the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Jun 2026
Latitudinal Variation of Ionospheric Total Electron Content During the 17 March 2015 Geomagnetic Storm: A Multi-Station GNSS Analysis
Abstract. We investigated the latitude-dependent ionospheric response to the St. Patrick’s Day geomagnetic storm of 17 March 2015 (Dst =−234 nT, Kp = 8) using GNSS-derived vertical total electron content (VTEC) from four stations spanning mid-to auroral latitudes (34–70° N) in the European–African sector: RABT (Rabat, Morocco), MADR (Madrid, Spain), BRUX (Brussels, Belgium), and TRO1 (Tromsø, Norway). VTEC was derived from dual-frequency pseudorange measurements with differential code bias corrections. Storm-time VTEC responses showed clear latitude-dependent variability. Relative enhancements ranged from +46 % at TRO1 to +224 % at BRUX, while the absolute peak VTEC decreased systematically with latitude. Linear regression of peak storm-time VTEC against geographic latitude yielded a gradient of −1.61±0.06 TECU per degree (R2 = 0.9968, p = 0.001578). Correlation analysis revealed statistically significant positive associations between hourly ∆VTEC and Kp at RABT and MADR, while BRUX showed a weaker positive but non-significant correlation under the adopted p < 0.01 criterion. These results quantify the latitudinal structuring of storm-time ionospheric variability in the European–African sector and highlight stronger positive responses at the mid-latitude stations than at auroral latitude during this event.
- Preprint
(3119 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2994', Anonymous Referee #1, 09 Jul 2026
-
AC1: 'Response to Referee #1 (RC1)', Nouhaila Bouhadi, 14 Sep 2026
We thank the Referees for their careful and constructive reviews. We have revised the manuscript to address every point raised. Below, each Referee comment (RC1) is followed by our response (AC) and the corresponding location in the revised manuscript.
RC1 — General comment
"The manuscript presents an analysis of the latitude-dependent ionospheric VTEC response to the 17 March 2015 geomagnetic storm... I recommend moderate revision to improve the clarity of the manuscript, and strengthen the discussion of novelty, data reproducibility, and the limitations of the analysis."
AC: We thank the Referee for this assessment. During revision, we performed an additional verification of the GNSS processing workflow and updated the station-level VTEC series accordingly. The revised tables, figures, and statistical results are based on a corrected and internally consistent processing applied uniformly to all four stations. We believe that this additional round of internal quality control strengthens the reproducibility and robustness of the reported results.
RC1 Comment 1 — Novelty of the study
"Could the authors clarify more explicitly what the principal novelty of the study is in comparison with previous investigations of this storm, particularly regarding the integrated multi-station analysis of the European–African sector and the quantitative assessment of the latitudinal gradient of peak storm-time VTEC?"
AC1: We agree and have clarified this explicitly in the Introduction.
Manuscript location: Sect. 1 (Introduction), penultimate paragraph:
"Although the St. Patrick’s Day storm of 17 March 2015 has been extensively documented, previous studies have mainly emphasized global responses, isolated regional sectors, or specific ionospheric manifestations, leaving the storm-time VTEC response in the European–African sector insufficiently constrained as an integrated latitudinal system. In contrast, the present work examines this sector as a coherent station transect within a unified statistical framework combining quiet-day baseline removal, storm-time enhancement analysis, regression of peak VTEC versus latitude, and correlation with geomagnetic indices."
This distinguishes the present study from prior regional or single-station analyses of this event (Astafyeva et al., 2015; Nava & Rodríguez-Zuluaga, 2016; Paul et al., 2018) by providing a single, internally consistent multi-station transect and an explicit quantitative latitudinal gradient (Sect. 4.3), rather than qualitative or piecemeal regional comparisons.
RC1 Comment 2 — Data sources and data processing (quality control)
"Could the authors provide additional details on the quality-control procedures applied to these datasets. How missing or low-quality GNSS observations were handled, how temporal consistency among the different datasets was ensured. Are there any limitations of using publicly available data. Did this affect the robustness of the derived VTEC estimates and subsequent statistical analyses?"
AC2: We have expanded the description of our quality-control procedures and added an explicit discussion of the limitations associated with using heterogeneous public GNSS station data.
