The Impact of the Disturbed Electric Field in the Sporadic E (Es) Layer Development Over Brazilian Region

During disturbed periods, E region electric fields can cause anomalous Es layer behavior, which is observed in the digital ionosonde data. To investigate the influence of these electric fields in the Es layer development, we analyzed a set of 20 magnetic storms from 2015 to 2018 over Boa Vista (BV, 2.8°N, 60.7°W, dip ∼18°), São Luís (SLZ, 2.3°S, 44.2°W, dip ∼8°), and Cachoeira Paulista (CXP, 22.41°S, 45°W, dip ∼35°). The electric field zonal components during the main and recovery phases of each magnetic storm are computed to study the corresponding characteristics of these Es seen in ionograms. Additionally, a numerical model (MIRE, Portuguese acronym for E Region Ionospheric Model) is used to analyze the Es layer dynamics modification around disturbed times. Using observation data and simulations, we were able to establish a threshold value for the electric field intensity for each region that can affect the Es layer formation. The results sustain that the strong Es layer in BV can be an indicator of the disturbed dynamo event. At SLZ, on the other hand, the Es layers are affected by the competition mechanisms of their formation, as equatorial electrojet irregularities and winds, during the main phase of the magnetic storm. Over CXP, the Es layer dynamics are dominated by the wind shear mechanism. Finally, this study provides new insights into the real impact of the electric field in the Es layer development over the Brazilian sector. Thus, our results lead to a better understanding of the underlying mechanisms related to the Es layer formation and dynamics.

a diffuse and non-blanketing Es trace named Es q (equatorial) layers (Chandra & Rastogi, 1975;Denardini et al., 2016;Resende et al., 2016;Moro et al., 2017). The signatures of the Es q layers appear as a scattering of the radio wave signal that covers most of the low-frequency scale and occurs very regularly when the polarization electric field and density gradient is well set at the E region heights (Resende et al., 2013). It is important to mention that in the vicinity of the magnetic equator, the wind shear mechanism is not effective due to the horizontal configuration of the magnetic field, not allowing the denser/blanketing layer formation in the E region.
The Es layer around the globe can suffer significant modifications due to the ionospheric electric field, mainly at locations near the geographic and magnetic equator. Resende et al. (2016) studied the competition between tidal winds and electric fields in the formation of blanketing sporadic E layers during quiet periods over São Luís, a region of transition from equatorial to low latitude due to an apparent northwestward movement of the magnetic equator. The authors showed that the blanketing sporadic E layers occur due to the vertical electric field weakening, caused by the departure of the magnetic equator from São Luís. Therefore, the tidal winds are more effective during some hours, forming denser Es layers. In other words, the vertical electric field component is responsible for Type II irregularity, and, consequently, for the Es q occurrence. Thus, when this electric field component is low, the wind shear mechanism becomes dominant. The same kind of analysis was performed by Moro et al. (2017) during disturbed periods, in which the authors found similar results about the relationship between the vertical electric field and blanketing Es layers. Hence, the blanketing Es layers occurrence at equatorial regions depends on the electric field vertical component. Abdu et al. (2003), Carrasco et al. (2007), and Abdu and Brum (2009) showed that the equatorial electric field during the evening pre-reversal enhancement (PRE) can cause Es layer intensification or disruption at low latitudes. The vertical electric field mapped from the equatorial F region to low latitudes can have some influence on the Es layer formation. These modifications in the PRE can be caused by an enhancement in the conductivity gradient near sunset due to the upward propagating planetary waves and tidal modes (Abdu & Brum, 2009;Abdu et al., 2006;Pancheva et al., 2003).
Recent studies have shown that the disturbed electric fields have some influence in the Es layer structures during geomagnetic storms. For instance, Abdu et al. (2014) reported on intensification or disruption of the Es layers associated with the Hall electric field induced by the prompt penetration electric fields (PPEFs) at low latitudes. More recently, Resende et al. (2020) detected strong Es layers in Boa Vista (BV, Geographic Coordinates: 2.8°N, 60.7°W, Magnetic Inclination: ∼18°), a station located near the geographic equator in the Brazilian sector, during the recovery phase of magnetic storms. They concluded that these anomalous Es layers are a consequence of the combined effect of the winds and disturbed electric fields. The authors concluded that the zonal westward electric field in the ionosphere due to the disturbance dynamo effect (DDEF) is the probable cause of such Es layer intensification over BV. However, since the disturbance dynamo is a global mechanism, these strong Es layers would also be expected in other regions. Since they did not observe this fact, they concluded that the real consequence of the electric field in the formation of Es layers during the disturbed periods was still unclear.
In light of the above discussion, this work analyses the electric field role in the Es layer formation during magnetic storms, providing novel insights about their coupling. First, we analyzed a set of days around 20 magnetic storms in three different regions over the Brazilian sector: an equatorial geographic station (BV), a transition station from equatorial to low latitude (São Luís (SLZ), Geographic Coordinates: 2.3°S, 44.2°W, Magnetic Inclination: ∼8°), and a low latitude station (Cachoeira Paulista (CXP), Geographic Coordinates: 22.41°S, 45°W, Magnetic Inclination: ∼35°). Afterward, we performed an in-depth analysis of the F region parameters to obtain the electric field values. Hence, it was possible to quantify the effect of the electric field in the Es layer formation for each analyzed region. The estimated electric fields were used as input to the E region ionospheric model (MIRE) (Resende et al., 2017a) to find the threshold value that is capable to cause the Es layer strengthening. Therefore, all this analysis allowed us to discuss the electric field role in the Es layer dynamics considering different locations during the disturbed periods, as shown in the following sections.

Methodology
Data from digital ionosondes and MIRE simulations were used to study the effect of the electric fields in the Es layer modification during 20 geomagnetic storms between 2015 and 2018. The methodology of the analysis is presented in the following sections.

