foF2 Seasonal Asymmetry Diurnal Variation Study during Very Quiet Geomagnetic Activity at Dakar Station

This paper aims to study the foF2 seasonal asymmetry diurnal variation at Dakar station from 1976 to 1995. We show that equinoctial asymmetry is less pronounced and somewhere is absent throughout 21 and 22 solar cycles. The absence of equinoctial asymmetry may be due to Russell-McPherron mechanism and the vertical drift E × B. The solstice anomaly or annual anomaly is always observed throughout both 21 and 22 solar cycles as measured at Dakar ionosonde. The maximum negative value of σfoF2, fairly equal to -65%, is observed during the decreasing phase at solstice time; this value appeared usually at 0200 LT except during the maximum phase during which it is observed at 2300 LT. The maximum positive value, fairly equal to +94%, is observed at 0600 LT during solar minimum at solstice time. This annual asymmetry may be due to neutral composition asymmetric variation and solar radiation annual asymmetry with the solstice time. The semiannual asymmetry is also observed during all solar cycle phases. The maximum positive value (+73%) is observed at 2300 LT during solar maximum, and its maximum negative (-12%) is observed during the increasing phase. We established, as the case of annual asymmetry, that this asymmetry could not be explained by the asymmetry in vertical velocity E × B phenomenon but by the axial mechanism, the “thermospheric spoon” mechanism, and the seasonally varying eddy mixing phenomenon.

The paper outline is as follows: the second section concerns data and methods of analysis, the third section treats our results and discussions, and a conclusion ends it.

Data and Methods
2.1. Data. For this paper, the data involved are (1) foF2 values carried out at Dakar station for ionosphere time variation. This station is operated from 1950 to December 1996. The involved data interval for our study is 1976-1995 and concerned SC 21 and SC 22; (2) Zurich sunspot number (Rz) for the influence of solar cycle phase; and (3) Mayaud [25][26][27][28] geomagnetic index aa for geomagnetic activity impacts. For this study, foF2 diurnal variation is analyzed by considering seasonal impacts.

Method for Determining the Geomagnetic Activity.
Based on the strong correlation between geomagnetic index Aa and solar wind velocity established by [18], it is wellknown, according to [29][30][31], that disturbed activities are characterized by Aa > 20 nT and are classified into three groups: (1) recurrent activity due to both the solar high wind stream and the corotating interaction region (CIRs), (2) shock activity due to the coronal mass ejections (CMEs) and the magnetic clouds, and (3) fluctuating activity; consequences of the fluctuation of solar heliosheet and quiet activities are characterized by Aa < 20 nT. For well-known geomagnetic quiet activity impact over a season, we do not use quiet days as the reference level of foF2 time variations as previously done by [32,33] but use the quietest days. It is important to note that, for a given period, [28] points out that the quietest days are obtained by taking Aa < 10 nT. Aa daily values are obtained from the following website: http:// isgi.unistra.fr/data_download.php.

Equinox and Solstice Month Determination.
Over a year, the seasons are considered as (1) spring (March, April, and May), (2) summer (June, July, and August), (3) autumn (September, October, and November), and (4) winter (December, January, and February). The present paper concerns the equinoctial month's:

Data Analysis
Method. This paper is aimed at studying the seasonal asymmetry observed in foF2 daily time variation during equinoctial and solstice time at Dakar station, throughout the SC 21 and SC 22.
The morphological or qualitative estimation of (1) equinoctial asymmetry observed in March-April and September-October equinoxes, respectively, noted by M-A and S-O, and (2) solstice asymmetry or semiannual asymmetry, observed in June-July and December-January solstices, noted, respectively, by J-J and D-J, is given by the following equation: where σ is the error bars putting in data graphs and V is the variance defined by ð1/NÞ∑ N i=1 ðx i − xÞ 2 with x mean value and N the total number of observations for a particular dataset.
The quantitative estimation of the asymmetry is given by σfoF2 expressed as where x F i concerns foF2 hourly mean value for March equinox (M-A) and June solstice (J-J) and x L i is devoted for September equinox (S-O) and December solstice (D-J).