Manuscript location: Sect. 3.3 (Station-Level VTEC Computation and Temporal Resampling), which already described the elevation cutoff (E ≥ 30∘), the minimum-satellite-count filter (N(t) < 4 excluded), and the treatment of missing hours as gaps rather than interpolated values, now also includes:
"Because the GNSS observations used in this study are publicly available IGS station data, receiver type, antenna configuration, tracking performance, and local multipath environment may differ from one station to another. These instrumental and site-dependent differences represent a limitation of the use of heterogeneous public GNSS data. Their impact was reduced by applying a common processing strategy to all stations, including DCB correction, an elevation-angle cutoff, epoch-level satellite-count filtering, hourly median aggregation, and the retention of missing values as gaps rather than interpolated data.”
Temporal consistency across stations is ensured by resampling all four station series to the same hourly UT grid and pairing them with Kp/Dst using an identical procedure (Sect. 3.3–3.4). The robustness of the corrected VTEC estimates is independently supported by the CODE GIM validation (Sect. 4.1, Table 2), which shows strong agreement at all four stations (r = 0.937–0.986,
RMSE = 1.71–4.25 TECU).
RC1 Comment 3 — Latitudinal gradient robustness
"The study reports a strong linear relationship between peak storm-time VTEC and geographic latitude based on four GNSS stations. Could the authors discuss this in details."
AC3: We agree that, with only four stations (two degrees of freedom), the reported R² should be interpreted cautiously, and we have added a quantitative sensitivity analysis rather than a qualitative caveat alone.
Manuscript location: Sect. 4.3 (Storm-Time VTEC Enhancements and Latitudinal Gradient), added paragraph:
"To further assess the sensitivity of the fitted gradient to the limited number of stations, a leave-one-out regression test was performed by repeating the latitude–VTEC fit after removing one station at a time. The resulting slopes remained negative in all cases, ranging from approximately −1.60 to −2.02 TECU per degree, with R2 values remaining close to unity. This indicates that the decreasing peak-VTEC trend with latitude is not produced by a single station alone. Nevertheless, because the regression is based on only four stations and one storm event, the fitted gradient should be interpreted as a first-order sectorial estimate rather than a universal storm-time scaling law.”
This shows the negative latitudinal trend is not an artifact of any single station, while still appropriately qualifying the generality of the fitted slope, consistent with the Referee's concern.
RC1 Comment 4 — Choice of geomagnetic indices (Kp vs. −Dst)
"The correlation analysis shows that Kp is more strongly associated with ΔVTEC than −Dst. Could the authors elaborate on the physical reasons for this difference and discuss whether other geomagnetic or solar-wind parameters, or interplanetary electric field components, might provide additional insight into the regional ionospheric response?"
AC4: We have expanded Sect. 4.4.2 with both the physical interpretation of the Kp/−Dst asymmetry and a discussion of complementary solar-wind/IMF parameters.
Manuscript location: Sect. 4.4.2 (Correlations with −Dst):
"... This weaker relationship likely reflects the fact that Dst primarily describes the intensity of the ring current. In contrast, ionospheric TEC variability in the European–African sector is additionally influenced by local time, latitude-dependent electrodynamics, thermospheric composition changes, and, particularly at auroral latitude, particle precipitation and auroral energy input.”
In this context, direct solar-wind and interplanetary magnetic field parameters would provide a more physically specific description of the storm-time drivers than Kp or Dst alone. In particular, IMF Bz, the solar-wind motional electric field Ey, and the auroral electrojet AE index are more direct proxies for magnetospheric convection, prompt-penetration electric fields, and auroral particle-precipitation forcing. Their inclusion would therefore help separate the respective roles of large-scale geomagnetic activity, electric-field penetration, and high-latitude auroral forcing in the regional TEC response. This represents a natural extension of the present correlation framework in future work."
RC1 Comment 5 — Code and data availability
"...could the authors provide the analysis scripts, processing workflow, and relevant documentation in a publicly accessible repository, e.g. GitHub or Zenodo?"
AC5: We have made the full processing pipeline publicly available.
Manuscript location: new section, "Code and data availability" (after Conclusions):
"The Python scripts and Jupyter notebooks used for GNSS-derived VTEC processing, quality control, quiet-day baseline removal, statistical analysis, and figure generation are available in a public GitHub repository: https://github.com/NouhaBouha/gnss-tec-march-2015-storm. The repository includes the processing workflow, station-level input files used for the corrected analysis, geomagnetic-index files, final tables, figures, and documentation required to reproduce the results. The original GNSS observation and navigation files are publicly available from the IGS/UNAVCO archive, CODE GIM and DCB products are available from CODE, the Kp index from GFZ, and the Dst index from the World Data Center for Geomagnetism, Kyoto." The repository includes a README, a step-by-step workflow document, and station-specific processing notes documenting quality-control choices in detail (including the corrected DCB handling referred to in our General Comment above).
