Geomagnetic Activity Control of Irregularities Occurrences Over the Crests of the African EIA

This paper investigated the behavior of ionospheric irregularities over the African equatorial ionization anomaly (EIA) crests during intense geomagnetic storms that occurred from 2012 to 2015. Irregularities were monitored using the rate of change of TEC index along with variations of the horizontal component of the Earth's magnetic field (H) and ionospheric electric current disturbance (Diono). The predictive capability of the Prompt Penetration Equatorial Electric Field Model (PPEFM) was assessed by comparing prompt penetration electric field (PPEF) inferred from interplanetary electric field and Diono with PPEF derived from the PPEFM, with emphasis on how well the model reproduced enhancement/reduction in the prereversal enhancement (PRE). Eastward PPEF triggered short duration irregularities on 23 April 2012, 17 March 2013, and 20 February 2014 while westward electric field reduced them thereafter. The PPEFM rightly predicted enhancement (reduction) in PRE on 17 March 2013 (19 February 2014) when irregularities were triggered (inhibited). It, however, showed no change in the PRE on 23 April 2012 and 20 February 2014. During the storms recoveries, irregularities were always inhibited/reduced over the trough by westward disturbance dynamo and the inhibition lasted longer during the superstorm of March 2015. Also, there was a hemispheric asymmetry in irregularities over the African EIA crests. On 16–17 July 2012, 15 November 2012, and 19 March 2013, there were differences in irregularities behavior. On these days, the asymmetry of the postsunset crests was pronounced in both hemispheres.


Introduction
Ionospheric irregularities are small to large-scale structures that form in the plasma density (Perkins, 1975). Their interaction with Global Navigation Satellite System (GNSS) signals could result in rapid fluctuations in the amplitude and/or phase of the signals, giving rise to a phenomenon known as scintillation (Aarons, 1982;Datta-Barua et al., 2015). During intense scintillation conditions, GNSS signals might suffer degradation, reduction in their information content, or failure in reception (Aarons et al., 1996;Kintner et al., 2007). The outcome could have disastrous effects on the life-critical GNSS applications especially, those utilized in navigation, positioning, search, and rescue, as well as military operations and surveying (Conker et al., 2003;Sunda et al., 2015). For this reason, adequate information about the actual state of the ionosphere is crucial for the smooth operation of the critical GNSS applications during all-weather conditions. The low-latitude ionosphere is mainly characterized by features such as the equatorial ionization anomaly (EIA) and ionospheric irregularities. The EIA results from the interaction of eastward electric field and the horizontal north-south geomagnetic field (Appleton, 1946;Namba & Maeda, 1939;Yue et al., 2015). Irregularities are generated by the generalized Rayleigh-Taylor instability (R-T instability) (Eccles, 2004;Farley et al., 1970;Yizengaw, Retterer, et al., 2013) in the postsunset when plasma is further lifted up under the action of the prereversal enhancement (PRE) in E × B drift velocities (Fejer et al., 1999). The plasma bubbles irregularities are depleted flux tubes that move upward, including the portion in the EIA crests (Groves et al., 1997). Parameters affecting the generation of irregularities are the (i) postsunset vertical drift, (ii) components of thermospheric winds, (iii) density gradient at the bottomside of the F layer, and (iv) initial seed perturbations due to gravity wave from the lower atmosphere (Haerendel, 1973;Ott, 1978;Sultan, 1996;Tsunoda et al., 2013).
During solar disturbances such as geomagnetic storms, electrodynamics of the equatorial/low-latitude ionosphere undergoes drastic variations. The equatorial region is mostly affected by mechanisms, such as the prompt penetration electric field (PPEF) (Abdu et al., 2018;Abdu, 2012;Kikuchi et al., 2008;, and disturbance dynamo electric field (DDEF) related to winds driven by Joule heating and ion drag. The time of occurrence, the polarity, and the combined effect of disturbed storm time electric fields are, therefore, crucial in controlling the variability of irregularities. PPEF has eastward (westward) polarity on the dayside (nightside), while DDEF is of opposite configuration (i.e., westward/eastward on the dayside/nightside) (Astafyeva et al., 2018;Blanc & Richmond, 1980;Yamazaki & Kosch, 2015). Eastward (westward) electric field occurring in the postsunset can enhance (weaken) the regular eastward vertical plasma drift, thereby affecting the F layer rise (Fejer & Scherliess, 1995) and consequently the generation of irregularities (Aarons, 1991;Abdu, 2012;Kelley, 1989;Shreedevi & Choudhary, 2017).