Results and Discussion
3.1. Results. Figures 1-4 concern foF2 and σfoF2 diurnal variation during the four solar cycle phases, minimum, ascending, maximum, and descending, respectively, for the solar cycles SC 21 and SC 22. The left column is devoted to foF2 diurnal profile and the right one for relative deviation diurnal variation. "  Figure 1(a) presents a foF2 profile (left column) with double peaks, a trough located at 1200 LT and an afternoon peak amplitude higher than the morning one. A night-time peak is observed at 2300 LT in the S-O curve. This observation is the same as that of [8] at Korhogo station. The S-O curve is above that of M-A between 1900 LT to 2100 LT and after 2200 LT to before 0000 LT with respect to error bars shown in the M-A graph: then, the equinoctial asymmetry is highlighted from this observation. In fact, the σfoF2 curve (right column) shows that the S-O foF2 values are always greater than those of M-A, except after 0100 LT to before 0300 LT, during daytime (after 0800 LT to before 1000 LT and from 1400 LT to 1700 LT) and during night time (after 2100 LT to before 2300 LT). The σfoF2 minimum value (-38%) is seen at 2300 LT, and its maximum value (+46%) is seen at 2200 LT. Figure 1(b) shows a foF2 profile (left column) with two peaks and a trough located at 1200 LT for the D-J curve. The morning peak amplitude (located at 1000 LT) is higher than the afternoon one located at 1400 LT. In the J-J curve, we observe an afternoon peak at 1700 LT. This graph 2 International Journal of Geophysics highlights solstice asymmetry due to the different morphology of profile; in fact, J-J foF2 values are greater than those of D-J from 0500 LT to 0700 LT and they are lower than the D-J foF2 values from 0800 LT to 1500 LT and 2000 LT to 0500 LT with respect to error bars shown in the J-J graph. σfoF2 values (in (b)) highlight that D-J foF2 values are always greater than those of J-J except after 0500 LT to 0700 LT and from 1600 LT to after 1900 LT. σfoF2's minimum value (-58%) is seen at 0200 LT, and its maximum value (+94%) is observed at 0600 LT. Figure 1(c) shows a foF2 profile (left column) with two peaks and a trough located at 1200 LT for both curves. The morning peaks' amplitude (located at 1000 LT for solstice month and 1100 LT for equinox time) is smaller than that   International Journal of Geophysics of the afternoon located at 1700 LT. The semiannual asymmetry is highlighted because the equinox curve is always above that of the solstice, mostly during daytime. σfoF2 values highlight that equinox foF2 values are always greater than those of solstice except during night time (2200 LT et de 0000-0100 LT) and before sunrise (0500 LT). The maximum negative value (-9%) is observed at 2200 LT, and its maximum positive value (+30%) is observed closely to the afternoon peak meantime (1600-1700 LT). Figure 2 is devoted to increasing phase. Figure 2(a) concerns the graphs of equinox months. The foF2 diurnal profile curves show double peaks with a trough at 1200 LT and an afternoon peak greater than the one of the morning. Only the S-O graph presents a night-time peak at   Figure 2(b) shows solstice month curves. Three peaks are observed with two troughs (one located at 1200 LT and another one at 1600 LT) for the D-J curve. The morning peak amplitude (located at 1000 LT) is higher than the one of the afternoon. The J-J curve shows an afternoon peak at 1700 LT, one hour after the second-afternoon peak of D-J. Only the D-   International Journal of Geophysics J graph presents a night-time peak at 0100 LT. The difference between both profiles highlights the asymmetry of the solstice; in fact, JJ's foF2 values are higher than those of DJ from 0500 LT to 0700 LT and they are lower from 0800 LT to 1500 LT and from 2000 LT to 0500 LT with respect to error bars shown in JJ's graph. σfoF2 values show that D-J foF2 values are always greater than those of J-J except from 0500 LT to 0700 LT and from 1600 LT to after 1900 LT. σfoF2's maxi-mum negative value (-63%) is observed at 0200 LT, and its maximum positive value (+70%) is observed at 0600 LT. Figure 2(c) shows a foF2 profile with two peaks and a trough located at 1200 LT for both curves, but the trough of equinox months is not very marked like the one of solstice month. The morning peak amplitude (located at 1000 LT) is smaller than that of the afternoon located at 1700 LT. Then, the semiannual asymmetry is highlighted because the   International Journal of Geophysics equinox curve is always above that of the solstice, mostly during daytime (1000-1800 LT). Besides, σfoF2 values show that equinox month foF2 values are always greater than those of solstice except around 0000 LT and before sunrise (0300-0500 LT). σfoF2's maximum negative value (-12%) is observed at 0300 LT, and its maximum positive value (+26%) is observed at 2300 LT. Figure 3 concerns maximum phase curves. Figure 3(a) shows equinox month graphs. The S-O curve presents double peaks with a trough at 1200 LT, and the one of M-A shows a morning peak. Only the S-O graph presents a night-time peak at 2300 LT, and it is important to note a lack of value observed at 2300 LT for the M-A graph. The equinoctial asymmetry observed during night time is more expressed by the difference in terms of values from 1900 to 0100 LT and from 0300 to 0500 LT with respect to error bars shown in the M-A graph. σfoF2 values are negatives from 0000 to 0200 LT, around 0700 LT and