RC1 Comment 6 — Conclusions and limitations (future extensions)
"The present study focuses on the ionospheric response to a single geomagnetic storm event... Could the authors discuss how the proposed analytical framework might be extended in future investigations. For example, by examining additional geomagnetic storms, including a denser network of GNSS stations, or integrating physics-based ionospheric models?"
AC6: We have expanded both the Limitations (Sect. 4.5) and Conclusions (Sect. 5) to explicitly address all three extensions suggested.
Manuscript location: Sect. 4.5 (Limitations of the Corrected Filtered Analysis):
"Several limitations should be noted. First, the latitudinal gradient is derived from only four stations; although the stations form a coherent transect from mid- to auroral latitudes, denser station coverage would be required for a full regional mapping.”
Manuscript location: Sect. 5 (Conclusions), final paragraph:
"Future work should extend this approach to multiple geomagnetic storms, additional longitude sectors, and coupled physics-based modelling in order to assess the generality of the observed relationships and improve predictive capability.”
Together with the solar-wind/IMF discussion added in response to Comment 4, these additions address all three specific directions suggested by the Referee: additional storms, denser station networks, and physics-based models.
We thank the Referees again for these comments, which we believe have substantially improved the clarity, reproducibility, and scientific rigor of our manuscript.
Citation: https://doi.org/10.5194/egusphere-2026-2994-AC1
-
AC1: 'Response to Referee #1 (RC1)', Nouhaila Bouhadi, 14 Sep 2026
-
RC2: 'Comment on egusphere-2026-2994', Anonymous Referee #2, 24 Aug 2026
The current paper analyses a storm of 17 March 2015 using GNSS data over four GPS locations in the northern hemispheric part of the European-African sector. It is important to mention that this particular storm has been comprehensively studied including over the region covered by the authors. As a result, the paper in its current form does not show additional understanding of this storm effect on the ionospheric dynamics and or electrodynamics. It is not clear why the authors use a limited dataset and yet Europe has sufficient GNSS data that would have been used to show detailed changes. In my view, the current paper should be reconsidered and the authors should identify the research knowledge gap and demonstrate how such is addressed. Below are comments that may be helpful
- There is sufficient data and it is not clear why the authors use a small dataset to perform a complex study that modified ionospheric changes during a strong storm under study. The authors should contextualize their study since this storm has been extensively studied.
- In the abstract, the percentage increases should be mentioned in the context of what is used as the quiet time variability. I can't understand what the correlation analysis of TEC with TEC aims to achieve here. Why correlate TEC with latitude?
- Line 45: This storm has been comprehensively investigated including over the African-European sector. While additional studies are welcome, I haven't seen any new additional findings over this sector during this storm period.
- The comparison with CODE TEC is not necessary. This is even low resolution compared to what would be derived from GNSS observations. Unless you are developing a new TEC calibration methodology and would like to evaluate its performance.
- Line 60: This looks to me like a research project-oriented report which the authors tried to convert into a scientific paper. These objectives don’t all meet the scientific peer-reviewing criteria unless contextualized.
- The text in line 80 is difficult to understand. How do stations ensure this? There are many stations in Europe which should have been used to do this.
- In relation to Figure 1 and associated text: What authors call time lag could be introduced by local time differences. TRO is almost 20 degrees East while RABT is beyond 5 degrees West. There is a local time difference exceeding one hour
- Two sentences before section 3: Does this fit here? What is the relevance?
- Subsection 3.1: Did the authors develop their TEC derivation software? If so, state so and if you used existing software, please provide a reference. The details in 3.1 don't need to be included in the scientific paper. Otherwise, if you would like to describe the TEC derivation software, I would suggest that you consider drafting a different paper describing the technique.
- Regarding the elevation angle of 10 degrees: Other previous studies have used elevation angles of about 20-30 degrees. Even 15 degrees is considered to be lower as you will be sampling a huge portions of the ionosphere if you use lower elevation angles.
- Subsections 3.1-3.3: The text under this is basic and shouldn’t appear in a scientific paper too; particularly given that most of this has been discussed in other sources. However, the quoted number of satellites look small at these latitudes. Please crosscheck. Why resample and not just take averaged TEC values at 1 hour resolution?
