Key Drivers of Flood Damage Differ Between Low and High Water Depths

ABSTRACT

Understanding which factors have the greatest influence on flood damage helps decision‐makers to target interventions more effectively. This study investigates whether the explanatory variables for residential flood damage differ between low‐ and high‐water depth conditions. We analyzed a transnational survey database ( n  = 1245) from the 2021 floods in Germany, Belgium and the Netherlands using machine learning algorithms and permutation feature importance to assess variable importance for building and content damage. The median water depth (1.15 m) was used as a threshold to define the low‐ and high‐water‐depth ranges. In the low‐water‐depth range, water depth is by far the most important factor in explaining unitary building and content damage. In the high‐water‐depth range, unitary building damage is primarily influenced by water depth, prior hazard knowledge, and human stability. For unitary content damage in the high‐water‐depth range, ownership status and footprint area are also key influencing factors alongside water depth. The results suggest that, while water depth is a good predictor of damage, additional variables become necessary to fully explain it under more extreme hazard conditions, underscoring the value of multivariable models. From a policymaker's perspective, the results emphasize the importance of raising awareness through target‐group‐specific emergency and risk communication (e.g., for tenants and homeowners).