Research

In this paper, we presented the e ﬀ ect of moderate geomagnetic storms on the TEC variation at the Koudougou station (Geo Lat 12 ° 15 ′ N; Geo Long: -2 ° 20 ′ E) in Burkina Faso (Africa) during the descending phase of solar cycle 24. For this purpose, four moderate geomagnetic storms without storm sudden commencement (SSC) or sudden impulse (SI) that occurred on May 13, 2015 (Dst: -76 nT), June 08, 2015 (Dst: -73 nT), September 11, 2015 (Dst: -80 nT), and May 08-09, 2016 (Dst: -88nT), were considered. These moderate storms were found to be associated with transients induced by fast solar winds. At the Koudougou station, TEC variation shows a positive response to the di ﬀ erent moderate geomagnetic storms studied, with increases of order of 2-21 TECU around 1300-1500 UT except for September 11, 2015, TEC variation which shows sometimes negative responses at a few hours (mainly at night). TEC increases observed are a function of geomagnetic parameter (magnitude and polarity) variation. Storm-induced electric ﬁ eld and neutral winds are the main drivers of TEC changes observed during the selected geomagnetic storms. In addition, it was found that the TEC peak on storm day behaves di ﬀ erently compared to the days before and after the storm depending on whether Dst is positive or negative before southward inversion. Indeed, a TEC small peak relative to the days before and after the storm is observed when Dst is negative before southward inversion, and a larger peak occurs in the opposite case. The reasons for these di ﬀ erences are not investigated in this paper.


Introduction
Sun is the solar system main source of energy. It continuously releases energy into interplanetary space in electromagnetic radiation form and charged particles that are responsible for the Sun-magnetosphere-ionosphere dynamics [1]. A variety of physical phenomena are associated with space weather, including geomagnetic storms, geomagnetic activity, ionospheric disturbances, flickers, auroras, and Earth-induced telluric currents [2]. Geomagnetic storms result from solar wind's interaction energy transferred to the Earth's magnetosphere through magnetic reconnection [3]. They are generally caused by coronal mass ejections (CMEs) from the Sun [4][5][6] and corotating interaction regions (CIRs) created by the interaction between slow and fast solar winds from coronal holes [7][8][9].
According to Perreault and Akasofu [10], geomagnetic storms can be defined as magnetosphere's response to intense solar wind flow impact in which intensity and direction of magnetic field vary in a complex manner. Conventionally, several geomagnetic indices are used to assess the strength of a geomagnetic storm. However, for equatorial regions, the most commonly used indices are (i) the disturbance storm time (Dst) index [11][12][13] and (ii) the Kp index which is an integer between 0 and 9 [13][14][15]. Gonzalez et al. [16] define three classes of geomagnetic storms according to their intensity based on the Dst index: (i) weak storms characterized by −50 nT < Dst ≤ −30 nT, (ii) moderate storms determined by −100 nT < Dst ≤ −50 nT, and (iii) intense storms with Dst ≤ −100 nT. Other authors such as Loewe and Prölss [17] even speak of severe storms when Dst ≤ − 200 nT. Here, our study concerns moderate storms.
Geomagnetic storm's main effect on the magnetosphere is the injection of many energetic ions and electrons from the tail, significantly increasing ring current. Depending on whether Earth's magnetosphere reaction time is fast or slow, two types of storms have been detected: (i) sudden storm commencement (SSC) [18][19][20] and (ii) progressive onset geomagnetic storms [21,22]. These latter are also called recurrent storms [23][24][25]. This paper discusses these types of geomagnetic storms during solar cycle 24 descending phase.
Geomagnetic storms can significantly alter ionosphere and have significant negative effects on space and ground systems [26]. During storms, solar wind/magnetosphere coupling leads to increased Joule heating. This results perturbations in the composition, temperature, density, and winds of the upper atmosphere [27][28][29] that influence aerospace systems and associated human technologies. Examples of activities directly impacted are radio wave propagation, sending signals between satellites and Earth, control of communication and navigation systems, etc. [30,16]. Therefore, it is important to know ionosphere in more depth and to study its behavior.
In communication and navigation systems' case, one of the ionospheric parameters with a dominant influence on system performance is total electronic content (TEC). GPS signals' range errors are directly proportional to TEC; therefore, any variation in this is a major concern [31]. Thus, several ionospheric models have been developed to better understand these variations. The best known are, for example, Klobuchar model [32], SAMI model [33], NTCM model [34], NeQuick model [35,36], or IRI (International Reference Ionosphere) model [37].