Analysis of the Vertical Drift and Es Layer Parameters
We used the data obtained from the digital ionosondes (Digisonde) installed in BV, SLZ, and CXP to collect the F region and Es layer parameters. This radar transmits radio waves continuously into the ionosphere ranging from 1 to 30 MHz (Reinisch et al., 2009).
The vertical drift velocity (V z ) is calculated as ∆hF⁄∆t where hF is the true height, corresponding to a defined frequency, obtained by the vertical electron density profiles of two consecutive ionograms, being ∆t the time interval between them. The frequencies at 4, 5, and 6 MHz were chosen for the calculation and the average was considered representative of V z (Abdu et al., 2010), which represents the vertical plasma drift over the magnetic equatorial regions. Since the three regions used in the present study are located a little outside (BV and SLZ) or far away (CXP) from the magnetic equator, it is necessary to consider the effect of the meridional wind in the vertical plasma motion (Rishbeth et al., 1978). Therefore, V z depends on the apparent vertical drifts (V ap ) (Nogueira et al., 2011). Furthermore, the recombination processes need to be taken into account for the drift velocity calculation since the F layer can be located at heights lower than 300 km (Bittencourt & Abdu, 1981). Thus, the final vertical drift (V zf ) and V ap are obtained as follows: where the β is the recombination coefficient, H is the scale height of ionization, I is the magnetic inclination angle (∼18° in BV, ∼7° in SLZ, ∼35° in CXP), U F is the meridional wind component in the F region (positive northward), and w D is the contribution of diffusion to the vertical plasma velocity given by equation   / D i w g , in which g is the gravity acceleration, and ν i is the ion-neutral collision frequency. A detailed methodology to obtain these parameters is described in Nogueira et al. (2011) and Resende et al. (2020).
Other three parameters were also necessary to analyze the electric field effect in the Es layer behavior: the virtual height of the F region (h'F in km), the Es blanketing frequency (fbEs), which corresponds to the frequency up to which the Es layer blocks the transmitted electromagnetic signal, and the top frequency (ftEs), which is the maximum frequency reflected by the Es layer. The frequency parameters are given in MHz. Additionally, we manually checked all the parameters used in this analysis to obtain a reliable ionospheric profile (Reinisch et al., 2004) since significant discrepancies are often found between the automatically generated and real ionospheric parameters in the studied regions.

The E Region Ionospheric Model-MIRE
MIRE is used to study the Es layer behavior at equatorial and low latitudes over the Brazilian sector (Carrasco et al., 2007;Resende et al. 2016Resende et al. , 2017aResende et al. , 2017b. The electron density is computed using the equations of continuity and momentum for the molecular/atomic ions ( and metallic ions (   Fe , Mg ). The system is solved using 0.05 km grid spacing in height, and 1 min time step between 00 UT and 24 UT.
The transport term of the continuity equation depends mainly on the meridional (U x ) and zonal (U y ) components of the tidal winds and the electric field components (E x,y,z ) as follows where ω i is ion gyrofrequency, v in is the ion-neutral collision frequency, m i is the mass of the ion, and e is the electric charge of the ion. Regarding the frame of reference, the X-axis points toward the south; the Y-axis points toward the east; and the Z-axis completes the right-handed coordinate system, pointing up.
The wind profile used as input to MIRE was obtained from the last version of the Global Scale Wave Model (GSWM-00). This wind model is derived from the resolution of the Navier-Stokes equations for tidal and planetary wave perturbations as a function of latitude and altitude, for a specific wave periodicity and zonal wavenumber (Hagan et al., 2002;Manson et al., 2002). Therefore, the GSWM-00 successfully describes the wind dynamics until 125 km, which predicts the diurnal (24 h) and the semidiurnal (12 h) tides that are necessary for the Es layer formation in MIRE. The parameters of the GSWM-00 are given by the High-Altitude Observatory of the National Center for Atmospheric Research in Colorado (http://www.hao.ucar.edu/ modeling/gswm/gswm.html). Resende et al. (2020) implemented the GSWM-00 parameters in MIRE to analyze the Es layers in BV. In the present study, we extended a similar analysis for SLZ and CXP using the GSWM-00 already included in the MIRE model for the first time for these regions. The horizontal tidal amplitudes (U x0 (z), U y0 (z)), the phases (t x0 (z), t y0 (z)), and the vertical wavelength (λ x , λ y ) of the respective diurnal (T = 24 h) and semidiurnal (T = 12 h) tides as provided by the GSWM-00 are used in the wind shear equations: where the subscripts x and y refer to the meridional and zonal directions, respectively, z 0 is a reference height, assumed as 100 km (Mathews & Bekeny, 1979;Resende et al., 2017a).
Finally, to perform the electric field effects analysis, we used the relationship that each variation of ∼40 m/s in the vertical drift velocity obtained in Equation 1 equals to the 1 mV/m in the zonal electric field (Fejer & Scherliess, 1995).

Five Stages of Data Processing
We used the Dst index to identify the geomagnetic storm periods. This data was acquired from the World Data Center in Kyoto (http//wdc.kugi.kyoto-u.ac.jp/dstae/index.html). We analyzed a set of 20 moderate/ intense magnetic storms (Dst < −50 nT) that occurred in 2015, 2016, 2017, and 2018. The analysis of this study consisted of: (a) identifying the presence of strong Es layers in the ionograms over BV station (b) analyzing the ionogram data available for the SLZ and CXP for the same period in the previous step (c) excluding the cases of Es c and Es h layers since they are formed only by the wind shear mechanism (d) identifying the magnetic storm phase in which the strong Es layers occurred for each region; and (e) obtaining the time variation of fbEs and ftEs from each selected period. Table 1 summarizes the data obtained using the described methodology. In this table, we show the quiet day used as reference, the maximum ftEs observed in this quiet day, the day of geomagnetic storm onset, the level of the magnetic storm represented by the Dst index, the maximum ftEs for each region together with the day that it was observed, and the magnetic storm phase in which these atypical Es layers occurred.
where ftEs Dist is the top frequency value for the abnormal Es layer during the disturbed day, and ftEs quiet is the quiet day value observed at the same local time as the abnormal Es layer. Equation 7 is generally used to quantify the positive and negative ionospheric storms in terms of the F2 layer critical frequency and its peak height (Blagoveshchensky & Sergeeva, 2020). The quiet days were selected to be the closest possible to the geomagnetic storms, and they come from the GeoForschungsZentrum Potsdam (http:// wdc.kugi.kyoto-u.ac.jp/qddays/index.html). Notice that we only considered events in which Es layers were observed on both quiet and disturbed days during the same local time. In other words, the DftEs values are extracted from the difference between the perturbed and quiet days in the Es layer observation instant.
RESENDE ET AL.

