the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Aug 2026
Influence of solar, solar wind, and geomagnetic activity on the ionospheric propagation factor M(3000)F2 over Ouagadougou in the African equatorial sector
Abstract. This study examines how solar, solar wind, and geomagnetic activity influence the ionospheric propagation factor M(3000)F2 over Ouagadougou (12.4° N, 358.5° E; dip latitude +1.45°), an equatorial station in the African sector. Hourly ionosonde observations from 1976 to 1997, spanning Solar Cycles 21 and 22, were analysed to investigate diurnal, seasonal, solar cycle, and storm-time variability of M(3000)F2. Representative years of high (1991), moderate (1993), and low (1995) solar activity were selected for climatological analyses, while seven geomagnetic storm events were examined to assess storm-time responses. Relationships between annual mean M(3000)F2 and selected solar, solar wind, and geomagnetic parameters were evaluated using correlation and linear regression analyses. M(3000)F2 exhibits pronounced diurnal and seasonal variability, characterised by higher nighttime and early-morning values, lower daytime values, and distinct equinoctial and solstitial differences associated with equatorial electrodynamics and thermospheric dynamics. An inverse dependence on solar activity is observed, with the highest M(3000)F2 values occurring during solar minimum. The strongest correlation is with the solar radio flux F10.7 (R = -0.810, R² = 0.657, p < 0.001), followed by the interplanetary magnetic field magnitude (B) (R = -0.689, R² = 0.475, p < 0.001) and the disturbance storm time (Dst) index (R = 0.527, R² = 0.277, p = 0.0117). Relationships with the planetary (Ap) index, the southward component of the interplanetary magnetic field (Bz), solar wind dynamic pressure (Psw), and solar-wind speed (Vsw) are weak and not statistically significant. During geomagnetic storms, M(3000)F2 generally decreases during the daytime main phase and recovers within one to two days, depending on storm intensity and background ionospheric conditions. The results indicate that long-term variability of M(3000)F2 over the African equatorial sector is primarily controlled by solar activity, with geomagnetic disturbances providing secondary modulation during storm periods. These findings contribute to a better understanding of equatorial ionospheric variability and provide useful information for HF radio wave propagation, empirical ionospheric modelling, and space weather applications.
- Preprint
(2439 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 05 Oct 2026)
-
RC1: 'Comment on egusphere-2026-4526', Anonymous Referee #1, 09 Sep 2026
reply
-
AC1: 'Reply on RC1:Response to Referee Comment', Abdullahi Kazeem, 15 Sep 2026
reply
Response to Referee Comment
Response to Comment 1: We thank the referee for highlighting the importance of local-time data coverage. We have clarified the temporal treatments used in the study and quantified the observational coverage. Daily M(3000)F2 values were calculated from all valid hourly observations available for each day. Monthly values were used to examine the relationship between M(3000)F2 and F10.7, and annual mean values were used for the correlation and regression analyses.
Hourly data coverage was substantially higher during the daytime than at night and in the early morning. For the representative years 1991, 1993, and 1995, the mean daily data coverage was 62.83%, 63.80%, and 70.71%, respectively. The corresponding coverage during 06–18 LT was 88.05%, 89.29%, and 95.64%. Thus, the incomplete nighttime and early-morning observations primarily limit the continuity and completeness of the 24-h records.
To evaluate the potential effect of incomplete daily sampling on the climatological means, we performed a sensitivity analysis using only days with at least 75% hourly data coverage. The resulting annual mean M(3000)F2 values differed from those obtained using all available days by 1.27%, 3.47%, and 1.19% for 1991, 1993, and 1995, respectively. The maximum absolute difference in monthly means was 6.20% in 1991, 3.97% in 1993, and 2.95% in 1995. These results indicate that incomplete daily sampling has a relatively modest effect on annual mean values, although it may affect some monthly variability.
