the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Physical Climate Drivers of East Africa’s March-April-May (MAM) seasonal rainfall Identified through Machine Learning Analysis
Abstract. East African March–May rainfall (MAM) remains difficult to predict despite its importance for agriculture, water resources, and disaster preparedness. This study identifies pre-season physical drivers of MAM rainfall and tests their value for probabilistic seasonal prediction. Predictor basins were derived from December and January sea surface temperature (SST), 2 m air temperature (T2), and sea-level pressure (SLP) anomalies relative to 1991–2020, using correlations with the leading mode of East African MAM rainfall and subsequent SHAP-based feature selection. The selected basin-derived indices were applied in Random Forest (RF) and Extreme Gradient Boosting (XGB) models. The dominant predictors appear to be the southern Indian Ocean T2 tendency, Australian and Eurasian T2 gradients, South Pacific and Antarctic T2 signals, Atlantic Niño tendency, and the Euro–African SLP gradient. T2-related predictors dominate both the January and December initialisations, showing that near-surface thermal gradients provide useful information in addition to SST memory. Walker-circulation diagnostics show that these drivers influence rainfall through pressure-gradient changes, tropical overturning, and upper-level wave-train development. For January initialisation, RF and XGB achieve spatially averaged Brier Skill Scores of 0.48 and 0.41, respectively, while the corresponding Area Under the Receiver Operating Characteristic Curve values amounting to 0.72 and 0.65. These results demonstrate that physically constrained machine learning provides promising probabilistic skill for East African MAM rainfall prediction.
- Preprint
(43033 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2832', Kalpesh Patil, 21 Aug 2026
-
AC2: 'Reply on RC1', Sinclair Chinyoka, 23 Oct 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2832/egusphere-2026-2832-AC2-supplement.pdf
-
AC2: 'Reply on RC1', Sinclair Chinyoka, 23 Oct 2026
-
RC2: 'Comment on egusphere-2026-2832', Anonymous Referee #2, 25 Sep 2026
Please see attached PDF for full comments.
-
AC1: 'Reply on RC2', Sinclair Chinyoka, 23 Oct 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2832/egusphere-2026-2832-AC1-supplement.pdf
-
AC3: 'Reply on RC2', Sinclair Chinyoka, 27 Oct 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-2832/egusphere-2026-2832-AC3-supplement.pdf
-
AC1: 'Reply on RC2', Sinclair Chinyoka, 23 Oct 2026
-
EC1: 'Comment on egusphere-2026-2832', Martin Jucker, 27 Oct 2026
Thanks to the reviewers for very thorough and helpful comments, I appreciate your time and effort.
Dear authors, your replies often reformulate the reviewer's comments instead of giving clear answers, and they are incomplete in some places. Some of the replies to the different reviewers are also inconsistent. For instance, in reply to RC1 you suggest changing the training to use 22 seasons as verification data in Comment 1, while in Comment 6 you state that this is already the case. It is not clear to me from the methods section in the current manuscript which is true. Another example is where you agree to no longer use EOF analysis in your reply to RC1, but in reply to RC2 you say you will "strengthen the EOF explanation". I agree with both reviewers that there are too many regions at the moment, and ocean versus land surfaces should be distinguished more carefully.
If you prepare a revised manuscript, please make sure these inconsistencies are resolved, and your revised manuscript reflects your response to the reviewers. I take the opportunity to also remind you of our AI policy which is available at https://www.weather-climate-dynamics.net/policies/ai_policy.html.
Citation: https://doi.org/10.5194/egusphere-2026-2832-EC1 -
AC4: 'Reply on EC1', Sinclair Chinyoka, 27 Oct 2026
Dear Editor,
Thank you for your reminder and for highlighting the inconsistencies in our responses to the reviewers. I would like to take this opportunity to clarify some of the issues raised.
Regarding the 22 seasons used for verification, we acknowledge that the current manuscript already indicates that forecast skill was evaluated using 22 seasons. However, our understanding was that the reviewers were requesting further clarification on how these seasons were selected and which specific seasons were included in the verification. Therefore, our proposed revision does not involve changing the verification period or the number of seasons used, but rather providing additional methodological details to improve transparency and reproducibility.
Regarding the use of Empirical Orthogonal Functions (EOFs), we acknowledge that our responses to the two reviewers may have created some confusion. This is because EOF/PCA was used for two distinct purposes in our study.
First, EOF analysis was used during the identification of potential predictor basins, where the first principal component (PC1) was used to represent rainfall variability. In response to the reviewers' comments, we propose replacing PC1 with grid-based standardised rainfall anomalies for this specific purpose.
