Mapping the World's River Levees: A Hyper‐Resolution Levee Database Based on Digital Elevation Models

Abstract The flooding of rivers can be both beneficial and disastrous to society. As global climate change progresses, hazard mapping utilizing river models has become crucial for mitigating disastrous consequences. Accurate inundation predictions from such models rely on precise data on the locations of levees. However, current levee databases have limited coverage. In this study, we propose a globally applicable method for detecting levees using morphological parameters derived from high‐resolution LiDAR Digital Elevation Models (DEMs). Our method shows adequate detection performance. Notably, even unrecorded natural and artificial levees were identified. Further, we identified 49,000 km of levee across five countries, 27,000 km of which protect urban areas and cropland. A portion of the remainder consists of natural levees within natural landscapes. The detected levees may enhance the accuracy of flood modeling and risk assessment. Our method possesses significant application potential, which will increase further as the availability of DEMs expands globally.