from 1500 to 2200 LT. Thus, M-A foF2 values are lower than those of S-O during these periods and they are greater before sunrise (0300-0600 LT) and during daytime (0800-1400 LT). σfoF2's maximum negative value (-28%) is observed at 2200 LT, and its maximum positive (+19%) is seen at 0400 LT. Figure 3(b) concerns solstice month's curves. Three peaks are observed with two troughs (first at 1700 LT and second at 2000 LT) for the D-J curve. This curve's morning peak amplitude (14.27 MHz), observed at 1000 LT, is higher than that of afternoon and night. In the J-J curve, an afternoon peak at 1600 LT can be observed. D-J's curve shows a night-time peak at 2100 LT, but the J-J one does not. This figure highlights solstice asymmetry due to the difference of both profiles; in fact, J-J foF2 values are lower than the D-J foF2 values all the time, except after sunrise (0500-0700 LT) with respect to error bars shown in the J-J graph. In fact, σfoF2 graph highlights that D-J foF2 values are always greater than those of J-J except after 0500 LT to 0700 LT. σfoF2's minimum value (-57%) is seen at 2300 LT, and its maximum value (+21%) is observed at 0600 LT. Figure 3(c) shows a foF2 profile with two peaks and a trough located at 1400 LT for the equinox month curve; the morning peak amplitude (13.78 MHz), located at 1100 LT, is fairly equal to the afternoon one located at 1700 LT. On the other hand, the solstice month curve presents a profile like a plateau with a light trough at 1200 LT. The equinox month curve shows a night-time peak (15.32 MHz) at 2300 LT, but one of solstice months does not. The semiannual asymmetry is highlighted because the equinox curve is always above that of solstice all the time except during the sunrise (0600-0700 LT) and the sunset (1900-2000 LT). σfoF2's values highlight that foF2 values of equinox months are always higher than those of solstice months because their values are positive with a maximum (+73%) observed at 2300 LT. Figure 4 concerns decreasing phase curves. Figure 4(a) presents equinox month curves. Both curves show an afternoon peak with a slight trough observed at 1300 LT and a night-time peak at 0300 LT. The M-A graph is always above that of S-O, except during night time (2300 LT-0100 LT). This situation shows equinoctial asymmetry which is more expressed during day time (0800-1600 LT) and during night time (0000 LT-0100 LT and 0200-0500 LT) with respect to error bars shown in the M-A graph. Besides, σfoF2 values are positives during all the time, except from 1900 to 2000 LT and from 2300 to 0100 LT. Therefore, M-A's foF2 values are lower than those of S-O during these periods and they are greater during the left time. The σfoF2 minimum value (-26%) is observed at 0100 LT, and the maximum one (+34%) is observed at 0200 LT. Figure 4(b) concerns solstice month curves. Three peaks are observed with two troughs (one located at 1400 LT and another one at 2200 LT) for the D-J curve. The morning peak amplitude, located at 1100 LT, is higher than the afternoon one. The J-J curve shows an afternoon peak at 1800 LT. D-J's graph presents a night-time peak at 2300 LT, but the J-J one does not. The difference of both profiles' morphology highlights solstice asymmetry; in fact, J-J foF2 values are lower than the D-J foF2 values during all the time, except during sunrise (0600-0800 LT) with respect to error bars shown in the J-J graph. σfoF2's values are always negative except at sunrise (0600-0700 LT) with a maximum negative value (-65%) observed at 0200 LT and a maximum positive one (+39%) observed at 0700 LT. Figure 4(c) presents equinox and solstice month graphs. All foF2's profiles show an afternoon peak with a slight trough located at 1300 LT. Both graphs present a nighttime peak, observed at 0200 LT for solstice months and at 0300 LT for equinoxes. The equinox curve is always above that of solstice all the time, except around 0200 LT and during the sunrise (0500-0600 LT); hence, the semiannual asymmetry is highlighted. σfoF2's values highlight the observation that the foF2 values of equinox months always remain higher than those of solstice months at all times except around 0200 LT and at sunrise (0500-0600 LT). The maximum positive value (+48%) is observed at 0000 LT, and the maximum negative one (-14%) is observed at 0500 LT.
3.2. Discussion. It emerges from this study that in general, equinoctial asymmetry is less pronounced at Dakar station [35] and somewhere is absent during day time for all solar cycle phases. Equinoctial asymmetry is due to an asymmetry of ionization during both equinoxes, despite the fact that solar zenithal angle kept the same characteristics during both equinoxes. This subject has been widely discussed in recent publications [8, 19-21, 36, 37]. Many mechanisms have been suggested to explain this asymmetry such as ionospheric electric field variation, geomagnetic activity, solar wind change through the equinoxes, thermosphere chemical composition asymmetric variation, and Russell-McPherron mechanism. [13] at Niamey station and [8] at the one of Korhogo suggest that the equinoctial asymmetry is mainly due to the Russell-McPherron mechanism for all seasons (spring and autumn) at those African equatorial anomaly regions of the ionosphere. Then, one can expect to have "no equinoctial asymmetry" during the equinox months.