Second, PCA and clustering were used to divide East Africa into three homogeneous rainfall regions, namely the Horn of Africa Zone (HoAZ), Equatorial Zone (EQZ), and Southern Sector Rainfall Zone (SSRZ). This regional classification methodology will remain unchanged. However, we will provide further details, including the explained variance and the procedures followed in identifying these regions.
We recognise that the distinction between these two applications was not sufficiently clear in our responses, and we will ensure that it is explicitly explained in the revised methodology.
Regarding the journal's AI policy, we appreciate your reminder. We have already included a statement in the Acknowledgements section of the manuscript describing how AI tools were used in preparing the paper.
Finally, I would like to sincerely apologise for some of the text that inadvertently appeared in our submitted responses to the reviewers. These passages originated from internal discussions with my co-authors during the preparation of our responses, and I regrettably failed to identify and remove them before submission. I take responsibility for this oversight and apologise for any confusion it may have caused.
Thank you once again for your careful review, constructive guidance, and patience. We appreciate the opportunity to clarify these matters and will ensure that the revised manuscript and accompanying responses are more consistent, precise, and transparent.
Citation: https://doi.org/10.5194/egusphere-2026-2832-AC4
-
AC4: 'Reply on EC1', Sinclair Chinyoka, 27 Oct 2026
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 350 | 274 | 68 | 692 | 50 | 49 |
- HTML: 350
- PDF: 274
- XML: 68
- Total: 692
- BibTeX: 50
- EndNote: 49
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Overall assessment
This manuscript addresses an important problem in East African seasonal rainfall prediction by combining physically based climate predictors with explainable machine-learning methods. The study is promising and potentially useful for seasonal forecasting, but several methodological and interpretive issues need clarification before the reported skill and physical conclusions can be considered robust.
Major contributions and achievements
The study provides a useful integration of climate dynamics, SHAP-based feature selection, and RF/XGB probabilistic forecasting. The identification of T2-based predictors in addition to SST signals is particularly interesting, while the circulation diagnostics provide a useful attempt to explain the physical pathways behind the statistical relationships. The reported BSS and AUC values also indicate potentially useful predictive skill.
Major comments
Predictor selection and information leakage: The manuscript should clearly explain whether correlation-based basin identification and SHAP feature selection were performed independently within each training fold. If the full dataset was used before validation, information leakage could inflate the reported forecast skill. A nested or strictly out-of-sample selection procedure is recommended.
Physical drivers versus statistical predictors: The terms “physical controls,” “drivers,” and “influence” appear stronger than supported by correlation and ML methods alone. The authors should distinguish predictive association from causality and moderate the wording unless stronger dynamical evidence is provided.
Justification of EOF1/PC1: The choice of rainfall PC1 as the predictand needs stronger justification. The authors should report its explained variance and spatial pattern and explain why it is preferable to a regional-mean rainfall index. A sensitivity test using regional mean MAM rainfall would strengthen the analysis.
Multiple testing: The use of p < 0.1 for global grid-point correlation screening may produce false-positive regions because many correlations are tested. The authors should address multiple comparisons and provide sensitivity tests using alternative significance thresholds or field-significance methods.
Additional value of T2 predictors: The conclusion that T2 provides information beyond SST should be demonstrated quantitatively. Comparisons of SST-only, T2-only, SLP-only, and combined models would show whether T2 predictors provide genuine incremental forecast skill.
Probabilistic skill assessment: The reported BSS and AUC values are promising, but the reference forecast, forecast categories, sample size, validation procedure, and uncertainty estimates should be clearly described. The meaning of “spatially averaged” BSS should also be clarified given that PC1 is described as the target.
Predictor robustness: The stability of the selected predictors across different training samples should be assessed. Reporting predictor-selection frequency or bootstrap stability would help establish whether the identified climate signals are robust rather than sample-dependent.
Physical mechanism: The proposed SST–T2–SLP–circulation pathway is plausible, but the current analysis appears mainly associative. Additional lagged relationships, circulation composites, moisture convergence, or moisture-budget diagnostics would strengthen the physical interpretation.
Operational relevance: The potential application to ICPAC and other climate services is valuable, but operational readiness should not be implied without independent hindcasts or real-time testing. The wording should be moderated to indicate potential operational usefulness.
Overall recommendation
The manuscript has a strong and relevant research focus, with promising contributions from physically informed ML and the identification of T2-related predictors. However, the reported skill and physical interpretations require stronger validation and clearer methodological documentation. I recommend major revision, particularly addressing information leakage, predictor robustness, incremental T2 skill, and the distinction between predictive association and causal physical drivers.