UAV-borne, LiDAR-based elevation modelling: a method for improving 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 modelling and assessments at the local scale. This is done by generating ultra-high spatial resolution digital terrain models (DTMs) featuring buildings and urban microtopographic structures that may 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, were used to supplement a 10 m resolution DTM covering the region’s entire catchment area. 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 data enabled the identification of archways, boundary walls and bridges that were critical when predicting precise run-off courses that could not be projected using the coarser DTM only. Variations in a stream’s geometry due to a one-year time gap between the satellite-based and UAV-LiDAR data sets were also observed. The application of the coarser DTM produced an overestimate of water flows equal to 15% for sloping terrain and up to 62.5% for flat areas when compared to the respective run-offs simulated from the DTMbs. The application of UAV-LiDAR may enhance the effectiveness of urban planning by projecting precisely the locations, extents and run-offs of flooded areas in dynamic urban settings.