- Equation 7 in relation to the background or quiet time ionospheric variability: This method should be reconsidered. For analysis, you performed re-sampling. For background, you simply take median TEC for each hour? There are widely accepted methodologies for doing this such as monthly medians or taking mean of the most quiet days in a month.
- Line 155: What is the use of background if you post-processed using running mean again?
- Equation 10: This is a bit misleading because TEC depends on several factors and not only latitude. You can't therefore express it as a linear function, unless you are doing long term trends and have already performed averaging over month or even year. Even if this is done, trends are studied by removing solar activity changes.
- Section 3.6: This step is really not necessary as both TEC datasets are based on the same GNSS observations. Comparison with GIM TEC is necessary if one is developing a new TEC derivation technique, not studying ionospheric dynamics and there is already TEC.
- Figure 2: If you are performing comparison across different locations, I suggest that the y-axis (limit is consistent)
- Figure 4: Data plotted in this figure has potential issues. Kp should be plotted in its form which will show the three hourly interval. VTEC has small kinks almost for all stations, but are more pronounced for TRO. The analysis should be repeated to cross-check data quality. There may have been errors with data processing which makes the rest of the paper difficult to evaluate.
In summary, there is sufficient dataset and the resolution used of an hour should be improved. Some physical mechanisms have time scales of an hour or less and this study does not take into account such dynamics.
Citation: https://doi.org/10.5194/egusphere-2026-2994-RC2 -
AC2: 'Response to Referee #2 (RC2)', Nouhaila Bouhadi, 14 Sep 2026
We thank the reviewers for their thorough and constructive reading of our manuscript. Below, we address each comment in turn, in the raised order. Reviewer comments are reproduced (paraphrased where noted) in the left column; our response and the corresponding change in the revised manuscript are given in the right column. Line numbers refer to the revised manuscript.
General comments
Reviewer’s comment
Author response
There is sufficient GNSS data available over Europe; it is unclear why only four stations were used for a storm that has already been extensively studied. The study should be contextualised and a clear knowledge gap identified.
We agree that the choice of four stations needed explicit justification. We reframed the study around a bounded research question rather than an implicit claim of regional completeness.
Added: "the study addresses the following question: to what extent can a small but consistently processed GNSS station transect resolve the sectorial structure of the ionospheric response..." (Introduction, lines 70–75).
Added justification after Table 1: the four stations were "a deliberate choice to construct a simple and traceable latitudinal transect... rather than to map the full European GNSS network" (lines 95–100).
Added a leave-one-out regression sensitivity test (§4.3, line 245) confirming the latitudinal gradient is not driven by a single station.
Novelty statement added: the aim is not to re-characterize the storm but to test how a uniformly processed station transect captures the sectorial VTEC structure (lines 55–60).
In the abstract, percentage increases should be stated relative to the quiet-time baseline used. It is unclear what the TEC–latitude correlation analysis is meant to achieve.
The abstract was rewritten to remove bare percentage figures and instead lead with absolute VTEC values, the fitted gradient, and validation statistics, all explicitly tied to the quiet-day baseline described in the Methods.
The correlation analysis is now framed only as a supporting diagnostic of Kp/Dst coupling, not as a latitude correlation (Introduction, lines 70–75).
Line 45: The storm has already been comprehensively studied over the African–European sector; no clearly new findings are presented.
An explicit novelty statement was added: the goal is an internally consistent, uniformly processed quantification of the sectorial gradient and station-dependent perturbation patterns, not the identification of new storm drivers (lines 55–60, 65–75).
The comparison with CODE TEC (GIM) is not necessary unless a new TEC calibration method is being evaluated.
We added an explicit statement of purpose for this comparison: "This comparison is not intended to validate a new TEC retrieval algorithm... It is used only as an independent consistency check to verify that the corrected station-level VTEC variations remain physically plausible" (§3.6, lines 190–195).
Line 60: The objectives read like a research-project report rather than a peer-reviewed paper unless properly contextualised.
A new paragraph frames the objectives around a single bounded research question and explicitly limits scope ("rather than aiming to provide a complete regional mapping..."), lines 70–75.
Line 80 is unclear: "stations ensure coverage across distinct ionospheric regimes" — how do the stations ensure this? Many more European stations exist.
Rewritten for clarity: "These stations sample distinct latitude-dependent ionospheric regimes, from mid-latitude conditions at RABT and MADR to higher-latitude and auroral-zone conditions at TRO1" (§2.1, lines 90–95). This states the sampling logic directly rather than implying exhaustive coverage.