The March 17, 2015 Geomagnetic Storm: An Example of a Case Study
The impact of Saint Patrick's magnetic storm in the ionosphere was well studied by several authors (De Michelis et al., 2020;Denardini et al., 2020;Maurya et al., 2018;Tulasi Ram et al., 2019;Spogli et al., 2016;Wu et al., 2016;Zhang et al., 2017). Here, we chose this event as an example of the effects in Es layer dynamics over the Brazilian region during the disturbed periods. Our motivation to study this magnetic storm is that the response of the lower ionospheric parameters is concentrated in the D region (Maurya et al., 2018); TEC distribution (Astafyeva et al., 2015;Spogli et al., 2016;Venkatesh et al., 2017;Wu et al., 2016), F region De Michelis et al., 2020;Venkatesh et al., 2017Venkatesh et al., , 2019, and in the space weather indices . The St. Patrick's Day storm and its consequences on Es layers were not deeply studied yet as far as we know. Figure 1 shows the temporal variation of the h'F parameter (red lines) over the three analyzed regions for the geomagnetic storm that occurred on March 16-18, 2015. Here, we used the h'F parameter to evaluate the physical mechanism related to the F layer's movement. The reference values (suffix Qd) are superimposed on the respective graphs (black lines), representing the quiet period (the quiet day considered for each geomagnetic storm is placed in the first column of Table 1 In this scenario, we observed two mechanisms acting in the ionosphere: the PPEF (Forbes et al., 1995) and the disturbance dynamo electric field (DDEF) (Blanc & Richmond, 1980). An abrupt increase in the AE parameter together with incursion of the IMF Bz to negative values was observed on March 17 at around 0600 UT, followed by a decrease in the base height of the layer over the regions, featuring a PPEF (see the line labeled "PPEF" in Figure 1). The PPEF events during this magnetic storm were described by Batista et al. (2017) and Venkatesh et al. (2019). They showed that an eastward PPEF in the afternoon acted as the main driver for the F3 layer formation in the equatorial and low latitudes. Other authors showed that the PPEF caused a strong EIA crest development due to the eastward intense electric field (Astafyeva et al., 2015;Venkatesh et al., 2017).
At 1500 UT on March 17, the AE index decreased, indicating the overshielding process (see the line labeled "OSD" in Figure 1). At around 2400 UT, the IMF Bz turned northward. Besides, close to 0300 UT on March 18, the h'F increased significantly for the three regions. From 0300 UT up to 1000 UT on March 18, the IMF Bz presented low values prevailing to the northward direction (positive values in Figure 1), and the AE index showed low values, oscillating near 100 nT with an elevation to 1000 nT at around 0900 UT. This behavior provides the appropriate conditions for the disturbance dynamo development. The significant increase of the h'F parameter at 0300 UT on March 18 confirms that an eastward electric field due to DDEF was present (see, for example, Fejer et al., 1983). The h'F parameter remains higher than its quiet time value until 1000 UT for the three regions, although for CXP and BV, it shows a tendency of recovery before 0600 UT followed by another increase after that. The double peak observations are out of the scope of the present work. Nevertheless, the main geomagnetic condition and ionospheric electrodynamics are clear indication of the DDEF occurrence, which is the focus of this analysis. At around 2100 UT until the end of the night on March 18, the Bz oscillated around zero, and the h'F for equatorial regions (BV and SLZ) decreased compared with the quiet reference day.
The significant modifications in the F region's electron density distribution over the equatorial and low latitudes were profoundly studied during the Saint Patrick's magnetic storm event. To analyze the Es layers behavior during this magnetic storm, Figure 2 shows the fbEs (orange line) and the ftEs (blue line) between March 16 and 18, 2015, for BV (a), SLZ (b), and CXP (c). The typical behavior of the frequency parameters is characterized by enhancement during the morning starting at around 0600 LT, reaching maxima values at around 1200 LT, followed by a steady decrease, reaching the quiescent values after 1800 LT (Resende et al., 2017a;2017b).
The frequency parameters of the BV region do not reach significant values under normal conditions, mainly during the daytime, because the tidal winds have low amplitudes in this location   On the other hand, the electric fields did not lead to significant modifications in the Es layers over CXP during quiet times (not shown here). In this station, located in a low latitude region, the winds are the primary mechanism responsible for the Es layer formation. They are driven by diurnal, semidiurnal, and terdiurnal tides in the E-region (Mathews, 1998;Pancheva et al. 2003;Resende et al. 2017a;Whitehead, 1961). The Es layer frequency parameters maintained a typical behavior with values below 6 MHz. We also observed the Es layer disruption during a long period between the main and recovery phases of the studied magnetic storm. This process started at 0800 UT and lasted up to 1120 UT. This is a blanketing Es layer, classified as "l" type. As mentioned before, the disturbance dynamo process was effective in these hours ( Figure 1). Thus, the zonal westward electric field is the most probable mechanism that acted during this atypical Es layer, agreeing with the previous study by Resende et al. (2020).
In SLZ, a strong Es layer appeared at 1900 UT on March 17, 2015. This Es layer reached a maximum frequency equal to almost 10 MHz at 2150 UT. This maximum frequency was determined from the last continuous point of the ionogram signal, as shown by the red arrow in Figure 3b. Afterward, the Es layer weakened and completely disappeared at 2340 UT. It is worth mentioning that two different Es layers seem to be present in hours around 2150 UT over SLZ. We observe an Es l layer, and in the background, we note the Es q layer occurrence, meaning that the EEJ irregularities can still be effective in SLZ. According to Forbes (1981), a region is generally considered equatorial if the magnetic inclination (dip angle) is up to 7°. Therefore, in 2015, this region was almost in the limit between equatorial and non-equatorial locations . Thus, the disturbed electric field in SLZ could have influenced the Es q layer intensification.
Over CXP we do not observe any expressive Es layer during this period. The ionograms in Figure 3c show a typical Es layer at 2100 UT that disappeared at 2220 UT on March 17, 2015, reappearing around 0340 UT in the next day. This Es layer persists until 0520 UT when it is disrupted again. The Es layer returned to RESENDE ET AL.
10.1029/2020JA028598 9 of 23 normal behavior during the nighttime. It is well-known that the electric field contributions in CXP to the Es layer formation are negligible (Resende et al., 2017a). However, some studies in the literature show that, in particular situations, the Es layer was intensified in this region (Abdu et al., 2014;Batista & Abdu, 1977), because it is under the influence of the SAMA. Thus, other mechanisms can influence the Es layer formation dynamics, such as particle precipitation (Batista & Abdu, 1977), and a significant conductivity enhancement (Abdu et al., 2014). Nevertheless, we did not observe any such behaviors in this case study.
Therefore, it is clear that the Es layer dynamics respond to the magnetic storm effects in different ways, depending on the location under analysis. Thus, we believe that the strong Es layer on BV and SLZ can be influenced by the disturbed electric field, whereas, on CXP, the winds are the principal agent in their formation. This same analysis was applied to the other magnetic storms to improve our understanding of the electric field role in the Es layer formation dynamics for these regions.