First evidence of geomagnetic disturbance on the horizontal component of the Earth's magnetic field (H) was presented by Chapman (1918). Later on, Nishida et al. (1966) and Nishida (1968) identified the disturbance polar no. 2 (DP2) current, while Vasyliunas (1970) presented a theoretical model for magnetospheric convection. Manoj et al. (2008) postulated that the maximum propagation time for interplanetary electric field (IEF) to travel from the nose of the bow shock to the equatorial ionosphere is about 17 min. Model results of Fejer and Scherliess (1997) revealed that PPEF vanished after 60 min due to the shielding effect of the ring current. However, cases of long-lasting penetration electric field in the equatorial ionosphere during periods of enhanced magnetospheric activity and continuous southward interplanetary magnetic field (IMF) have been reported (Huang et al., 2005(Huang et al., , 2010. Huang (2019) recently showed that penetration electric fields dominated the equatorial plasma drifts for well over 14 h, including 3 h of the main phase and the first 11 h into the recovery of the storm of December 2006. In line with the disturbance dynamo theory of Blanc and Richmond (1980), Le Huy andAmory-Mazaudier (2005, 2008) linked the magnetic signatures of the reversed solar quiet (Sq) current at low latitude to the ionospheric disturbance dynamo (Ddyn), which is the equivalent current system associated to DDEF.
Despite its wider spatial coverage over the low-latitude region, Africa has the fewer number of studies in terms of ionospheric response to storm time electric fields. The lack of studies for Africa has been attributed to the long time absence of ionospheric observational tools over this sector (Paznukhov et al., 2012;Yizengaw, Doherty, & Fuller-Rowell, 2013) that constitutes an impediment to global modeling. Over the last decade, however, the availability of ground-based instruments thanks to projects such as the International Equatorial Electrojet Year, the International Heliophysical Year, and the International Space Weather Initiative has helped in improving the knowledge the ionosphere over Africa.
For example, evidence of the inhibition of irregularities in Nairobi, Kenya, and Kampala, Uganda, has been presented during the storm of 6/8 April 2011 (Ngwira et al., 2013). Similarly, the suppressing effect of westward storm time electric field on the PRE and the ensuing inhibition of irregularities over East Africa together with the reduction in the virtual height of the F 2 layer (h'F2) (at Ascension Island) was presented by Azzouzi et al. (2015) for the October 2013 event. In addition, Ddyn and DP2 signals have been separated during several storms over Africa (Amaechi et al., 2018a;Azzouzi et al., 2015;Fathy et al., 2014;Nava et al., 2016). These works revealed characteristics of DP2 and Ddyn in terms of their source/origin, period, ionospheric responses, and longitudinal behavior  as well as amplitude, hemispherical behavior, and asymmetry (Zaourar et al., 2017). Also, model studies using the Thermosphere-Ionosphere-Electrodynamics General Circulation Model (Carter et al., 2014) and the Prompt Penetration Electric Field Model (PPEFM) (Nayak et al., 2016) have been done to assess the ability of storm to enhance/suppress irregularities and predict PPEF and its effect on irregularities, respectively. Such studies are more than ever needed over the African EIA during storms.
Despite all these works, a better understanding of the storm time behavior of ionospheric irregularities is still needed over the African EIA. Features such as their simultaneous response to PPEF and DDEF and the combination of both disturbed electric fields over the crests in both hemispheres are yet to be investigated during various storms. This poses not only a limitation to global modeling but also challenge to forecasting space weather that has long been the goal of the Space Physics and Aeronomy community. This paper investigates variations of irregularities over the crests of the African EIA in both hemispheres during intense geomagnetic storms of the ascending phase of solar cycle 24. The study further performed, for the first time over this region, comparison between PPEFs derived from ground-based magnetometer data and inferred from the PPEFM (Manoj & Maus, 2012). To this end, section 2 describes the data sets and method of analysis, while section 3 gives a highlight of the results obtained. The discussion and conclusion are presented in sections 4 and 5, respectively.

