UAV-Borne, LiDAR-Based Elevation Modelling: An Effective Tool for Improved Local Scale Urban Flood Risk Assessment

Abstract In this study we present the first findings of the potential utility of miniaturized Light and Detection Ranging (LiDAR) scanners mounted on Unmanned Aerial Vehicles (UAVs) for improving urban flood assessment at the local scale. This is done by generating high spatial resolution Digital Terrain Models (DTM) featuring buildings and urban microtopographic structures that can affect floodwater pathways (DTMbs). The accuracy and level of detail of the flooded areas simulated by a hydrologic screening model (Arc-Malstrøm), were vastly improved when DTMbs of 0.3 m resolution representing three urban sites surveyed by a UAV-LiDAR in Accra, Ghana, supplemented a commercially available 10 m resolution DTM covering the full catchment area of the region. The generation of DTMbs necessitated the effective classification of UAV-LiDAR point clouds using a morphological and a triangulated irregular network method for hilly and flat landscapes, respectively. The UAV-LiDAR enabled the identification of archways, boundary walls and bridges that were critical when predicting precise runoff courses that could not be projected using the DTM only. Variations in a stream’s geometry due to a one-year time gap between the satellite-based and UAV-LiDAR datasets were also observed. The application of the coarser DTM produced an overestimation of water flows equal to 15% for sloping terrain and up to 62.5% for flat areas when compared to the respective runoff simulated using the DTMbs. The application of UAV-LiDAR may enhance the effectiveness of urban planning by projecting precisely the location, extent and runoff of flooded areas in dynamic urban settings.