Correlation Between Es Layers and Disturbed Electric Fields
We applied the analysis shown in the previous section to the other magnetic storms listed in Table 1. Hence, we observed the Es layer development in the three Brazilian regions during the magnetic storm phases. In this analysis, we computed the drift velocity (V zf ) using Equation 1 during the nighttime hours to obtain the electric field component using the relationship mentioned in Section 2.2. Given that, we considered the most atypical Es layer occurrence in each storm phase and calculated the deviation DftEs defined in Equation 7. Thus, we built a relationship between the DftEs and the electric field values for each station. The results are shown in Figure 4 through the three-dimensional graphs. In this figure, the Es layer statistical analyses at BV (blue bars), SLZ (orange bars), and CXP (gray bars) are presented considering the main phase ( Figure 4a) and the recovery phase ( Figure 4b) for each one of the magnetic storms analyzed.
It is possible to conclude from Figure 4 that, over the BV station, the electric field influence in the Es layer during the main phase is weak compared with the recovery phase. In the former, the Es layer strength did not show strong variation with the electric field, whereas, in the later, the Es layers intensified for higher electric fields. Furthermore, the Es layers in BV during the recovery phase were significantly stronger than those of the main phase, with a density increase from 90% up to 500%.
In SLZ, we observe an inverse behavior because the Es layers were almost 90% stronger in the main phase compared to the quiet times. This occurred for the electric field values ranging from 0.5 to 1.4 mV/m. In this region, it is not possible to infer any correlation between the electric field intensity and the Es layer density during both phases.
Over the CXP region, only one case showed a significant density increase (97%) in relation to the quiet period which occurred during the main phase. Furthermore, it is also not possible to observe a clear correlation between the electric field intensity and the Es layer density during both phases of the magnetic storm.
RESENDE ET AL.

10.1029/2020JA028598
10 of 23 To provide a broader view of the relationship between the Es layer development and electric field values, Table 2 presents the ftEs variation concerning the quiet reference day (Equation 7) as well as the respective electric field at each magnetic storm phase. It is important to mention that the electric field calculated for the equatorial station SLZ was used for all the regions analyzed since the electric field can be mapped from the equatorial F region to low latitudes through the magnetic field lines (Abdu et al., 2014). Furthermore, the locations of the regions in this study have similar longitudes. Thus, the longitudinal variation does not cause a significant influence on the electric field values.
We calculated the vertical drift velocity around the atypical Es layer occurrence hours. This drift is then used to obtain the zonal electric field (Abdu et al., 2005;Kelley, 1989) from the drift equation V =  2 / E B B , in which the International Geomagnetic Reference Field provides the magnetic field B. This relationship concludes that a 1 mV/m zonal electric field produces a vertical drift of 40 m/s. In this analysis, the maximum computed drift velocity was considered when obtaining the electric field. Notice that we do not estimate the other electric field components because they do not significantly influence the Es layer dynamics over the analyzed regions (Abdu et al., 2014;Resende et al., 2020).
As mentioned before, the Es layers in CXP had almost no significant changes when the two magnetic storm phases are compared. At low latitudes, the wind shear mechanism is predominant, and therefore, the electric field plays only a secondary role in the Es layer formation (Haldoupis, 2011;Haldoupis et al. 2006;Whitehead, 1961). Prassad et al. (2012) studied the Es layer behavior in different regions of the globe, and they concluded that the magnetic storm effect in the Es layer at middle latitudes is very weak, considering it negligible. During disturbed times, the only mechanism that modifies the Es layers over the CXP region is the particle precipitation. As mentioned earlier, the effectiveness of this mechanism is verified through the Es a layers, shown in Batista and Abdu (1977). However, the Es a was not observed in our data. Therefore, we believe that this mechanism did not act in the period studied despite the intense geomagnetic storm.
Over SLZ, a region closes to the magnetic equator, our results pointed out that the disturbed zonal electric field can cause modifications in the Es layer formation, mainly during the magnetic storm main phase. However, this behavior does not follow a specific pattern. It is important to mention here that the geomagnetic field inclination in the Brazilian sector varies at a rate of 20′ per year, corresponding to an apparent northwestward movement of the magnetic equator . Thus, the SLZ site is not considered an equatorial station nowadays since the magnetic equator is departing from this region. Although the wind shear mechanism is efficient at the SLZ region, forming blanketing layers, we also observe the Es q and other Es layer types in these disturbed periods. It means that the electric field of the EEJ instabilities could still work together with the tidal winds in the Es layer formation process. Depending on the electric field direction, the EEJ plasma irregularities could be stronger, leading to higher ftEs (Resende & Denardini, 2012;Resende et al., 2013). This fact explains some Es layers being strengthened in some cases during the main phase of the magnetic storms in SLZ.
The strong Es layers that occurred in BV confirm the hypothesis in Resende et al. (2020), in which the most probable mechanism acting during these atypical Es layers was the zonal westward electric field caused by a disturbance dynamo. The authors did a deep case study using observational data and simulations to show some evidence that the Es layer density is significantly enhanced when the disturbed zonal electric field is present. However, they affirmed that the electric field effect in the Es layer formation over BV still needed more in-depth analysis. In the present work, the strong Es layer in BV was observed in all events during the analyzed magnetic storm recovery phases. In the following sections, we discuss the physics of these Es layers in BV to show their relationship with the DDEF.
Lastly, the electric field value presented in Table 2 varies between 0.2 and 2 mV/m, agreeing with the intensity of the disturbed electric that penetrates to the ionosphere (Gonzalez et al.,1994;Tsurutani et al., 2008). As we did not find in the literature a quantitative electric field effect on the Es layer formation, we provide this analysis as follows.