Data Sets
All storm events under investigation were associated with coronal mass ejection (CME) (see the Solar and Heliospheric Observatory/Large Angle and Spectrometric Coronagraph CME catalog at https://cdaw. gsfc.nasa.gov/CME_list/ for a description of basic attributes of these events). Their evolution in the interplanetary medium was monitored using the z component of interplanetary magnetic field (IMF Bz) and x component of the solar wind speed (Vx). Both data sets recorded onboard the Advanced Composition Explorer satellite with time resolution of 64 s were time shifted by about 52 min to account for propagation delays to the Earth's magnetosphere (Chakrabarty et al., 2005). They were utilized to compute the y component of interplanetary electric field (IEFy). Changes in the Earth's magnetosphere were examined using the symmetric H index (SYM-H) and horizontal component of the geomagnetic field (H). SYM-H data were provided by the International Service of Geomagnetic Indices. This data set with 1 min resolution is more suitable for monitoring changes in the solar wind dynamic pressure (Wanliss & Showalter, 2006) and related current in the magnetosphere during storms. The ionospheric response vis-à-vis irregularities variation was analyzed using indices derived from GNSS observables.
H was computed using magnetometer data at the Addis Ababa station that is managed by the Institut de Physique du Globe de Paris, and the data are distributed via the International Real-Time Magnetic Observatory Network. GNSS observables for stations located in the African equatorial/low-latitude region around longitude 37°E were obtained from University NAVSTAR Consortium. These data with resolution of 30 s were used to estimate vertical total electron content (VTEC) and derive the rate of change of total electron content (TEC) index (ROTI). The coordinates of the magnetometer and GNSS stations are given in Table 1.

Methods of Analysis
The y component of the IEFy was estimated using the formula IEF y = −V × × IMF Bz (Kelley, 1990) where V × and IMF Bz are the x component of the solar wind speed and z component of the interplanetary magnetic field, respectively. Also the horizontal component of the geomagnetic field (H) was computed using the north (X ) and east (Y ) components of the field (i.e., H ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ). According to Cole (1966), the observed H is a combination of currents systems flowing in the magnetosphere-ionosphere (MI) system and is given by where S R is the daily solar regular variation of the Earth's magnetic field associated to the regular ionospheric dynamo due to solar heating (Mayaud, 1965) and D is the integrated effects of disturbances coming from various current systems flowing in the Magnetosphere-Thermosphere system (Zaourar et al., 2017). Neglecting the effect of induced ground currents (Sabaka et al., 2004) as well as Chapman Ferraro currents (Chapman & Ferraro, 1931) and the tail currents in the presence of the generally strongest ring current, H can be written as where SYM − H × cosδ is the symmetric component of the ring current and δ is the geomagnetic latitude of the station. DP is the disturbance polar currents (Kamide & Fukushima, 1972;Nishida et al., 1966) and Ddyn is the ionospheric disturbed dynamo currents (Blanc & Richmond, 1980;Le Huy & Amory-Mazaudier, 2005).
The term DP + Ddyn is known as the ionospheric electric current disturbance (Diono) Zaourar et al., 2017). Diono combines the effects of (i) the disturbance polar no.1 (DP1), (ii) disturbance polar no.2 (DP2), (iii) the disturbance polar no.3 (DP3), and (iv) the disturbance polar no.4 (DP4), as well as ionospheric disturbed dynamo currents (Ddyn). DP1 is one cell current system on the nightside associated with substorm (Rostoker, 1967(Rostoker, , 1969, while DP2 is the one expanding from pole to equator due to convection electric field (Nishida, 1968). DP3 is a system of current flowing in the polar cap with direction opposite to that of DP2 (Kuznetsov & Troshichev, 1977;Troschichev & Janzhura, 2012). DP4 represents the current system of the disturbance related to the azimuthal component of IMF (Svalgaard, 1968). DP1, which is on the nightside, is negligible along with DP3 and DP4 that are restricted to the polar cap (Stauning, 2012). Based on these, at middle and low latitudes we can write DP2 is the equivalent current system due to PPEF (Nishida, 1968) and Ddyn is the equivalent current system associated to DDEF (Blanc & Richmond, 1980).
