Developing a Levee Module for Global Flood Modeling With a Reach‐Level Parameterization Approach

Levees are critical infrastructure for mitigating flood hazards worldwide. Despite their importance, current global flood models inadequately consider levees due to the complexity of flow mechanisms involving levees and lack of levee data. This research develops a levee module for global flood modeling and proposes a reach-level parameterization approach to estimate levee parameters globally. Specifically, we developed a simplified levee module within the CaMa-Flood flood model, requiring only two levee parameters: levee unprotected fraction and equivalent levee height. We then identified leveed river reaches globally and estimated the levee parameters for these reaches using open-access land use and levee standard data. Finally, we evaluated the model's performance by comparing changes in river hydrodynamics and flood hazard maps, with and without levees, against observed data or official flood maps from representative case studies. The results showed that (a) the proposed approach successfully identified protected reaches with a hit rate of 82.1% in data-rich regions, (b) the levee unprotected fraction can be accurately estimated based on land-use data, and equivalent levee height can be derived from flood defense data using the global flood model, and (c) the enhanced CaMa-Flood model, incorporating the levee module, accurately simulated both river hydrodynamics and flood hazard mapping, improving the mean Nash-Sutcliffe Efficiency of water levels from 0.68 to 0.84 and increasing the accuracy of flood hazard mapping from 0.76 to 0.87 in five US regions. Validation on flood mapping in other eight representative regions worldwide confirms robust performance and strong potential for global applications.