The Threshold Value of the Electric Field in the Es Layer Formation
The MIRE model has been used to simulate the Es layers with a high confidence level over the Brazilian sector (  field was neglected without affecting the simulation accuracy (Resende et al., 2017a;2017b). For regions around the magnetic equator, the electric field of the EEJ current was used to study the Es layer development (Moro et al., 2017;Resende et al., 2016). Recently, using the MIRE simulations, Resende et al. (2020) showed a possible connection between the westward electric field and the strong Es layer that occurred in the geographic equator region.
In the present work, based on the electric field values derived from drift velocity (Table 2), we analyzed their influence at each region considered in this study. First, we fitted the wind profile for the three stations using the GSWM-00. This process is required to obtain values that represent better the Es layer dynamics. We used 80% of the wind amplitudes for São Luís and 90% of the amplitudes for Cachoeira Paulista, which best resolve the Es layer formation in our simulations. Figure 5 shows the wind amplitudes (color scale) computed using the wind parameters in Equations 5 and 6. The map format in this figure given in height versus universal time (UT) refers to the diurnal and semidiurnal tidal wind for the meridional and zonal components over BV (a), SLZ (b), and CXP (c). The wind parameters for March were used to generate the maps. Although the tidal winds have a seasonal variation, the purpose of this work is to verify the electric field effect. Thus, the GSWM-00 model's amplitudes are satisfactory to be used as a background profile for this purpose. Finally, notice the presence of wind shearing in all components (zeros in the profile), which is a necessary condition to produce Es layers.
The results in Figure 5 show that the wind amplitudes are much lower over the BV region than at SLZ and CXP stations. This behavior corroborates with the fact that weak Es layers are seen in the observational data at BV. In this station, we observed that the meridional wind amplitude is almost equal to the zonal amplitude, with values around 30 m/s. SLZ and CXP present the most intense wind profiles. In general, the wind RESENDE ET AL.

10.1029/2020JA028598
13 of 23 behavior over SLZ and CXP agrees with the previous analysis in Resende et al. (2017aResende et al. ( , 2017b, in which they used data from a meteor radar to obtain the wind components. In this work, we chose the GSWM-00 model since it yields good results for our interest zone, which is around 100-110 km. We also observed that the wind behavior is similar over BV and SLZ, which is expected since we are considering the equinoctial condition, and the distances of those regions to the geographic equator are almost the same.  simulated performed a downward movement, which is the typical behavior of their electrodynamics (Haldoupis et al., 2011). This behavior occurs due to the semidiurnal and diurnal periodicity that characterizes the Es layer movement (Resende et al., 2017a).
During the daytime, the electron density of Es layers in BV is very close to the E region peak density. This behavior occurs due to the low wind amplitudes around the geographic equator regions, as mentioned before. These Es layers are more expressive during the nighttime, agreeing with what we have seen in observational data. Over SLZ, the Es layer is denser in the daytime than in the night hours. In CXP, the Es layers occur during all day, with similar density. We also noticed that the simulated Es layers are in high altitudes in some hours with an evident downward movement. This fact reinforces that the winds play a fundamental role in the Es layer formation over these regions.
To analyze the electric field effect in the Es layer development, we performed multiple simulation scenarios considering a constant westward zonal electric field with values ranging from 0.25 mV/m to 3.0 mV/m and a step of 0.25 mV/m. This analysis aims to estimate the Es layer fraction that strengthens due to the electric field value. The simulations indicated that there are different threshold values for each region. To exemplify some results, we show the HTI maps in Figure 7 for ( Comparing these results with the previous one, which considers only tidal winds (Figure 6), it is noted that the Es layer electron density increases as the electric field component increases. In BV, the simulated thin layers that characterize the Es layer are not forming with 3.0 mV/m. On the other hand, for the same electric field value, a denser and thin Es layer is observed during the daytime in SLZ. This result shows that RESENDE ET AL.

