MUST

MUST is a cloud-native flood risk decision-support toolkit for East Africa that transforms complex ECMWF ensemble forecasts into clear and actionable flood warnings for national meteorological services, disaster management agencies and humanitarian actors. The toolkit standardizes ~1 petabyte of forecast data into efficient Zarr/IceChunk formats and computes exceedance probabilities across rainfall accumulation windows from 3 hours to 7 days and return periods from 2 to 100 years, identifying when and where flood risk is highest at admin-1 level. The system further evaluates early warning performance through a Missed Opportunity Index that classifies each forecast as effective, late, or missed against observed flood events, giving a rigorous read on early warning performance. The outputs are then delivered through an interactive calendar and storymap interface, allowing decision-makers to explore daily risk signals, observed rainfall, and historical flood events in an intuitive way. By bridging the gap between forecast availability and real-world decision-making, the project enhances anticipatory action, reduces missed opportunities, and ultimately aims to save lives and protect livelihoods. Its full pipeline is open-source and cloud-deployed, supporting reproducibility and adaptation across the region.