In the past, many studies have been conducted on storm effects on the ionosphere at high and midlatitudes [38,39] and low latitudes [40][41][42][43][44]. Pedatella et al. [45] investigated the TEC variation during December 15, 2006, storm over Pacific Ocean region using multi-instrument data. Rama Rao et al. [46] studied the TEC variation at different latitudes over the Indian sectors and geomagnetic storms impact on navigation systems by considering two successive storms that occurred between November 8 and 12, 2004. Kumar et al. [47] reported that electric field induced by the storm can trigger the growth of the Rayleigh-Taylor instability and consequently the development of plasma bubble. Singh et al. [48] studied four intense geomagnetic storm effects on low-latitude TEC during ascending phase of solar cycle 24. An analysis of ionospheric TEC from GPS, GIM, and global ionospheric models during moderate, strong, and extreme geomagnetic storms over the Indian region was done by Reddybattula et al. [49].
However, the complex processes that occur in the upper atmosphere during geomagnetic storms make accurate modeling difficult. Existing models can only provide monthly averages of actual variability, especially during periods of magnetic quiet [27]. Geomagnetic phenomena's complexity makes modeling hard during disturbed period complex, hence interest in studying response of Koudougou station' TEC data during disturbed periods in general and particularly during moderate geomagnetic storms.
Data and analysis method are presented in Section 2. Section 3 concerns results and discussion. Conclusion is presented in Section 4. Solar wind parameters have been used to evaluate Sun's contribution on geomagnetic storms. These parameters are available on the OMNIWeb database (https://omniweb.gsfc .nasa.gov/form/dx1.html). Solar wind parameters such as speed (SSW), temperature (SWPT), and plasma pressure (SWP) as well as magnetic field magnitudes (B Z component and total magnetic field B T ) were used. Also the data of y component of electric field (IEF Ey) also used in this work are downloadable on the same site.

Data and Analysis
Geomagnetic indices were used to select geomagnetic activity different days according to existing classes. These indices are available on the International Service of Geomagnetic Indices (ISGI) website (http://isgi.unistra.fr/data_ download.php). Indices used in this work are (i) Dst index and (ii) Kp index. Dst index indicates horizontal component's hourly variation of Earth's magnetic field [3]. It is storm intensity indicator but is not a geomagnetic activity's good tracer in usual sense, for which Kp index should be considered instead [50]. Interplanetary index Kp indicates geomagnetic activity's level. It varies between 0 and 9 according to the intensity of terrestrial magnetosphere's disturbance. But Kp on OMNIWeb base has been subject to a special treatment, explained in the OMNIWeb site at https://omniweb.gsfc.nasa.gov/html/ow_data.html.

Methods of Analysis.
In the present study, recurrent geomagnetic storm day's selection was performed using a pixel diagram constructed from Kp OMNI data. Each line represents a Bartels rotation, the double black circles represent SSC dates, and the single black circles represent SI dates. The method was first used by Legrand and Simon [23] with the geomagnetic index values aa for geomagnetic activity day classification. It was then improved by Ouattara and Amory-Mazaudier [24] and a little later by Zerbo et al. [25]. More recently, the authors have appreciated Kp index's reliability compared to the aa index. Kp index combined with other indices such as Dst or SymH is an excellent indicator for 2 International Journal of Geophysics geomagnetic activity day selection [51]. It is used for the selection of international calm and disturbed days in the month [52]. Figure 1 shows an example of a pixel diagram constructed using Kp OMNI. Recurrent days are identified by Kp ≥ 30 spanning one or more Bartels rotations without SSC or SI (sudden impulsion). The moderate storm day selection criteria are (i) −100 nT < Dst ≤ −50 nT, (ii) SWPT ≥ 105 K, and (iii) SSW ≥ 500 km/s); (iv) day selected must not be preceded by an SSC or SI. Taking these criteria into account, four storms were selected. Table 1 contains the dates of the selected storms and Kp OMNI and Dst indices maximum and minimum values and Dst indices, respectively. TEC measurements' relative deviation ðδTECÞ on disturbed days compared to those on calm days is calculated using the following equation: where TEC S is TEC for stormy day and TEC Q is daily average TEC for month quiet days considerate. Sandwidi and Ouattara [53] used this method to study the impact of recurrent events on foF2 critical frequency at Dakar station (Lat: 14.8°N, Long: 342.6°E, Senegal).
