Drivers of sub‐seasonal extreme rainfall and their representation in ECMWF forecasts during the Eastern African March‐to‐May seasons of 2018–2020

Abstract In recent years, Eastern Africa has been severely impacted by extreme climate events such as droughts and flooding. In a region where people's livelihoods are heavily dependent on climate conditions, extreme hydrometeorological events can exacerbate existing vulnerabilities. For example, suppressed rainfall during the March to May 2019 rainy season led to substantial food insecurity. In order to enhance preparedness against forecasted extreme events, it is critical to assess rainfall predictions and their known drivers in forecast models. In this study, we take a case study approach and evaluate drivers during March to May seasons of 2018, 2019 and 2020. We use observations, reanalysis and predictions from the European Centre for Medium‐Range Weather Forecasts (ECMWF) to identify and evaluate rainfall drivers. Extreme rainfall during March to May 2018 and 2020 was associated with an active Madden–Julian Oscillation (MJO) in Phases 1–4, or/and a tropical cyclone to the east of Madagascar. On the other hand, the dry 2019 March to May MAM season, which included a delayed rainfall onset, was associated with tropical cyclones to the west of Madagascar. In general, whilst ECMWF forecasts correctly capture temporal variations in anomalous rainfall, they generally underestimate rainfall intensities. Further analysis shows that underestimated rainfall is linked to a weak forecasted MJO and errors in the location and intensity of tropical cyclones. Taking a case study approach motivates further study to determine the best application of our understanding of rainfall drivers. Communicated effectively, knowledge of rainfall drivers and forecast uncertainty will inform preparedness actions and reduce climate‐driven social and economic consequences.