ≥ 25 over Santa Maria (∼30ºS) and Cachoeira Paulista (14º S) after 15:30 UT on August 27, which means a strong development of the southern EIA crest farther from the magnetic equator as compared to the reference quiet day. Small enhancement in plasma density is also observed over Port Stanley (∼42ºS). The TEC slightly decreased on the following days, August 28-29, but it is still intense compared with the geomagnetically quiet day (August 6). It is still possible to observe large enhancements of plasma density (TEC > 30) from 15:00 UT to 16:30 UT on August 28, and from 15:30 UT to 16:00 UT on August 29. Overall, the maps show no suppression/weakening of dayside EIA, which indicates no DDEF effects on TEC (and Digisonde) data during the RP (August 27-29, 2018) of the geomagnetic storm. Lissa et al. (2020) studied the ionosphere responses on August 25, 2018 geomagnetic storm using TEC data from GPS stations distributes over India, Sri Lanka, China, Taiwan, and the Philippines. The authors also observed large positive effects in the daytime hours for three consecutive days from August 26 to 29. They explained the drivers of the positive effects in terms of the recombination process, plasma diffusion caused by disturbed winds and enhanced thermospheric [O]/[N 2 ] ratio over the considered longitudes (80ºE-120ºE). In addition to these drivers, it is also suggested in the present work that the HSS had the potential to play a key role in causing the positive effects during the RP. The daytime intensification of the EIA may be caused by PPEF of eastward polarity, which in turn, depends on the IMF B z component. Wei et al. (2008) explained that the PPEF can be classified into "single penetration" and "multiple penetrations". The former is characterized by a southward or northward abrupt reversal of the IMF B z component, and the latter is the oscillation between northward and southward directions. shown in Figure 2d could excite discontinuous magnetic reconnection. As shown before, IEF E y component is controlled by solar wind speed as E y = (V x )(B z ), which peaked 630 km/s at 17:00 UT on August 27 (Figure 2c). Therefore, it is suggested that the high V SW effectively contributed to the enhancement of the PPEF (eastward polarity). As a result, it was observed positive ionospheric effects centered at around ±15°-±30° magnetic latitudes (note that there is no station at +15º magnetic latitude) due to the PPEF. It explains the higher Df o F 2 peaks observed over Santa Maria as compared to Cachoeira Paulista during the RP.
In a recent work, Ren et al. (2020) used TEC data collected by the Beidou geostationary satellites to study the impact of the HSS in the ionosphere during August 27-29, 2018. Using multiple observations and the Thermosphere-ionosphere-electrodynamics general circulation model (TIEGCM), the authors suggested that the HSS modulates the low latitude ionosphere during the RP. The TIEGCM could well reproduce the positive ionospheric storm during the RP, although it underestimated the TEC enhancements. Denton et al. (2009) reported that during the passage of a typical HSS over the magnetosphere, it induces several phenomena. Therefore, it is important to understand the impact of this HSS to identify the main dynamic mechanisms responsible for the long-last intense positive ionospheric effects in Digisonde data over Santa Maria and Cachoeira Paulista on August 27-29. Thereby, a close inspection from August 27 in the IMF features through the linear correlation calculated between the variation of the solar wind velocity components at L1 location (ACE data) versus the corresponding Alfvén velocity components (figures not shown) confirms that the fluctuations are Alfvénic. Alfvénic fluctuations may also occur concomitantly with the southward IMF B z (Gonzalez et al., 1994) and can start the reconnection in the Earth's magnetosphere, which can trigger the main drives to moderate geomagnetic storms (Tsurutani et al., 1995), followed by a sequence of the substorms, as observed on August 27-29. Substorms are an important process for the injection of low energy electron particles in the inner magnetosphere (Jaynes et al., 2015). Substorms accompanied by the IMF B z and IMF Alfvénic fluctuations play a crucial role in generating Ultralow Frequency (ULF) waves in the magnetosphere, consequently, may occur wave-particle interaction . This dynamic in the magnetosphere observed during the influences of these Alfvénic fluctuations associated with HSS in the study agrees with the results discussed by Denton et al. (2009). As far as the author's know, it seems to be the first time that these results are discussed in this geomagnetic storm. Denton et al. (2009) also reported that disturbed f o F 2 and h m F 2 values last for at least 4 days in the middle latitudes. Therefore, the HSS on August 26 seems to be one of the factors that caused the long-last intense positive ionospheric storm effects in Digisonde data over Santa Maria and Cachoeira Paulista between August 27 and 29, and over Ramey (summer) in a less degree of intensity. Santa Maria and Cachoeira Paulista stations are in the winter, which is well known that positive phases show a maximum occurrence. However, the effects reported here were quite dramatic due to the complex drivers acting during the geomagnetic storm.
