Stochastic coastal flood risk modelling for the east coast of Africa

Coastal flooding resulting from tropical cyclones can have large repercussions in many low-lying regions around the world. Accurate flood risk assessments are crucial for designing measures to reduce the societal impacts of coastal flooding. At continental to global scales, however, traditional flood risk assessments mostly use methods that do not capture the spatiotemporal dynamics of coastal flood risk patterns. In this study, we address these deficiencies by applying a novel modelling framework that dynamically simulates stochastic coastal flood risk for the east coast of Africa. Using 10,000 years of synthetic tropical cyclones and a cascade of hydrodynamic models to simulate storm tides and flooding, we calculate the damage of each individual tropical cyclone event and empirically derive the risk curve for each country. Results show that the largest aggregated annual losses in the region come from multiple events rather than from a single low-probability event. Results also reveal that events with the highest return periods in terms of storm surge residual levels and flood extents are not necessarily the most damaging events. Here, the 1 in 10,000-year damage event is associated with a 1 in 45-year event in terms of flood extent, showing that addressing exposure and vulnerability is essential in determining risk. Our modelling framework enables a high-resolution continental-scale risk analysis that takes the spatial dependencies of flood events into account.