Missing observations were not interpolated. For the storm-time analysis, incomplete nighttime and early-morning observations resulted in discontinuous temporal coverage. Therefore, the storm analysis was deliberately restricted to the observed daytime interval, using 06–18 LT as the nominal interval, with actual coverage varying slightly among events. The storm-time coverage within 06–18 LT was 88.05%, 89.29%, and 95.64% for 1991, 1993, and 1995, respectively. We have therefore clarified throughout the manuscript that the storm-time results represent the observed daytime storm-time response and should not be interpreted as a complete 24-h response.
Comment 2: The modified Lloyd seasonal grouping needs physical justification, significance testing against day-to-day variability, and clearer justification of the seasonal scheme.
Response: We thank the referee for this valuable comment. We have clarified the rationale for adopting the modified Lloyd seasonal classification and have treated it as an analytical framework for resolving seasonal variability rather than as an assumption of sharply separated physical regimes. The classification separates the two equinoxes and solstitial periods and has been used in previous studies of equatorial ionospheric variability (Eleman, 1973; Rabiu et al., 2007; Kazeem et al., 2020). We have also clarified that the seasonal means presented in Fig. 3 and Table 2 were calculated from all available hourly M(3000)F2 observations within the months assigned to each season.
In addition, to assess whether the observed seasonal differences were significant relative to day-to-day variability, we derived daily mean M(3000)F2 values from the available hourly observations and compared the four seasonal distributions using the Kruskal-Wallis test. The differences were statistically significant for 1991 (H = 32.81, p = 3.54 × 10⁻⁷), 1993 (H = 71.94, p = 1.64 × 10⁻¹⁵), and 1995 (H = 26.65, p = 6.96 × 10⁻⁶). These results provide statistical support for the presence of seasonal differences beyond day-to-day variability.
We have also corrected an error in the original description of the seasonal maxima and minima. The December solstice has the highest seasonal mean in 1991 and 1995, whereas the September equinox has the highest value in 1993. The lowest value occurs at the June solstice in 1991 and at the March equinox in 1993 and 1995. Accordingly, we have revised the text to avoid implying a consistent seasonal maximum or minimum across all solar activity conditions.
Comment 3: The monthly time series are not perfectly inverse; for example, during Solar Cycle 21 around 1980, F10.7 reaches a maximum while M(3000)F2 appears to increase around 1981. Please examine possible lagged relationships and discuss whether this indicates a delayed ionospheric response. Also, the conclusion that F10.7 is the “dominant driver” should be moderated.
Response: We thank the referee for this important observation. We agree that the relationship between F10.7 and M(3000)F2 is not perfectly inverse at all times and that departures from the overall relationship may reflect temporal variability and the coupled nature of solar, solar wind, and geomagnetic forcing.
In response, we have added a lagged Pearson correlation analysis between the annual mean F10.7 solar flux and annual mean M(3000)F2 for the 1976–1997 dataset. Correlations were evaluated for lags from −3 to +3 years to investigate whether a delayed annual association was stronger than the contemporaneous relationship. The results show that the zero-lag correlation is the strongest (r=−0.810, R2=0.657, p<0.001), while the preceding one-year lag is very similar (r=−0.804). The correlation at a one-year positive lag is weaker (r=−0.719), and the association decreases substantially at longer positive lags. Thus, the analysis provides no evidence that a delayed annual relationship is stronger than the contemporaneous association, although the similarity between the zero- and 1-year correlations indicates that the annual relationship is not strictly instantaneous and should not be interpreted as evidence of a simple one-to-one response.
Comment 4
Response: We thank the referee for this comment. We have clarified that the storm-time analysis is a case-study analysis of seven selected events during the representative years 1991, 1993, and 1995, rather than a comprehensive statistical assessment of storm-time behaviour. The storm events were selected based on their geomagnetic activity and the availability of sufficient M(3000)F2 observations. We also clarified the treatment of missing observations and excluded storm phases where substantial data gaps prevented reliable interpretation.
