FLEMOflash: The probabilistic flash flood loss model for quantifying direct economic losses with uncertainty information

In light of the increasing losses from flash floods, exacerbated by climate change, there is a pressing need for robust flash flood loss models to support risk analyses and mitigation strategies. Existing residential sector loss models predominantly focus on fluvial flood processes; while the key drivers of flash flood losses remain poorly understood. Applying Machine Learning on empirical data reveals key drivers of flash flood losses such as flow velocity and emergency response. We introduce FLEMO_flash (Flood Loss Estimation MOdel for flash floods), a novel multivariate probabilistic model to estimate losses to residential buildings and contents from flash floods. Model based assessments reveal that households with clear knowledge of emergency action during high water levels can reduce building losses by up to 78% and contents losses by up to 31%. Thus, FLEMO_flash can provide differential loss estimates based on varying levels of risk preparedness.