Predicting East Africa’s rainfall extremes with calibrated, hybrid physical and AI systems

Accurate and reliable rainfall forecasts are crucial for Early Warning Systems (EWS). Hybrid Artificial Intelligence (AI) approaches, which integrate global physical forecasts with AI, enable low-cost ensemble generation, extending skill to medium-range lead times and offering scalable systems to inform EWS particularly under resource-constrained settings. However, assessing their practical value in representing user-relevant rainfall extremes and guiding decisions for EWS remains critical. We present a low-cost, fine-tuning method that calibrates ensemble forecasts to probabilities of exceedances for rainfall extremes. Applied to physical and hybrid AI rainfall forecasts over East Africa, case-study results show encouraging skill improvement from hybrid AI systems at higher rainfall thresholds. Sub-regional analysis shows the hybrid AI systems better reflecting the observed chances of extremes at longer lead times. Forecast skill decomposed across rainfall and probability thresholds further highlights the ranges of rainfall and risk tolerances where most actionable improvement arises, whether using satellite or rain-gauge data as reference. This tailored evaluation helps guide forecast adoption for EWS.