Because nighttime and early-morning observations were less complete, the storm analysis was restricted to the observed daytime interval (06–18 LT), for which data coverage was approximately 88–96% across the three representative years. The manuscript now consistently refers to these results as daytime storm-time responses, rather than complete 24-h responses. Finally, the Discussion and Conclusion have been revised to restrict the interpretation of the seven events and avoid generalising the observed responses to all geomagnetic storms.
Citation: https://doi.org/10.5194/egusphere-2026-4526-AC1
-
AC1: 'Reply on RC1:Response to Referee Comment', Abdullahi Kazeem, 15 Sep 2026
reply
-
RC2: 'Comment on egusphere-2026-4526', Anonymous Referee #2, 20 Sep 2026
reply
see attach file
-
AC2: 'Reply on RC2', Abdullahi Kazeem, 02 Oct 2026
reply
Response to Referee 2
We thank the referee for the careful and constructive comments. We revised the manuscript accordingly, paying particular attention to the consistency of the statistical analysis, the interpretation of physical mechanisms, figure presentation, and the scope of our conclusions.
Comment 1 – Longitudinal comparison
Response: We agree that the analysis did not sufficiently support the previous comparison with other longitudinal sectors. We have removed this comparison and restricted the interpretation to Ouagadougou and the low-latitude West African ionosphere.
Comment 2 – Use of annual means
Response: Annual means are used to examine interannual associations between M(3000)F2 and annual solar, solar-wind, and geomagnetic parameters. This averaging reduces short-term variability, while monthly and seasonal analyses are retained to examine shorter-timescale variability. The treatment of dispersion has also been clarified where appropriate.
Comment 3 – Modified Lloyd classification
Response: The modified Lloyd classification uses an analytical framework for grouping observations into equinoctial and solstitial intervals rather than as a claim of four sharply separated physical seasons. The corresponding month ranges are now explicitly stated in the manuscript.
Comment 4 – Disturbed conditions and storm selection
Response: We now explicitly define disturbed conditions using Kp≥4Kp and Ap≥16Ap. The seven selected storms were chosen as representative case studies based on their Dst evolution, geomagnetic intensity, and availability of suitable M(3000)F2 observations. They are therefore not intended to constitute a storm climatology.
Comment 5 – Excluded storm observations
Response: The main phase of the 4–7 April 1993 storm was excluded from detailed interpretation because of substantial gaps in the M(3000)F2 observations. The other selected events were retained where sufficient daytime observations were available, and no interpolation was introduced to fill missing values.
Comment 6 – Figure 1
Response: We thank the referee for this helpful comment. Figure 1 has been replotted using 15 instead of 30 contour levels and a more gradual colour scale, while retaining the same colour-bar scale for all three years. The differences among 1991, 1993, and 1995 have been retained because they represent the observed variability under high-, moderate-, and low-solar activity conditions.
Comment 7 – Extreme values in 1993
Response: The extreme M(3000)F2M(3000)F2 values in 1993 were checked against the archived observations and processing records. No interpolation was used to generate missing values, and the revised text avoids overinterpreting isolated extrema in periods with incomplete sampling.
Comment 8 – Organisation of Section 3
Response: Section 3 has been reorganised to separate the diurnal, seasonal, interannual, and storm-time analyses more clearly. This restructuring avoids mixing results obtained at different temporal scales.
Comment 9 – Solar activity relationship in the seasonal discussion
Response: The inverse relationship between M(3000)F2 and F10.7 has been moved from the seasonal discussion to the interannual solar-activity subsection. The seasonal section now focuses specifically on the variability within the modified Lloyd intervals.
Comment 10 – Physical interpretation of seasonal variability
Response: The discussion of possible physical mechanisms has been made more cautious. The observed seasonal behaviour is discussed in the context of known equatorial ionospheric dynamics, but we do not attribute the observed variations quantitatively to individual electrodynamic or thermospheric mechanisms because the corresponding measurements are not available in this study.