Using Equations 2 and 3, Diono can then be derived as Short-term oscillations of about 2 h associated with southward turning of IMF Bz are the signature of DP2 (Nava et al., 2016), while diurnal oscillations are attributed to Ddyn (Le Huy and Amory-Mazaudier, 2005). In the equatorial region, Ddyn is absent at the beginning of the storm since it requires a few hours with respect to the start of PPEF and/or storm onset (Abdu, 2012;Huang, 2013), typically 2-3 h at nighttime, to get to the low latitudes. In this situation, DP2 becomes significant and can be approximated to Diono. Also, when a magnetic quiet day immediately follows a storm and there is no auroral activity and by implication weak convection electric field that is different from that during the storm, DP2 becomes zero and Ddyn can also be approximated to Diono . Based on these, a running average filter that takes the mean value of 4 h of Diono data with sliding of 1 h was used for the separation of Ddyn from DP2 (Amaechi et al., 2018b;Azzouzi et al., 2015;Fathy et al., 2014). This method of separation and related assumptions remain valid only for short-duration PPEF (≤3 h) but not during cases of long lasting PPEF. The filter, nevertheless, is still useful in isolating Ddyn signal related to DDEF (which takes typically 2-3 h to reach the low latitude after the beginning of the main phase). We therefore employed IEFy to further identify long-lasting PPEF. As such, when IEFy is eastward there is PPEF; hence, DP2 occurs.
To gain more insight into variation of PPEF during the main phase of storms, the PPEFM was used to estimate PPEF in the African sector around longitude 37°E. This model is mainly a transfer function that models daily variations of equatorial ionospheric electric fields using IEF data mapped in the solar wind. Details about it can be found in Manoj and Maus (2012). The input parameters are time and location, while the output parameters are estimated values of equatorial electric field (EEF) mainly the (i) background electric field (i.e., quiet electric field obtained during geomagnetic quiet conditions) and (ii) the total electric field (i.e., the sum of quiet and prompt penetration electric fields).

10.1029/2020EA001183
Earth and Space Science AMAECHI ET AL.
GNSS observables in Receiver Independent Exchange format were subjected to quality check using the Translating Editing and Quality Checking software (Estey & Meertens, 1999). Relative Slant TEC (STEC) was thereafter estimated by leveling the carrier phase with the pseudorange measurements (Hansen et al., 2000). Prior to that, eventual cycle slips in the phase data were detected and corrected (Blewitt, 1990). Absolute STEC was derived from relative STEC by removing satellite and receiver biases (Sardon et al., 1994). This was finally converted to VTEC using a suitable mapping function with ionospheric pierce point assumed at a height of 350 km (Mannucci et al., 1993). Details about TEC processing technique and software used can be found in Seemala (2010) and Seemala and Valladares (2011). The elevation cutoff angle of 40°was adopted in order to reduce multipath (Amaechi et al., 2018a) as well as to reduce errors related with varying ionospheric pierce point due to potential ionospheric gradients that are characteristic of the low-latitude ionosphere (Rama Rao et al., 2006). The rate of change of TEC (ROT) was calculated and converted to the unit of TECU/min according to Equation 5.
The rate of change of TEC index (ROTI) was further computed as the standard deviation of ROT over 5 min ) (Pi et al., 1997). ROTI is a good proxy for scintillation index (S 4 ) (Basu et al., 1999). In this work, ROTI values for all available satellites above the elevation of 40°were averaged at a given epoch (5 min) (Amaechi et al., 2018a;Jacobsen & Dähnn, 2014) and a threshold of 0.5 TECU/min was set as the limit for the detection of irregularities (Ma & Maruyama, 2006).