10.1029/2020JA028598
15 of 23 although the wind profiles were symmetric, the electric field influence is different in these regions. Also, in CXP, the electric field does not seem to have an immediate influence as observed over BV, in which the electron density of the Es layers intensifies gradually. The main difference in simulations is that the electric field variation does not cause a significant modification in the Es layer process in CXP, as shown in the results for 0.25 mV/m (density increase of ∼6%) and 1.5 mV/m (density increase of ∼38%). However, the same variation causes a significant strengthening of the Es layer in simulations over BV and SLZ, reaching a density increase of ∼236% and ∼890%, respectively for an electric field of 1.5 mV/m.
To further support our claim that the electric field is the main responsible for the Es layer strengthening in regions near the equator, Table 3 shows the maximum intensity in the Es layer occurrence for each electric field value tested in MIRE. Additionally, we show the percentage of the Es layer increases concerning the same hour of the simulations that consider only the wind profiles. It is essential to mention here that the Es layer's maximum density occurred around 0300 UT, 0500 UT, and 0400 UT for BV, SLZ, and CXP, respectively.
In Figure 7 we observe a significant intensification during the nighttime in SLZ. The Es layer increase in simulations can be greater than 1000%, which does not agree with the observational data shown in Figure 4.
Electric field values up to 0.75 mV/m in the model simulate Es layers that show good correlation with the observational data. For higher electric field values the simulation results were unrealistic. Thus, the threshold (limit) value of the electric field to form the Es layer for this region is 0.75 mV/m during the disturbed period. We believe that the unsatisfactory results in this region can be due to other Es layer formation mechanisms not included in MIRE, such as EEJ instabilities. We will discuss this behavior in more detail in the next section.
Over BV, the intensification values in simulations have a reasonable correlation with the observational data, showing that the model reacts well to the wind and electric field behavior in this region. We notice that the threshold electric field value to form the Es layer in BV is 2.5 mV/m, that is when the model still converges, and the electron density is satisfactory compared with the observational data for the Es layer reported in the literature (Resende et al., 2017a;2017b;Resende et al., 2020). The Es layer becomes extensive with unrealistic values when the electric field is equal to 3 mV/m, as shown in Figure 7.
Finally, the electric field variation causes low modification in the Es layer formation in CXP, showing that tidal winds are dominant over this station. The Es layer increased by a maximum of 55% in simulations, while the observational data show intensifications around 60% in most cases of this analysis. The model was able to simulate the Es layers for electric field values up to 3.7 mV/m, but the electron density did not continue to increase significantly. Therefore, in the next section, we try to discuss the competing roles of electrodynamical and dynamical processes in the Es layer formation over these three regions studied.