The results presented in this work show that the F 2 layer electron density was significantly disturbed during the MP and (especially) in the RP. Moreover, the southern EIA crest achieved the Santa Maria region as seen in f o F 2 and confirmed in TEC data. It caused the most pronounced positive effects in the electron density with respect to all stations analyzed here. Despite being an intense geomagnetic storm, the EIA was not suppressed due to the DDEF process during the RP. Finally, this work shows how unpredictable is the storm-time ionosphere, especially in the SAMA region, which is an additional difficulty imposed in the analysis of the I-T system responses and the development of models.

Conclusions
The data from the new Digisonde station in Santa Maria were used for the first time in this work to study a geomagnetic storm. The August 25-29, 2018 storm is the third most intense of the Solar Cycle 24 and is considered complex due to the interaction of the two consecutive ICMEs and followed by a HSS. The responses of the F 2 layer are studied considering the critical frequency f o F 2 and its peak height h m F 2 . The plasma parameters are compared with data from the Digisondes installed at Millstone Hill, Ramey, Jicamarca, São Luís, Cachoeira Paulista, and Port Stanley in the NH and SH in the American sector. The observed positive and negative ionospheric storm effects are quantified by the deviation of f o F 2 (Df o F 2 ) and h m F 2 (Dh m F 2 ) in percent.
The results show that the development of negative and positive phases had different durations, intensities, and onset. In general, the F region behavior displayed a remarkable asymmetry in the response of the geomagnetic disturbances during the RP. The main conclusions of the study are as follows: 1. At the beginning of the MP (16:00 UT on August 25), the increases in the F 2 layer height over Jicamarca and São Luís are a response to a weak undershielding PPEF of eastward polarity. Further, on August 25-26 during the MP, the positive ionospheric effects were observed in all stations and could be caused by several factors. During the daytime hours on August 25, the increase in electron density over Millstone Hill seems to be an effect of TIDs at higher latitudes, which may propagate equatorward. The effect of an overshielding process could be noticed after sunset as an increase of the h m F 2 in Jicamarca and São Luís. Consequently, the effects of positive ionospheric storms were observed in middle and low latitudes due to the development of the EIA crests. Interestingly, the overshielding during nighttime hours on August 26 occurred when IMF . Besides, several oscillations in f o F 2 were observed in the SH almost in phase with IMF B z (and IEF E y ) component increasing, even more, the positive phase, which resulted in higher deviations in Df o F 2 and Dh m F 2 . However, such oscillations were not observed over Millstone Hill and Ramey due to the strong decrease in electron density and its peak height. The interaction of the two mechanisms (changes in the thermospheric composition and electric field) had the potential to play a key role in supporting the large F 2 layer electron density on August 26 over the stations in the SH 3. The incidence of the HSS seems to modulate the Earth's magnetosphere and could disturb the middle latitude ionosphere in the SH (especially) for multiple days, from August 27 to 29 during the RP. A close analysis of the IMF features on August 27 through the linear correlation between the variation of the solar wind velocity components and the corresponding Alfvén velocity components confirm that the fluctuations are Alfvénic in nature. This is one important result since it helps to explain the intensification of the EIA crests over Santa Maria in terms of f o F 2 and TEC data during daytime hours from August 27 to 29 4. During the RP, the highest deviations were always observed over Santa Maria due to the changes in the southern EIA crest position Finally, the results show how unpredictable is the storm-time ionosphere over Santa Maria and, in general, in the wide region close to the SAMA center, which presents an additional complication in understanding the I-T responses and the development of models in a region with an anomalously weak geomagnetic field.

Data Availability Statement
The GUVI data used here (http://guvitimed.jhuapl.edu/) are provided through support from the NASA MO&DA program. The Santa Maria Digisonde data can be also downloaded upon registration at the Embrace webpage from INPE Space Weather Program (http://www2.inpe.br/climaespacial/portal/en). The Digisonde data are provided by GIRODidbase (http://umlcar.uml.edu/).