The perturbation in TEC (ΔTEC), which is the difference between observed TEC and quiet TEC, was computed for longitude 37°with a latitudinal coverage of ±30°and resolution of 1 h × 1°(time/latitude). Quiet TEC was obtained by taking the average of VTEC during the five most geomagnetically quiet days in the month. The day-to-day variability was taken cared by computing the standard deviation of TEC during these quiet days. ΔTEC was, hence, utilized to examine the contribution of the asymmetry of the EIA and its potential association with differences in irregularities variations especially on 16-17 July 2012, 15 November 2012, and 19 March 2013 when such differences were pronounced. The selection of storm events was done based on (i) their intensity (SYM-H < −100 nT) and (ii) season of occurrence, and (iii) simultaneous availability of GNSS data over stations located in both hemispheres.

Results
In this section each storm event has been analyzed using the SYM-H index, solar wind parameters (IMF Bz and IEFy), ground-based magnetometer data (H), and the derived current systems (Diono and Ddyn), as well as ROTI over stations in the crests and trough of the African EIA. All the storms were CME driven except Case 1 which is a "wake of CME" (www.spaceweather.com). We included some days before and after the storms that acted as quiet time reference.

Storm Case Studies 3.1.1. Storm Period of 23-24 April 2012
From Figure 1, the Sudden Storm Commencement (SSC) (vertical dashed lines) occurred on 23 April (panel 1) while SYM-H reached a minimum value of −124.00 nT at 03:50 UT on 24 April. IMF Bz (second panel) went south with a minimum of −14.19 nT at 03:50 UT and turned north sharply. Thereafter, there was a long duration southward IMF Bz with a minimum of −15.30 nT at 17:40 UT. Prior to that there was another period of southward IMF Bz followed by a northward return at about 15:00 UT. Southward incursions were also observed on 24 and 25 April in the postsunset period. IEFy (third panel) was eastward with peak of 4.01 and 5.70 mV/m when IMF was southward on 23 April. The observed H (panel 4, red curve) had been superimposed on its quiet time reference level (blue curve) along with its day-by-day variation (light blue area). H clearly replicated the well-known regular pattern of the low-latitude Sq for Addis Ababa. However, it increased and fluctuated at the time of SSC and was clearly below the limit of day-by-day variation in the postsunset period on 23 April as well as at about 04:30 UT on 24 April. Oscillations in H were also observed in the postsunset period on 24 and 25 April. Diono increased at the time of SSC (panel 5, magenta curve) and fluctuated with minima on 23-25 April. Ddyn amplitude (panel 5, black curve) was undisturbed before the storm day. It decreased with minima on 23-25 April. (see Table 2 for the time of occurrence and magnitude of Diono and Ddyn minima). TEC irregularities were weaker over the trough and stronger over the crests with some hemispheric asymmetry especially before and after the main phase of the storm. On 23 April 2012, irregularities were inhibited over the trough (panel 7) and crests (panels 6 and 8) from 16:30 to 18:00 UT ( (Table 2). Weak irregularities were observed (10 and 12 November 2012) before the storm, with complete inhibition during the storm main phase (13 November) and first recovery day (14 November), followed by much stronger and asymmetric irregularities on 15 and 17 November.

10.1029/2020EA001183
Earth and Space Science 13:00 UT). In the postsunset, the enhancement was confined within the magnetic equator. On 15 November 2012, the perturbation was localized within the magnetic equator during noon albeit stronger in the Northern Hemisphere. The postsunset crests were enhanced and asymmetric. 19 March 2013 was relatively quiet, although noon enhancement could be observed in both hemispheres around the crests. However, there was only one crest in the postsunset period especially at about 17:00 UT (20:00 LT). The second panel of Figure 9 shows that ionization varied differently in the noon and postnoon periods before and after the storm. It was more enhanced with well-developed postsunset crest after the storm.