Comment 11 – Selection of months in Figure 3
Response: Seasonal means were calculated using all available hourly observations within the four defined intervals: March–April, May–August, September–October, and November–February. Individual months were therefore not selected arbitrarily; the observed maxima and minima also vary among the three representative years.
Comment 12 – Contribution relative to existing models.
Response: We do not claim that the inverse solar activity relationship is itself a new physical relationship. The contribution of this study is the long-term, station-specific observational characterisation of M(3000)F2 at Ouagadougou, including its associations with solar, solar wind, and geomagnetic parameters, lagged solar activity relationships, seasonal and diurnal behaviour, and storm-time responses. The results are presented as observational evidence that complements, rather than replaces, empirical ionospheric models.
Comment 13 – Storm modelling and selection of seven events
Response: We agree that M(3000)F2 alone cannot provide a complete reconstruction of the storm-time vertical ionisation profile. The seven events are therefore treated as case studies illustrating observed responses of the propagation factor during different storm phases rather than as evidence for a universal storm-time response.
Comment 14 – Selection of IMF parameters.
Response: We focus on IMF magnitude B and Bz because B represents the overall IMF strength, whereas Bz is directly relevant to solar wind–magnetosphere coupling. Bx and By were not analysed separately because their geoeffective interpretation depends on magnetic field orientation and coordinate system; their omission is therefore a scope choice rather than an assumption that they are physically unimportant.
Comment 15 – “More comprehensive assessment” and regional uniqueness
Response: We have removed statements implying that the study provides a uniquely comprehensive assessment or that the behaviour is specific to a particular African longitude sector. The contribution is now framed as a multi-parameter observational assessment at Ouagadougou, with particular relevance to a region where long-duration ionospheric observations remain relatively sparse.
Comment 16 – PPEF, DDEF, and thermospheric circulation
Response: The discussion now distinguishes between prompt electric field effects, particularly during the initial and main phases of storms, and later disturbance-dynamo and thermospheric/neutral-wind responses that may become important during the recovery phase. These mechanisms are presented as possible contributors rather than as uniquely identified causes of the observed variations.
Comment 17 – Generalisation to the African equatorial sector
Response: We agree that observations from a single station cannot be generalised to the entire African equatorial sector. The conclusions have therefore been restricted primarily to Ouagadougou and the low-latitude West African ionosphere.
Comment 18 – Data availability and provenance
Response: The data sources and provenance have been clarified. The M(3000)F2 observations are identified as archived ionosonde data from Ouagadougou, while the solar activity, solar wind, and geomagnetic parameters are obtained from the NASA SPDF/OMNI database. Information on the availability and source of the Ouagadougou observations has also been clarified.
Comment 19 – Overall contribution and HF propagation.
Response: The manuscript now emphasises that the main contribution is a long-term observational characterisation of M(3000)F2 at Ouagadougou across interannual, lagged, diurnal, seasonal, and storm-time scales. Although M(3000)F2 is relevant to the maximum usable frequency of HF propagation, this study is not a direct HF propagation campaign or a complete validation of an ionospheric model; rather, it provides station-specific observational information that can contribute to the evaluation of empirical representations.
General revision
We have also revised the manuscript for consistency and scientific precision, distinguishing observed variability from proposed mechanisms, statistical association from causality, station-specific results from regional generalisation, and storm case studies from storm climatology. We have additionally reduced repetitive descriptive text and strengthened the quantitative interpretation of the results.
Citation: https://doi.org/10.5194/egusphere-2026-4526-AC2
-
AC2: 'Reply on RC2', Abdullahi Kazeem, 02 Oct 2026
reply
-
RC3: 'Comment on egusphere-2026-4526', Anonymous Referee #3, 26 Sep 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-4526/egusphere-2026-4526-RC3-supplement.pdf
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 100 | 67 | 77 | 244 | 51 | 67 |
- HTML: 100
- PDF: 67
- XML: 77
- Total: 244
- BibTeX: 51
- EndNote: 67
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript “Influence of solar, solar wind, and geomagnetic activity on the ionospheric propagation factor M(3000)F2 over Ouagadougou in the African equatorial sector” analyses long‑term ionosonde observations from Ouagadougou (1976–1997) together with solar, solar‑wind, and geomagnetic indices. The author investigates diurnal, seasonal, solar‑cycle, and storm‑time variability of M(3000)F2 and uses linear regression to quantify relationships between yearly mean M(3000)F2 and several geophysical drivers.