Discussions
In this work, we evaluate the electric field effect in the strengthening of the Es layer over the Brazilian sector. The results have shown that the electric field is effective in the Es layer formation during the recovery phase of magnetic storms in BV and during the main phase of magnetic storms in SLZ. Nevertheless, over the CXP region the electric field does not seem to cause a significant influence in the Es layer formation.  Therefore, to investigate which phenomenon occurs in areas around the equatorial/magnetic equator, we used also additional techniques to those shown in previous sections. First, we emphasize here that the wind shear mechanism is the main driver for the Es layer formation over the Brazilian sector. We believe that during the disturbed periods, the electric field is superposed to tidal winds reinforcing the Es layers.
In relation to the BV region, Resende et al. (2020) showed that DDEF is a possible mechanism to trigger the atypical layers in this region. They observed three cases that the DDEF leads the observed Es layer strengthening. However, they affirm that further study would be required to compute the statistical behavior of such occurrences. As described in Resende et al. (2020), to analyze the electric field performance, we processed the vertical drift during our interest hours, that is when the atypical Es layer occurs. The positive drift values mean that the electric field points eastward, whereas negative drift values indicate that it points westward. Here, we found westward electric fields during all the cases under analysis.
Therefore, to confirm that the DDEF is acting, we used the total electron content (TEC) to analyze the weakening of the equatorial ionization anomaly (EIA) in some hours, which indicates the DDEF occurrence. Therefore, maps of the South America TEC were analyzed. The TEC is calculated using a technique developed by Otsuka et al. (2002), in which a weighted least squares fitting determines the instrumental biases, assuming that the hourly TEC average is uniform. The maps show the TEC in a two-dimensional form, and they are available in Brazilian Studies and Monitoring of Space Weather (Embrace -http://www2.inpe.br/ climaespacial/portal/en/). These TEC maps have a 10 min time resolution and 0.5 × 0.5° of spatial resolution in latitude and longitude. More details about the TEC maps methodology can be found in Takahashi et al. (2014Takahashi et al. ( , 2016. The quiet time behavior of the TEC in the maps shows a high-density observation area extending between 20°-30°S and 40°-60°W, which characterizes the EIA. This anomaly occurs due to the fountain effect in which the plasma along the geomagnetic equator is raised under the action of the E×B drift and subsequently it moves downwards along magnetic field lines, under the action of diffusion and gravity, generating plasma crests over the off-equatorial region, in the Northern and Southern Hemispheres. Generally, the EIA is well developed at 2300 UT (19-20 LT in Brazilian longitudes) and its peaks can be observed around ±15° magnetic latitude in TEC maps. Figure 8 shows the TEC maps over South America during the days around the St. Patrick magnetic storm that is used as an example in our analysis. The TEC map during the reference quiet day (March 10, 2015) at 2300 UT is presented in Figure 8a Figure 8a). Hence, the EIA's weakening shows that the zonal westward electric field caused by a DDEF is present. Although the DDEF started from 0300 UT (Figure 1), the TEC maps suggest that it lasted until ∼15 h later. The high values of ftEs that also occurred in the nighttime period on March 18, 2015 (see Figure 2a), support this statement in BV.
To see the DDEF effect and the connection with the strong Es layers in all magnetic storms used in this study, we analyze the relative difference (RD) parameter over the TEC maps. The RD (Equation 8) is calculated through the TEC maps for each day that we observe the strong Es layers (TEC Dist ) concerning the typical (non-disturbed) day (TEC Ref ) of each magnetic storm used in this analysis given in Table 1. We used the EIA peak between 20° and 30°S at 2300 UT to perform this analysis, selecting the highest RD value in this latitude range. RD is computed here as follows: Thus, we correlated the DftEs during the recovery phase of the magnetic storms with the RD parameter. Here, we used as a TEC Dist reference, the day that the strong Es layer occurred in BV. Figure 9 shows a 3D graph distribution for each station in this analysis at BV (blue bars), SLZ (orange bars), and CXP (gray bars). In the most extreme case, the Es layer density increased more than 500% in BV, whereas the EIA weakened almost 100%, i.e.,  Dist TEC 0. Therefore, this analysis indicates that DDEF action is a significant factor in the anomalous Es layer formation in BV, confirming the assumption in Resende et al. (2020). This correlation is important since the presence of a strong Es layer in the ionograms at BV can be an indicator of the DDEF when the TEC data are not available.
During the magnetic storm recovery phases, the electric fields do not have a significant effect in SLZ and CXP. Therefore, we observed a low correlation between the EIA weakness (due DDEF effect) and the Es layer formation in these regions. We believe that the DDEF does not cause notable influences in SLZ or CXP because other mechanisms act together in the Es layer formation in these regions. These Es layer modifications may even be due to the seasonal wind variation, mainly in CXP (Resende et al., 2017a). One crucial factor is that the tidal wind amplitudes are higher in SLZ and CXP than in BV, showing that the wind shear mechanism can give rise to the denser Es layers in such regions.
The Es layers reached high frequencies in some cases over SLZ during the main phase of the magnetic storms, as shown in Figure 4. A study performed by Rastogi et al. (2012) related the Es layer modifications with the undershielding/overshielding electric fields over the equatorial and low latitude regions. They observed that these layers occurred during the main phase of magnetic storms in the equatorial region and are associated with the large westward PPEF. Additionally, Abdu et al. (2014) affirm that the zonal westward electric field during disturbed times (overshielding) contributes to the formation of the Es layer around equatorial/low latitudes. They attributed these anomalous Es layers to the enhanced ratio of the field line integrated Hall to Pedersen conductivity (∑H/∑P), which is possible due to the SAMA presence. However, in the main phase of the Saint Patrick's magnetic storm, TEC data showed an intensification of the EIA, representing an undershielding process (eastward electric field) during the daytime, and it is expected that the Es layer disappears (Abdu et al., 2014). Thus, we believe that the electric field effect in SLZ indicates that other mechanisms could be acting. This fact corroborates with the irregular pattern of the strong Es layers in this station, observed through the low correlation between the electric field intensity and the Es layer density in the statistical study. Furthermore, we do not find any evidence of particle precipitation over SLZ and CXP, thereby discarding the SAMA influence.
In our study, some ionograms in SLZ caught our attention. One example is the Es intensification in SLZ during the main phase of the St. Patrick magnetic storm (Figure 3b). In this case, it is evident that the Es layers are blanketing layers ("f" type), but it also seems to have an irregularity layer (Es q layer) in the background. Here, we strong believe that Es q layer occurs because of the high frequency that this layer reached (ftEs > 9 MHz) since the tidal wind behavior are not so strong in this region, as mentioned in Resende et al. (2017b). Also, at the same time, the Es layers blocked the upper regions meaning that the Es f is present. Therefore, we conclude that 2 Es layers were present on the same day, and the SLZ location might still be under some influence of the EEJ current. This result closely resembles those obtained by Devasia et al. (2006), in which they studied the characteristics of different Es layer types and their association with the plasma density irregularities over the magnetic equator. The authors concluded that the Gradient Drift instability structures may occur when the EEJ current is not strong. In such cases, the Es q layers can be formed simultaneously with other Es layer types. Thus, we believe that, in the analyzed case, an electric field imposed during the magnetic storm main phase may have contributed to form the Gradient Drift instability. In turn, this mechanism produced the Es q layer, even if the studied station is located in the magnetic equator border.
To corroborate our hypothesis that there is more than one mechanism acting in the Es layer formation in SLZ, we evaluated the magnetometer data for the St. Patrick magnetic storm event. Figure 10 shows the EEJ ground-strength variation as a function of UT for the disturbed periods analyzed (March 17 and 18, 2015). The magnetic data treatment consists of analyzing the five quietest days in a respective month and obtaining the average local midnight values. Thus, this average is subtracted from each value of the H component, providing the ∆H for the corresponding station. To evaluate the EEJ current, we used data from an equatorial station SLZ, and from an off equatorial station, Eusébio (EUS, 03.89°S, 38.44°W, dip: −14.83). Thus, the variation of the EEJ ground strength is estimated by taking the difference between the ∆H values of these stations (∆H SLZ −∆H EUS ). A more detailed explanation of the magnetic data treatment and the use of the two stations can be found in Denardini et al. (2009).
Two important characteristics were observed in Figure 10 on March 17, 2015, such as (i) there is a period of reversed electrojet currents (CEJ events) during the daytime that coincides with the absence of Es layer (around 13-16 UT in Figure 2), and (ii) the EEJ current is intensified during the same time that the strong Es layer occurs in SLZ (around 20-23 UT in Figure 2). These characteristics are a real manifestation of the Gradient Drift instability, which is driven by the vertical polarization Hall electric field and the density gradient causing diffuse traces in ionograms, the Es q layers. It is important to mention here that the eastward electric field intensifies these irregularities. Thus, the unrealistic results of MIRE model probably occur because the Gradient Drift instability is not considered in the simulations, as seen in Resende et al. (2016) and Moro et al. (2017). In fact, in transition regions such as SLZ, under the limiting modification of influence of the EEJ, the competing roles of electrodynamical and neutral-dynamical processes (wind shears) in the Es layer formation are frequent, and we observed different types of Es layers in ionograms . Thus, we have here evidence that there is a possibility that the EEJ current still causes irregularities over the SLZ station.
The above affirmation is supported in some previous works. The regions are considered an equatorial station until a magnetic inclination of 7° (Forbes, 1981), and consequently, the Es q layers occurred in such areas. However, Moro et al. (2016) showed that the vertical electric field of the EEJ current, responsible for the Gradient Drift instability occurrence, is strong around 100 km in recent years in SLZ (   7 ), I where Es q usually appears. Furthermore, Resende et al. (2016) showed that in 2015, in which the magnetic inclination in SLZ was around 7.8°, the Es q layers still occurred during a few hours. The authors also showed that during the disturbed periods in the daytime, the weakness/disruption of this Es layer type is frequent due to the CEJ events Resende & Denardini, 2012;Resende et al., 2013). The Es q layers only returned when the EEJ would establish for its typical conditions, eastward (positive) at daytime. This behavior seems to have occurred in our analysis, as seen in the correlation between Figures 2 and 10. The Es q and the Es b layer occurrences show that the strengthening Es layer during the main phase in some cases is due to the EEJ electric field that operates in increasing instabilities in this region. Thus, it is not a rule that the disturbed electric field and winds can cause an intense Es layer over SLZ as at the BV region. Additionally, as mentioned before, we believe that the model used here does not show satisfactory results when the electric field is included in SLZ because different mechanisms acted in the Es layer development in this region. In other words, we credit these discrepancies in our simulations because we do not consider the EEJ dynamics in MIRE.
Also, our results over CXP agree with the recent analysis performed by Conceição-Santos et al. (2019), in which they analyzed the different types of Es layer during four months in São José dos Campos (23.2°S, 45.8°W, dip: ∼21.0°), a station near to CXP. They did not observe the Es a layers, indicating the absence of the particle precipitation mechanism, as mentioned before. Nevertheless, they found one case in which a moderate increase occurred in the Es layer frequencies during the disturbed period of 01-05 September 2016. In this magnetic storm, the Es layer density in CXP increased by 34% in the main phase, as shown in Table 2. Conceição-Santos et al. (2019) suggested that the Es electron density enhancements are associated with redistribution of ionization driven by the wind shear mechanism. Thus, we believe that the only effective mechanism in the Es layer development during these years is the tidal winds. Therefore, the variations concerning the quiet periods in CXP can be associated with the seasonal tidal wind variabilities, as shown in Resende et al. (2017a).
Finally, it is remarkable that the electric field influence in the strong Es layers is specific of the regions near the geographic equator. Our analysis corroborates with the proposal of Resende et al. (2020), in which the intensifications observed in BV are associated with DDEF. The low amplitudes of the winds over the BV station seem to favor the electric field influences in the Es layer modifications. Furthermore, in SLZ, we notice some irregular intensifications that occurred during the magnetic storm's main phase. Our results give an indication that the EEJ instabilities might still be present over the SLZ station. Over CXP, the tidal winds are the principal agent for the Es layer formation, and for this reason, there are no significant modifications during the disturbed periods.