Discussion
It is well known that storm time PPEF and DDEF can affect the regular electric field (i.e., E × B drift) hence, the formation of ionospheric irregularities (Abdu et al., 2009(Abdu et al., , 2018. When IMF Bz turns south and convection electric field (IEFy) increases, PPEF whose signature is the DP2 signal can penetrate into the low-latitude ionosphere (Kikuchi & Araki, 1979).  Figure 6). The result of 23 April 2012 is particularly interesting giving that irregularities were earlier inhibited by westward electric field (Figure 1, green box). In line with our observations on 17 March, Kalita et al. (2016) found that short duration irregularities were triggered when IMF Bz turned south in the postsunset period over 100°E longitude. Also, Kassa and Damtie (2017) reported that irregularities were triggered in Bahir Dah (11°N, 38°E), Ethiopia, by an enhanced drift that had favored the postsunset lifting of the F layer (Joshi et al., 2015) to altitude where irregularities were generated by the R-T instability mechanism (Kelley, 1990). Similar triggering of irregularities by eastward PPEF with corresponding deep density depletions and EPBs was captured using the Defense Meteorological Satellite Program and ground-based GNSS, respectively, during the main phase of the storm of 13 September 2004 in Africa (Ngwira et al., 2013). In line with these observations, Zakharenkova and Astafyeva (2015) reported that eastward PPEF significantly enhanced CHAMP ROTI and increased the fluctuations level of ROT data from ground-based GNSS in Africa during the main phase of the storm of 30 August 2004.
Conversely, DDEF is more active during the recovery phase of a geomagnetic storm. The decay in H several hours after the beginning of the disturbance is the signature of Ddyn current system (Azzouzi et al., 2015) related to westward DDEF (Le Huy & Amory-Mazaudier, 2008). DDEF is driven by increased heating of the thermosphere at high latitude ensuing from the energy input during storms (Danilov & Lastovicka, 2001) and the resultant change in global circulation. Their presence in the postsunset period can inhibit the development of irregularities (Abdu et al., 1995).  (1-2 days). Although Ddyn lasted 6 days, its amplitude was well reduced from 20 to 22 March (Nava et al., 2016). The magnetometer at Addis Ababa also showed a reduction in Ddyn amplitude on 20 March while there was no data on 21 March. It is known that the necessary condition for the generation of postsunset irregularities is the rise of the F layer to higher height under the influence of PRE. It is thus, very likely that the weak Ddyn on 19-21 March 2015 was not efficient enough to prevent the rise of the F layer, which was controlled by higher drift velocity typical of months of high solar activity (Fejer et al., 1999).
Additionally, overshielding penetration electric field of westward polarity can occur when the southward IMF Bz is followed by a rapid turning to northward and there is rapid decrease in convection during the main phase (Kikuchi et al., 2000). Stressing further, Kikuchi et al. (2003) postulated that when R2-FACs build up following the rapid decrease in R1-FACs, due to the northward turning of the IMF and concomitant decreased in convection, electric field in the equatorial region reverse from eastward to westward under the so-called dominant shielding electric field (Kelley et al., 1979). From 15:00 to 18:00 UT on 23 April 2012, there were sharp fluctuations in IMF Bz with corresponding peak in westward IEFy. The westward overshielding electric field could have acted to suppress the PRE, hence inhibit irregularities from 17:00 to 18:00 UT. Another contributing factor could have been the existence of westward DDEF described in the previous paragraph. In fact, Chakraborty et al. (2015) reported enhancement of the AE index on 23 April 2012 at about 06:00 UT and 15:00 UT. These implied the presence of a heating source at high latitude that may have led to the development of DDEF on this day. Recently, Huang (2018) confirmed that it takes about 4.7 h after the onset of a storm for the effect of disturbance dynamo to reach the equatorial region. It is thus, inferred that westward electric field (overshielding electric field and DDEF) were responsible for the inhibition of irregularities from 17:00 to 18:00 UT on 23 April (Figure 1, green box).