But, [6] suggests that, even if the Russell-McPherron impact is the main explicative mechanism of equinoctial asymmetry, it remains important to find other mechanisms which could explain this phenomenon. From their work on the satellite ROCSAT-1 observation from 1994 to 2004, [21,38] suggest that equinoctial asymmetry is due to the asymmetry in vertical drift velocity E × B at the ionosphere of equatorial latitude. They also suggest that the asymmetry in the PRE mechanism may be related to the asymmetry in the eastward thermosphere winds. Nevertheless, we observe a PRE asymmetry because it occurs during autumn and not during spring. Then, vertical drift velocity E × B at the equatorial latitude ionosphere has also been suggested to explain the equinoctial asymmetry as suggested by [20].
The solstice asymmetry, called winter anomaly, is observed throughout all solar cycle phases. σfoF2's maximum negative value (-65%) is observed during the decreasing phase, and its maximum positive one (+94%) is observed during the minimum phase. In general, the night-time peak observed in D-J's curve is not observed in J-J's graph. Besides, D-J's curve (winter) is over the J-J's one (summer); then, the winter anomaly is obvious and it remains during night time contrary to the conclusion of [39]. Some authors suggest that this asymmetry may be due to neutral composition variation [8,[39][40][41][42][43][44][45]. According to [43], this change may induce an enhancement of the O/N 2 ratio due to convection of atomic oxygen from the summer hemisphere to the winter one. Otherwise, for [46,47], this abnormal variation of the neutral composition of the ionosphere's F region may be associated with the seasonal variation of the thermosphere winds. According to [48], asymmetric variations of aeronomic parameters between winter and summer, the main ones being the thermospheric parameters, the intensity of EUV radiations, the recombination rate β, and the atomic oxygen concentration [O] are responsible for the winter anomaly phenomenon. If according to [23], this anomaly is due to the intensity of solar radiation intensity due to change in the Earth-Solar distance between December solstice (Earth is at perigee) and June solstice (Earth is at apogee), the flux of solar ionization radiation cannot be the only factor as concluded by [22]. Our results show that the difference between winter and summer is mainly due to the amplitude of solar high-speed wind. As this type of wind can be a source of interplanetary corpuscular radiation, then, the annual asymmetry observed may be due to [24]'s process.
The semiannual asymmetry is observed during the day over all solar cycle phases. It is more observed during night time for maximum and decreasing phases than the minimum and increasing phases. This phenomenon is characterized by a predominance of ionization during equinoctial months than during solstices. Numerous researches have been done concerning it. [49] suggests that at the equatorial ionosphere latitudes, it may be explained by asymmetry observed in vertical drift velocity E × B. Nevertheless, we observe an absence of asymmetry in the vertical drift velocity E × B; thus, the vertical drift velocity E × B cannot be the source of the semiannual anomaly. Otherwise, if [50] suggests that it can be explained by the variation of the terrestrial high atmosphere temperature variation, according to [40,51], it is due to semiannual variation of neutral atmosphere density at low latitudes with the geomagnetic activity. The axial mechanism suggested by [52] contributes widely to explain the semiannual anomaly. In fact, during equinox months, the Earth is exposed to the solar high wind stream because of its heliographic latitude of about 7°. Further, [53] suggests the existence of a phenomenon called "thermospheric spoon" which is an interhemispheric circulation on a global scale at the solstice. It would act like a huge turbulent vortex mixing the major thermosphere's species. These vortices tend to reduce the oxygen atomic density of solstice months compared to the equinox. This phenomenon causes atmosphere "compression" at solstices which may explain the semiannual anomaly. In addition, from the analysis of satellites TIME-D/GUVI's observations and model studies (e.g., TIE-GCM), [54] shows that there is a diffusion of vortices in the mesopause region. These vortex diffusions are called the vortex mixing phenomenon, being more important during solstices than equinoxes; they would therefore be one of the causes of semiannual asymmetry.

Conclusion
This paper shows that the equinoctial asymmetry is less pronounced at Dakar station. The solstice asymmetry (annual anomaly) is observed during all solar cycle phases. The semiannual asymmetry is observed over all solar cycle phases; it is more expressed during the night for maximum and decreasing phases than during minimum and increasing phases. The maximum negative value of σfoF2 (-65%) is observed at the decreasing phase, and its maximum positive one (+94%) is observed at the minimum phase. In general, night-time peaks observed during the months of D-J and S-O are not observed in those of J-J and M-A, respectively. Thus, there is an equinoctial, annual, and semiannual asymmetry in the PRE mechanism.

Data Availability
The sunspot data used to support the findings of our study are available at http://www.sidc.be/silso/datafiles. The geomagnetic aa index data are available at http://isgi.unistra.fr/ data_download.php.

Conflicts of Interest
The authors declare no conflicts of interest.