On May 14, 2015, first day after the storm, Dst index returns to the normal variation of a calm day but still negative with a first low value (-33nT) at 0500 UT and a second (-32nT) at 1600 UT. This characterizes the substorm signature. The TEC increased sharply mainly during the daytime hours with a peak of 78.12 TECU at 1400 UT, either an increment of 21.51 TECU. However, the B Z is north-facing and the IEF Ey is west-facing. Also, low values of solar wind temperature and pressure compared to the day of the storm are observed. Thus, it is clear that this TEC enhancement is neither associated with the PPEF nor with the dynamic solar wind pressure. These enhancements could be due to the storm-induced neutral wind effect. This phenomenon may result from the superposition of two substorms successive indicated by two successive southward Dst perturbations. These results are consistent with the results reported by Jin et al. [29] for the March 2015 geomagnetic storm. May 15, 2015, is marked by a positive TEC of 17.34 TECU. During this time the B Z is oriented to the south and Ey to the east. Therefore, the possible cause of this increase could be the effect of the PP electric field. The variation of TEC on 09 September 2015 is less than regular variation on calm days between 0000 UT and 1100 UT. This is despite a southern orientation of B Z and an eastern orientation of IEF Ey, respectively, during these hours. According to Fejer [55], Huang et al. [56], and Singh et al. [54], when the IMF B Z is south oriented and the dawndusk component of interplanetary electric field (IEF) Ey computed by Zhao et al. [60] is east oriented, a probable increase in TEC is expected. This is not the case here. It is noticed that solar wind speed is low (less than 450 km/s), as well as solar wind temperature, but a weak oscillation of solar wind pressure is noted from 0000 UT to 1300 UT. The possible cause of this TEC depression could be the slow decay rate of ring current observed during the main phase of this storm shown by the Dst index. After the negative variation, a positive TEC with a very small increase (2 TECU) is observed between 1200 UT and 1500 UT. This corresponds to periods of intense sunshine at the Koudougou station. Since the PP electric field is in the same direction as the ambient electric field (IEF Ey towards the east), it pushes up the E × B drift towards higher altitudes [61]. Due to the larger production/loss ratio at higher altitudes, the TEC increases during sunlight hours (B. [62]). Reddybattula et al. [49] had also found a positive TEC during the main phase of September 09, 2015, storm over the Indian region. The TEC becomes negative again from 1700 UT to 0500 UT of September 10. This period corresponds to the night hours at the Koudougou station. Since at night, the PP electric field and the ambient field are of opposite polarity [54], which could be the cause of this depression.
September 10, 2015, located on the recovery phase of September 09 storm, has a higher TEC peak than the day of September 09 and 10. The TEC is negative during the night hours (0100 to 0500 UT and 1900 to 2300 UT) and during hours of low sunshine (0600 to 0900 UT) and positive during hours of high sunshine (between 1000 and 1700 UT) at the Koudougou station. During the whole day of September 10, B Z and Ey oscillate between positive and negative values. This does not give a clear direction of IMF  International Journal of Geophysics has shifted southward with a decrease of 12.7 nT (2.1 nT to -10.8 nT), and the IEF Ey shows an eastward orientation from 0700 to 1300 UT. Indeed, the increase in the TEC observed between 1100 and 1700 UT may be associated with the PP electric field caused by the direct and rapid penetration of electric field at dawn [63]. The onset of September 11 storm is also marked by an increased growth of solar wind speed and proton temperature. However, a decrease in pressure is observed. This could also justify the TEC increase by the Joule heating effect of thermospheric neutral winds. The low pressure rate of solar wind dynamics could be the cause of the small variation of TEC observed. From 1600 UT on September 11, Dst index started to turn northward and reached its calm day background level at 1900 UT on September 12. Thus, the storm is in its recovery phase. As on previous days, the TEC is positive during the daytime hours and negative during the nighttime hours but with a peak (47.21 TECU) higher than on September 11 (42.56 TECU). However, B Z shows fluctuations, sometimes southward, sometimes northward. IEF Ey undergoes the same orientation variation but from east-west. The effect of neutral wind induced by storm could play on this increase. On September 13, 2015, the TEC curve is almost confused with the curve of the regular variation of calm days except between 1300 and 1600 UT that a notable difference is observed.