The occurrences of irregularities near the trough cannot be explained by electric fields only. The PRE occurs in the postsunset when solar photoionization decreases rapidly; consequently, there are large density gradients thus, irregularities develop thanks to the R-T instability. Background density is, therefore, a crucial factor to reckon with during the formation of irregularities. The storm of November 2012 occurred during winter when irregularities are weak and their occurrence quite random (Akala et al., 2014). To examine the observed difference in irregularities before (10 and 12 November) and after the storm (15 and 17 November), we examined the TEC profile in the postsunset of 10 and 15 November (Figure 9, second panel). It was found that the crests were not well formed before the storm while they were well developed after it. Also, the dynamism of ionization on 10 November was not too different from that during quiet days. Therefore, the difference in irregularities could be attributed to background ionization in addition to recombination processes. Oppositely, the inhibition of irregularities during the storm's main phase (13 November) and first recovery day (14 November) was related to the presence of westward DDEF. Olwendo et al. (2015) identified westward electric fields using magnetic field variations (dH) during this storm.
The storm of June 2013 was particularly interesting with its long and smooth main phase, as well as slow changing dH/dt and asymmetric irregularities practically during all the days ( Figure 5). This storm occurred in June solstice during which nighttime to postmidnight irregularities are frequent (Akala et al., 2014;Yizengaw, Retterer, et al., 2013). In line with this, irregularities were observed from night to postmidnight before the storm. On 28 June during the main phase, they were triggered in the postsunset by the long duration PPEF. On 29 June, westward DDEF prevented them from occurring. On 30 June, however, the resurgence of postsunset irregularities might have been associated with the northward incursion of IEFy around 18:00 UT.
Fluctuations in Diono and Ddyn around 11 UT on 17 February when SYM-H, IMF Bz, and IEFy were completely quiet were related to the reduction in H slightly below the day-to-day variability. This interesting phenomenon could have been caused by (i) solar flare and/or (ii) westward electric field. In effect, four C class solar flares that were well above the background flux F B7. Ionospheric irregularities over Africa are less frequent in solstices because of the weaker drifts (Wiens et al., 2006;Yizengaw et al., 2014) associated with this season. They occur predominantly during equinox season (Oladipo et al., 2013;Paznukhov et al., 2012;Wiens et al., 2006) when there is a good alignment between the solar terminator and the geomagnetic meridian (Tsunoda, 1985). This scenario gives rise to increase in conductivity gradient, hence eastward electric field and the consequent optimum vertical drift needed to lift the F layer to altitude favorable for the development of irregularities (Eccles et al., 2015). Other studies have shown increase occurrence rate in March equinox than September equinox (e.g., Mungufeni et al., 2016;Oladipo et al., 2013;Olwendo et al., 2013) with some occurrences in summer (Akala et al., 2014). This seasonal behavior was well captured during our events, with stronger (weaker) irregularities during the April 2012 (July 2012/November 2012) and March 2013 (June 2013) events. During storm, nonetheless, this expected seasonal pattern can be altered significantly under the dictate of electric field and the intensity of the ring current as shown earlier.
As for solar activity, it was observed that strongest irregularities (ROTI > 3 TECU/min) occurred in February 2014 which also registered the highest solar flux (170.3 sfu) while weakest irregularities (ROTI < 0.5 TECU/ min) occurred in June 2013 with corresponding lowest solar flux (110.74 sfu). This was a fairly reflection of the solar activity control of irregularities whereby higher solar flux was associated with stronger irregularities (Aarons, 1991) in line with the increase in postsunset drifts with solar flux (Fejer et al., 2008). Previous studies have examined the solar activity control of quiet time irregularities over Africa (Akala et al., 2014;Mungufeni et al., 2016). During storm, nevertheless, it was found that irregularities were generally inhibited during various solar cycle phases in line with past observations (Dugassa et al., 2020;Ngwira et al., 2013;Seba & Nigussie, 2016).
