Assessing the riverine flood forecast skill of GloFAS and Google Flood Hub with impact data and discharge observations to support early actions in Mali

This study evaluates opportunities to enhance the Mali Red Cross Early Action Protocol (EAP) for riverine flooding by comparing global flood forecasting systems against the country's current trigger model. Riverine floods are among Mali's most destructive natural hazards, yet the existing EAP trigger model relies solely on upstream water levels from the National Directorate of Hydraulics (DNH). Because it excludes meteorological forcing, this current setup restricts forecast lead times to a maximum of 4 days. By leveraging medium-range weather forecasts, physics-based models (GloFAS v3.0 and v4.0) and AI-based models (Google Flood Hub) were evaluated alongside the current model across various lead times and discharge thresholds to test their viability for anticipatory action.

Using discharge observations and district-level impact data sourced from multiple databases and text-mined news articles, the study demonstrates that GloFAS and Google Flood Hub possess sufficient skill for early action beyond 4 days in frequently flooded regions. These global systems offer larger spatial coverage than the current trigger model, indicating that early action plans in Mali could operate with a 7-day lead time and cover a broader area. Overall, the research assesses the usability of discharge versus impact data for forecast validation, evaluates the feasibility of expanding the current model's lead time and spatial extent, and highlights the potential and challenges of user-oriented forecast skill in flood-prone, data-scarce regions.