In general, May TEC values at the Koudougou GPS station are higher than those of September. This difference could be justified by the seasonal influence on variation of TEC [64]. During the study of these two events (storm of 11-15 May 2015 and storm of 09-13 September 2015) at the Koudougou station, it is found that when Dst index turns southward at the beginning of storm without changing sign, the TEC peak of day following storm is always higher than that of storm day. At the Koudougou station, the difference varies between 2 and 5 TECU depending on the storm. This shows the important contribution of the winds induced by storms on the variation of TEC, since these moments are much more influenced by these winds.  Figure 4(a) shows that Dst index started to turn southward at 0300 UT on 08 June 2015 and reached the lowest value of -73 nT at 0800 UT. It is also observed that IMF B Z turned south at 0400 UT and remained in this direction until 0800 UT when it reached a value of -14.6 nT. At the same time, IEF Ey values ranged from -1.63 to 6.74 mV/m (8.37 mV/m increment). Subsequently, IMF B Z turned south again at 1400 UT and reached the lowest value of magnitude -5.6 nT. It remained southward until 1700 UT on June 08. Meanwhile, IEF Ey turned eastward and changed from 3.78 mV/m (-0.13) to 3.65 mV/m. The rapid upward and downward changes in IMF B Z show the presence of a sudden storm commencement (SSC) [54]. This is not the case here as no SSC was reported on this day on the International Service of Geomagnetic Indices (ISGI) site. This shows that gradual onset storms can also have rapid recursions on IMF B Z . The storm main phase is marked by a rapid increase in solar wind parameters.
The Dst variation shows that 06 and 07 June 2015 are very calm geomagnetic days with Dst turned northward and of positive values. Despite this, a positive TEC is observed during these days mainly during the daytime hours. However, the exact mechanism responsible for the increase in ionospheric electron density prior to the storm is still a matter of debate [65,66]. The TEC amplitude on 07 June is larger than that on 06 June, and the cause could not be the effect of PP electric field because the B Z magnetic field faces north and the IEF Ey faces west from 0800 UT to 1900 UT on 07 June 2015.
The TEC variation is positive all day on June 08, 2015, at the Koudougou station, even during the night hours. Nevertheless, it is during the daytime hours that a strong increase is observed, i.e., a magnitude of 15.13 TECU at 1400 UT. During this day, Dst is facing south; IMF B Z is facing south from 0400 to 0800 UT and then from 1200 UT to 1700 UT. At this time, IEF Ey is facing east. These indications show that the increase in TEC could be associated with the convection electric field in the magnetosphere described as the primary source of the PP electric field [54]. The high peak in the TEC on this day could also be the contribution of effect heating caused by the high solar wind temperature or the high solar wind dynamic pressure, since an enhanced growth of solar wind parameters is observed. The increase of TEC during the storm of 08 June 2015 at the station of Koudougou is evaluated at 34.43%, and that of the day before and after the storm is evaluated, respectively, at 16.84% and 26.82%.
On 08 June around 1800 UT, the Dst index started to turn northward and reached its calm day background level at 0800 UT on 09 June 2015. Thus, a long recovery phase was maintained for the 08 June 2015 storm. However, a transient behavior of Dst index is observed during the recovery phase where it showed twice a south polarity, the first at 1800 UT on June 09 (-42 nT) and the second at 1500 UT on June 10 (-31 nT). The north-south oscillation of IMF B Z and Dst index is associated with an increase of solar wind speed which reaches 650 km/s at 1800 UT on June 10. This shows the low intensity substorm condition of 09-10 June 2015. These substorm processes could be associated with the increases in TEC amplitudes observed over almost the entire period [57]. The large increases in TEC observed from 09-10 June during daylight hours may be due to the effect of neutral wind induced by the 08 June storm.  International Journal of Geophysics 26.47% and 33.05%, respectively, from the level of quiet day. The increase in TEC was likely associated with the PP electric field as confirmed by the change in IEF Ey value during the increase in TEC values. However, a time interval of 3 to 4 hours is required to observe any disturbance in the vertical drift of IEF [61].