Previous work by Amaechi et al., (2018aAmaechi et al., ( , 2018b had presented the effect of storms on irregularities near the magnetic equator. The present study emphasized on the simultaneous behavior of irregularities over the crests in both hemispheres and investigated plausible mechanism responsible for their behavior. Irregularity occurrence over the crests is quite complex and least studied over the African EIA. Ordinarily, EPBs generated at the magnetic equator extend along the magnetic field lines and the crests. Generally, irregularities were strong over the crests and weak over the trough during most of our events mainly because of the larger background ionization at the crests. Additionally, most TEC irregularities occured at roughly the same time over the crests and trough, indicating that they were from the same origin. However, there were differences in the irregularity behavior over the crests especially on 17 July 2012 and on 19 March 2013, which will be examined in the next section along with the TEC perturbation profile. From Figure 2, it is evident that PPEF and DDEF acted on 16 July 2012. The eastward PPEF had modulated the enhancement in TEC from 06:00 to 12:00 UT (Figure 9, first panel). On the other hand, westward DDEF acted to reduce irregularities over all stations till 22:00 UT. TEC irregularities, nonetheless, reappeared at the southern crest and trough but not at the northern crest from 22:00-24:00 UT on this day. From Figure 9 (first panel) maximum ionization was confined within the magnetic equator in the postsunset. On 15 November 2012, the presence of DDEF (with reduced amplitude) and the observed weak background ionization at the magnetic equator accounted for the presence of weak irregularities at the trough. ΔVTEC profile in the postsunset period showed enhancement in ionization an obvious hemispheric asymmetry in TEC (Figure 9, middle panel) that could have been responsible for the difference in TEC irregularities over both crests. On 19 March 2013, however, magnetic conditions were relatively quiet yet there was only one postsunset crest in the Southern Hemisphere (Figure 9, panel 3). This observed hemispheric asymmetry is indication of the presence of transequatorial neutral wind and other processes, such as composition change during the recovery phase. Maruyama and Matuura (1984) showed that various forms of plasma transport by wind from one hemisphere to another can affect the conductivity, thus the instability growth rate. Nicolls et al. (2006) had noted that even in the presence of westward electric field, a contribution from meridional equatorward wind of the order of about 30 m/s and plasma movement driven by latitudinal gradient in electron density could lead to an uplift of the F layer to heights where the growth rate can trigger irregularities. The influence of wind and asymmetry of the EIA crests, as well as perturbations from lower atmosphere, on the behavior of irregularities during the recovery phase of storms still require further investigations over Africa.
First result for the assessment of the capability of the PPEFM over the African longitude revealed that the model is capable of reproducing enhancement and reduction in the PRE caused by eastward and westward PPEF as well as the corresponding effect on the behavior of irregularities during storms. However, the model could not accurately reproduce long duration PPEF. One of the plausible reasons could be that during events characterized by long duration IMF Bz, DDEF might be active in addition to PPEF. A contributing factor could have also been the magnitude of auroral activity during such events (Huang, 2019). Nayak et al. (2016) had used the PPEFM to highlight the effect of eastward PPEF/westward DDEF on the PRE, thus on the generation/inhibition of irregularities in the Indian/Taiwanese sectors during the storm of 17 March 2015.
Validating the PPEFM in the present study further reinforces the role of modeling in increasing our understanding of storm time electric field effect on irregularities over the low-latitude African sector.

Conclusions
The behavior of ionospheric irregularities over the crests of the African EIA in both hemispheres has been studied and the predictive capability of PPEFM along longitude 37°E assessed during intense geomagnetic storms. It was found that: 1. Ionospheric irregularities over the magnetic equator and crests of the EIA were simultaneously suppressed by westward DDEF for over 1 h on 23 April 2012 and subsequently triggered by an eastward PPEF from about 21:00 LT. 2. Similarly, irregularities were triggered by eastward PPEF that occurred at about 21:00 LT on 20 February 2014 as well as 1 h earlier (20:00 LT) on 17 March 2013. They were inhibited thereafter (by westward DDEF) over the trough and crests. 3. The duration of irregularities inhibition during the recovery phases was related to the amplitude and duration of the magnetic perturbation Ddyn that lasted longer during the superstorm of March 2015. 4. There was a hemispheric asymmetry in irregularities strength over the crests that might have been linked to the asymmetry in the magnitude and position of the EIA crests over Africa. In particular, the southern crests experienced irregularities on 16 July 2012 while the northern crest and trough did not. Also, irregularities were stronger (weaker) over the southern ( A better understanding of irregularities over the African EIA is, however, limited by the absence of observational tools such as incoherent scatter radar that could have given us more insight into the variations of electric field as well as the contribution of perturbation originating from the lower atmosphere during these events.