The topic is relevant to HF propagation, empirical ionospheric modelling, and equatorial space‑weather studies, and the long data record from a data‑poor region is valuable. However, several methodological and interpretational issues need clarification before the manuscript can be considered for publication.
Major comments
Daily values are defined as arithmetic means over all hourly M(3000)F2 observations, which are then averaged to obtain monthly and yearly means. At the same time, the author explicitly notes frequent missing nighttime and early‑morning observations and restricts the storm‑time analysis to daytime hours. This implies that “daily averages” may in practice be daytime‑weighted means rather than true 24‑hour averages, which can bias monthly and yearly means at an equatorial station where M(3000)F2 shows strong nighttime and early‑morning maxima. The author should quantify local‑time data coverage, demonstrate whether missing hours affect long‑term averages, and consistently reflect these limitations in the climatological (diurnal, seasonal, solar‑cycle) analysis as well as in the storm‑time section.
Seasonal variability is analysed using a modified Lloyd classification: March equinox (March–April), June solstice (May–August), September equinox (September–October), and December solstice (November–February). While this scheme is referenced to earlier work, the manuscript does not clearly articulate a physical hypothesis for why these four seasons are the most appropriate framework for M(3000)F2 at a near‑equatorial station. The results show December solstice as having the highest mean M(3000)F2, with more modest differences among other seasons, yet the manuscript speaks of “marked seasonal variability”. The author should (i) explicitly state the expected seasonal differences and their physical basis, (ii) assess whether seasonal differences are statistically significant compared to day‑to‑day variability, and (iii) clarify why this particular seasonal grouping is optimal in the equatorial context.
The time series of monthly mean M(3000)F2 and F10.7 for 1976–1997 and the strong negative correlation of yearly means (R = −0.81, R² = 0.657) support an overall inverse relationship, and the author concludes that F10.7 is the dominant driver of long‑term M(3000)F2 variability. However, there are extended intervals where the behaviour is not strictly inverse—for example, around the maximum of Solar Cycle 21, F10.7 peaks around 1980 while M(3000)F2 rises from a minimum towards a higher value around 1981, during persistently high F10.7. These intervals suggest phase lags or additional influences and should be explicitly discussed. I recommend examining lagged correlations, highlighting periods where the inverse behaviour breaks down, and moderating the conclusion about F10.7 as the “dominant driver”, framing it as the parameter with the strongest statistical association among several coupled processes.
The storm‑time analysis is based on seven geomagnetic storms in three representative years (1991, 1993, 1995), with some phases missing due to data gaps. Despite this small and incomplete sample, the manuscript generalises that geomagnetic disturbances “generally” cause daytime depletion during the main phase and recovery over one to two days. These generalisations appear too strong for a limited set of case studies. The author should clearly state the selection criteria for storms, emphasise the case‑study nature of the results, and either expand the storm sample (if possible) or restrict conclusions to the analysed events without implying a robust climatology. Given that the analysis is restricted to daytime hours, the text should consistently refer to “daytime storm‑time response” and avoid suggesting that full diurnal behaviour has been characterised.
Recommendation
The manuscript addresses an important topic and makes good use of a valuable long‑term dataset, but the issues above require substantial revision. I therefore recommend publication after major revision, provided that the author can address the concerns regarding data gaps and averaging procedures, the physical motivation and significance of seasonal variability, the interpretation of the M(3000)F2–F10.7 relationship, and the limitations of the storm‑time analysis.