Before the event of 08-09 May 2016, a rise in solar wind speed values is observed on 07 May at 0200 UT where it reached 521 km/s. But well before this date, a peak of 8.14 nPa is observed in SWP values on 06 May at 1900 UT. During this time, the B Z is turned southward between 1100 and 1800 UT, but the variation of IEF Ey shows a westward orientation except at 1800 UT where it is directed eastward. These observed characteristics could be the condition of passage of a fast solar wind. All day on May 07, the IEF Ey was facing west and the IMF B Z and the Dst are oriented to the north. This could justify the low amplitude of TEC observed this day. Indeed, if the IMF B Z is northward, there is no reconnection between the IMF and the Earth's magnetic field during the day. The discontinuity of SWP curve between 0000 and 0600 UT on May 06 is due to the lack of data recorded in the OMNIWeb database.
On May 10 and 11, 2016, TEC shows positive variation from the quiet day especially during the daytime hours but with a small amplitude compared to those of 08-09 May. During these two days, IEF Ey is almost westward with no polarity inversion and IMF B Z is northward with a southward direction on May 10 from 1400 to 1700 UT and on May 11 from 1000 to 1300 UT. However, the likely cause of the increased TEC on these days is the effect of storminduced neutral winds.
The global study of events of 08 June 2015 and 08-09 May 2016 shows an almost identical behavior of TEC. Indeed, during these storms, TEC is positive during almost the whole day but weak during the night hours. The Dst shifts to the north direction at storm beginning before turning south during the main phase. The observation is that the TEC amplitude on the day of storm is greater than on the day after storm. The opposite effect is observed at the May 13 and September 11, 2015, storms. Thus, this behavior of TEC can be explained by the behavior of Dst index.

Conclusions
We examined solar wind parameters' roles in the formation of a few selected recurrent type moderate geomagnetic storms during the solar cycle 24 descending phase and the response of equatorial GPS-TEC at the Koudougou station to these storms. All of the moderate geomagnetic storms studied were found to be associated with manifestations of fast solar winds (SSW ≥ 500 km/s). These storms are unique in terms of the Dst index behavior prior to the onset of event (negative in May 13 and September 11, 2015, storms and positive in June 08, 2015, and May 08-09, 2016, storms). Our results show TEC increase as a function of geomagnetic parameter (magnitude and polarity) variation and local weather at the Koudougou station, and TEC slightly decreases during the September 11, 2015, storm at a few hours of the day over the period of September 09-13, 2015.
The results of this work also show that the storm-induced electric field and neutral winds are the main drivers of observed TEC changes during the selected geomagnetic storms. However, some differences were observed in the behavior of TEC as a function of variation of Dst: (1) when the Dst goes through positive values (north direction) before reversing to negative values (south direction) at the storm beginning, the IMF is southward and the solar wind pressure is increased during the main phase of storm. And the TEC amplitude observed on storm day is higher than that on the days before and after the storm. (2) But when the Dst passes to the south direction without sign change at storm beginning, IMF B Z and IEF Ey oscillate without a specific direction during the main of storm. At this time, the TEC peak during the storm is small compared to the days before and after the storm. The physical phenomenon associated with such TEC behavior is not revealed, but future studies will justify this. Of all the geomagnetic storms considered, the positive ionospheric TEC effect was more pronounced during the June 08, 2015, event and much less pronounced during the September 11 event. Specifically, a moderate geomagnetic storm of recurrent type increases the maximum TEC at the Koudougou station by about 2-21 TECU.
This study provides an overview of the origins and progression of moderate recurrent geomagnetic storms and their impacts on the total electron content (TEC) of the ionosphere at the Koudougou station, an equatorial station located in West African region; this will help to develop mechanisms for predicting the response of African equatorial TEC to different geomagnetic storms.

Data Availability
The total electron content (TEC) data used to support the findings of this study are included within the supplementary information file(s). The solar wind parameter (speed, pressure, and temperature) data used are available at the OMNIWeb site (https://omniweb.gsfc.nasa.gov/form/dx1 .html). The Bz component and B scalar of the interplanetary magnetic field data used are available at the OMNIWeb site (https://omniweb.gsfc.nasa.gov/form/dx1.html). The geomagnetic index (Dst) data used are available at the ISGI site (http://isgi.unistra.fr/data_download.php). The geomagnetic index (Kp) data used are available at the OMNIWeb site (https://omniweb.gsfc.nasa.gov/form/dx1.html). The solar sunspot number index data used are available at the SILSO site (https://www.sidc.be/silso/datafiles).

Conflicts of Interest
The authors have not declared any conflict of interests.