Conclusions
We performed a comprehensive study of the disturbed electric field influences in the Es layer formation during 20 magnetic storms. We analyzed the digisonde data and modeling results during the main and recovery phases of magnetic storms in three regions over the Brazilian sector, BV, SLZ, and CXP.
As a case study, we analyzed the Saint Patrick's magnetic storm event. We observed a clear relationship between the anomalous Es layers and the disturbance dynamo in BV. The same finding was statistically verified in the other 20 events studied here. Therefore, this work confirms that the Es layer behavior is strongly affected by the zonal DDEF in this region. This correlation is very important since the presence of a strong Es layer in the ionograms at BV can be an indicator of the DDEF when the TEC data is not available.
Over SLZ, we noticed that the electric field has an influence on the Es layer formation during the main phase of the magnetic storms. However, due to the low correlation between the electric field intensity and the Es layer density in the statistical analysis, we assumed that the EEJ current was still effective in SLZ even though this region is located near the magnetic equator influence border. In the ionograms on March 17, 2015, the Es q and the Es b layers occurred simultaneously. The magnetometer data reveals that the EEJ current was intensified during the main phase of this magnetic storm, reaching values high enough to develop the Gradient Drift instability structure. Thus, the irregularity layers were formed, leading to high ftEs values.
The Es layer behavior in CXP was not influenced by the electric field at any phase of the magnetic storms. The wind shear mechanism was predominant, and therefore, the electric field played only a secondary role in the Es layer formation process. The only mechanism that could modify the Es layers over the CXP region is the particle precipitation during the disturbed times, as seen in previous works. However, we did not find any evidence of this mechanism in CXP, confirming that only the wind shear mechanism operated in our cases.
The results obtained from the theoretical model showed a good agreement with the observed Es layer formation over BV and CXP. The electric field threshold in BV is 2.5 mV/m, meaning that the numerical simulation provided satisfactory electron density for values lower than that. For higher values, the simulated Es layer density becomes unrealistic. In CXP, the electric field variation in simulations caused few modifications in the Es layer formation, showing that tidal winds are dominant in this region. In this case, the maximum increase of the Es layer density was 55%, whereas the observational data showed an intensification of ∼60% in most cases of this analysis. At last, MIRE did not provide satisfactory results when the electric field was included in SLZ. We credited these discrepancies to the fact that the EEJ effect is not yet modeled in MIRE.
Therefore, this study shows that the disturbed electric field can impact the Es layer formation at regions in the geographic/magnetic equator. We notice that the zonal eastward electric field in the main phase of the magnetic storm can cause an equatorial Es layer in SLZ, whereas it can weaken or not cause changes in the Es layer in regions such as BV and CXP. During the recovery phase, the zonal westward electric field contributes to forming the Es layer in BV. Finally, the electric field role in the Es layer dynamics in this work provides a significant contribution to our understanding of these competing mechanisms in the Es layer formation during disturbed periods.

Data Availability Statement
The authors thank the High Altitude Observatory (HAO) of the National Center for Atmospheric Research (NCAR), in Colorado (http://www.hao.ucar.edu/modeling/gswm/gswm.html) for providing wind data used in MIRE model. The authors thank the OMNIWEB for providing IMF Bz, AE and Dst parameters used in the classification of the days and the GeoForschungsZentrum (GFZ) Potsdam for providing the list of geomagnetically quiet days (http://wdc.kugi.kyoto-u.ac.jp/qddays/index.html). The Digisonde data from Boa Vista, São Luís and Cachoeira Paulista, TEC data, and Magnetometer data can be downloaded upon registration at the Embrace webpage from INPE Space Weather Program in the following link: http:// www2.inpe.br/climaespacial/portal/en/.