Figure 1 / associated text: the apparent time lag between stations could be due to local time differences rather than a physical effect (TRO1 ~20°E vs RABT ~5°W, >1 h local-time offset).
Addressed explicitly in the revised Limitations (§4.5): "the longitude separation between RABT and TRO1 corresponds to a solar local-time difference of more than one hour, so the reported behaviour should be interpreted as a sectorial latitude–longitude response rather than a purely latitudinal effect" (lines 300–305).
Two sentences before Section 3: does this fit here? What is its relevance?
The tangential sentences on K-index-based prediction studies (Nebdi 2010; Chafik et al. 2024a,b) were condensed from several sentences to a single, clearly-scoped reference sentence at the end of §2.2 (line 105).
Section-specific comments
Reviewer comment
Author response
Subsection 3.1: Did the authors develop their own TEC-derivation software? If existing software was used, cite it; if the intent is to describe new software, this belongs in a separate paper.
A "Code and data availability" section was added (Section 6, lines 335–340) with a public GitHub repository containing the full processing workflow, so the origin and reproducibility of the processing chain is now transparent. The pseudorange-to-VTEC formulation itself follows standard, cited methods (Schaer, 1999; Hofmann-Wellenhof et al., 1992).
The elevation cutoff of 10° is low compared with the 20–30° typically used in the literature; even 15° samples a very large portion of the ionosphere.
The elevation cutoff was raised to E ≥ 30° throughout the processing chain (§3.2, lines 130–135), removing the earlier inconsistent reference to a 10° threshold. This is now stated consistently in the Abstract, §3.1, and §3.2.
Subsections 3.1–3.3 contain basic material that need not appear in a research paper. The reported satellite counts look low for these latitudes — please cross-check. Why resample instead of directly averaging TEC at 1-hour resolution?
Satellite counts (Table 1 / §3.3) were re-verified against the reprocessed dataset and are consistent with the stricter E ≥ 30° cutoff and per-epoch quality control now applied.
We added an explicit rationale for hourly median aggregation: "Hourly median aggregation was used instead of simple hourly means in order to reduce the influence of residual outliers, short data gaps, multipath-contaminated epochs, and uneven satellite sampling within each hour" (§3.3, lines 140–145).
Equation 7 (quiet-time baseline): using the median of a single quiet day is questionable. Established approaches use monthly medians or the mean of the quietest days in a month.
This is acknowledged directly in the revised Limitations: the event-centred single-day baseline "is less robust than climatological baselines based on monthly medians or averages over several geomagnetically quiet days," and reported ∆VTEC values are explicitly framed as "event-relative perturbations, not climatological storm deviations" (§4.5, lines 300–310).
We retained the single quiet-day baseline for this event-centred analysis (rather than switching methods) because it isolates the storm-specific perturbation most directly; the climatological alternative is flagged as a natural extension for future work.
Line 155: What is the purpose of the baseline if the data are then post-processed with a running mean?
The additional running-mean/smoothing step referred to by the reviewer was removed from the processing chain; ∆VTEC is now computed directly as VTEC(t) − Baseline(hour(t)) with no further smoothing (§3.4, Eq. 8, lines 155–165), so the role of the baseline is unambiguous.
Equation 10: expressing VTEC as a linear function of latitude alone is misleading, since TEC depends on multiple factors; this would only be appropriate for long-term, solar-activity-corrected trend analysis.
An explicit caveat was added immediately after Eq. 10: "This regression is not intended to imply that VTEC is controlled by latitude alone; rather, it provides a compact first-order description of the peak-VTEC decrease across the selected station transect during this specific event" (§3.5, lines 175–180), reiterated in §4.3 (lines 240–245).
Section 3.6: The GIM comparison is unnecessary since both datasets derive from the same underlying GNSS observations; it is only relevant when developing a new TEC-derivation technique.
As above, an explicit statement of intent was added: the GIM comparison is used only as an independent consistency check on the corrected station-level VTEC series, not as a validation of a new retrieval technique (§3.6, lines 192–195).
Figures
Reviewer comment
Author response
Figure 2: when comparing across stations, the y-axis limits should be consistent.
The station-comparison VTEC panels now use a common y-axis from 0 to 80 TECU to facilitate inter-station comparison.
Figure 4: Kp should be plotted as the step function it actually is (3-hour cadence). VTEC shows small kinks, most pronounced at TRO1 — the analysis should be repeated to check data quality, as processing errors would make the rest of the paper difficult to evaluate.
