Three-Dimensional Inundation Mapping Using UAV Image Segmentation and Digital Surface Model

Flood occurrence is increasing due to the expansion of urbanization and extreme weather like hurricanes; hence, research on methods inundation monitoring mapping has increased reduce severe impacts flood disasters. This studies compares two for depth estimation using UAV images topographic data. The consist three main stages: (1) extracting flooded areas create 2D polygons deep learning; (2) reconstructing 3D water surface data; (3) deriving a map reconstructed pre-flood DEM. are different at (stage 2). first method uses structure from motion (SfM) creating point cloud area overlapping images, resulted stage 1 applied classification. While second reconstructs by intersecting DEM created LiDAR We evaluate proposed over Town Princeville during flooding event Hurricane Matthew. compared validated USGS gauge level data acquired event. RMSEs SfM integrated based learning were 0.34m 0.26m, respectively.