Rainy Season Climatology and Trends in Southern Africa Using Multiple Rainfall Datasets

Rainfall variability critically affects rainfed agriculture and water resources across the Southern Africa subcontinent, where communities are highly vulnerable to shifts in rainy season dynamics. This study provides a comprehensive assessment of the rainy season’s modality, onset, cessation, duration, and trend over the Southern African subcontinent using multiple and recent gridded rainfall datasets together with an extensive collection of in-situ observations. The analysis uses the rainy season definition methodology based on Liebmann & Marengo (2001) that can be applied to a range of climates ranging from semi-arid to humid. The method is applied to daily satellite, satellite-gauge-calibrated, gauge-only, and reanalysis datasets and a large collection of daily station data from about 1980 to 2020.  The approach is complemented by Fast Fourier Transform (FFT) analysis to enhance the robustness of seasonal signal detection rainy season modality.

Our results reveal a clear north-to-south rainfall gradient, with wetter equatorial regions and drier southwestern areas. This gradient is less pronounced southeastwards. Rainfall modality varies spatially, with bimodal regimes dominating the equatorial zone linked to the north-south movement of the rain belt (aka. Intertropical Convergence Zone (ITCZ)), while interior and southeastern zones exhibit unimodal summer rainfall peaks. The southwestern tip of South Africa displays a distinctive winter rainfall peak, mostly driven by extratropical low pressure systems. Transitional zones with complex orography as well as coastal zones show larger dataset disagreement, bringing challenges in capturing rainfall seasonality.

Trend analysis over recent decades indicates a trend towards delayed onsets (~2-3 days/year), earlier cessations (-3 to <~-4 days/year) and shortened durations of the rainy season (-2 to ~-4 days/year ) in regions such as Angola, Namibia and western South Africa. Cessation trends show higher spatial variability than onset trends. These changes are more pronounced in gridded datasets but also appear in station records, reinforcing confidence in the observed tendencies. The findings align with future climate projections under high-emission scenarios, highlighting risks for water availability, agricultural planning, and food security. The study emphasizes the need for improved observational coverage and integration of onset and cessation monitoring into early warning and climate adaptation systems.