Kp is now plotted as bars at its native 3-hour cadence, as indicated in the revised Figure 4 caption.
The full dataset was reprocessed with stricter quality control: an epoch-level minimum satellite count, a hard E ≥ 30° elevation cutoff, and a minimum-valid-epoch requirement per hourly bin, with missing hours retained as gaps rather than interpolated (§3.3, lines 149–154). This reprocessing changed the reported headline values (e.g., RABT peak VTEC and the GIM validation correlations), and is documented in full in the new §4.5 "Limitations of the Corrected Filtered Analysis," including the exact retained/rejected epoch counts for TRO1 (lines 295–300).
Closing comment
Reviewer comment
Author response
There is sufficient data, and the 1-hour resolution used should be improved: some of the relevant physical mechanisms operate on timescales of an hour or less, which this study does not capture.
We agree this is a genuine limitation and now state it explicitly rather than leaving it implicit. A new closing paragraph in §4.5 (lines 310–315) reads: "the use of hourly median VTEC values limits the ability of the present analysis to resolve sub-hourly electrodynamic processes such as prompt-penetration electric fields or rapid auroral forcing. The hourly resolution provides a robust inter-station comparison and is consistent with the cadence of Dst, but it is not designed to capture short-timescale ionospheric dynamics... future work should use the original 30-second GNSS series together with higher-cadence solar-wind, IMF, and auroral indices to investigate faster responses."
We chose not to attempt a full sub-hourly reanalysis in this revision, since doing so reliably (with matched sub-hourly geomagnetic/solar-wind drivers) is a substantial undertaking in its own right; we present it here as a clearly scoped direction to follow-on work rather than overstating what the current hourly-resolution framework can support.
We believe these revisions directly address the reviewer’s concerns regarding dataset scope, methodological transparency, data quality, and the interpretive limits of the analysis, while preserving the paper’s core contribution: an internally consistent, multi-station quantification of storm-time VTEC variability across the European–African sector. We thank the reviewer again for the careful and detailed evaluations, which have substantially improved the manuscript.
Citation: https://doi.org/10.5194/egusphere-2026-2994-AC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 72 | 34 | 13 | 119 | 7 | 8 |
- HTML: 72
- PDF: 34
- XML: 13
- Total: 119
- BibTeX: 7
- EndNote: 8
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript presents a analysis of the latitude-dependent ionospheric VTEC response to the 17 March 2015 geomagnetic storm, using publicly available GNSS and geomagnetic datasets together. The results are generally consistent with previous studies of the St. Patrick’s Day storm and provide a quantitative assessment of the European–African sector. However, I recommend moderate revision to improve the clarity of the manuscript, and strengthen the discussion of novelty, data reproducibility, and the limitations of the analysis.
Novelty of the study: Could the authors clarify more explicitly what the principal novelty of the study is in comparison with previous investigations of this storm. Particularly regarding the integrated multi-station analysis of the European-African sector and the quantitative assessment of the latitudinal gradient of peak storm-time VTEC?
Data sources and Data processing: The study relies on publicly available GNSS observations. Could the authors provide additional details on the quality-control procedures applied to these datasets. How missing or low-quality GNSS observations were handled, how temporal consistency among the different datasets was ensured. Are there any limitations of using publicly available data may. Did this affect the robustness of the derived VTEC estimates and subsequent statistical analyses?
Latitudinal gradient robustness: The study reports a strong linear relationship between peak storm-time VTEC and geographic latitude based on four GNSS stations. Could the authors discuss this in details.
Choice of geomagnetic indices: The correlation analysis shows that Kp is more strongly associated with ΔVTEC than −Dst. Could the authors elaborate on the physical reasons for this difference and discuss whether other geomagnetic or solar-wind parameters, or interplanetary electric field components, might provide additional insight into the regional ionospheric response?
Code & Data availability: The authors states that the data processing and statistical analyses were performed using Python scripts that are available from the corresponding author upon reasonable request. To enhance transparency and reproducibility, could the authors provide the analysis scripts, processing workflow, and relevant documentation in a publicly accessible repository, e.g. GitHub or Zenodo?
Conclusions & limitations: The present study focuses on the ionospheric response to a single geomagnetic storm event in the European - African sector. Could the authors discuss how the proposed analytical framework might be extended in future investigations. For example, by examining additional geomagnetic storms, including a denser network of GNSS stations, or integrating physics-based ionospheric models?