Unmanned Aerial Vehicle and Structure from Motion Approach for Flood Assessment in Coastal Channels

Leal-Alves, D.C.; Weschenfelder, J.; Almeida, J.C.D.; Albuquerque, M.G.; Espinoza, J.M., and Gonzaga, B.A., 2020. Unmanned Aerial Vehicle and Structure from Motion approach for flood assessment in coastal channels. In: Malvárez, G. and Navas, F. (eds.), Global Coastal Issues of 2020. Journal of Coastal Research, Special Issue No. 95, pp. 1162–1166. Coconut Creek (Florida), ISSN 0749-0208. High-resolution topographic data are the primary inputs for various scientific applications. For years, fine-scale digital elevations models have been restricted to LiDAR systems. Currently, the use of consumer-grade cameras coupled with Unmanned Aerial Vehicles (UAV) is a consolidated alternative and widely used in several research fields. The rapid diffusion of the topography UAV-based was possible by the combination of three factors: assimilation of photogrammetric principles; low-cost of planning and execution; computational efficiency of the Structure from Motion (SfM) algorithms. The systematization of these factors, combined with the ground control points GNSS-RTK, provides accurate results. Using the UAV-SfM approach, we performed low-altitude aerial surveys (50 meters) for two coastal channels on a low-lying sandy beach in southernmost Brazil. The main objectives were: (1) three-dimensional reconstruction of the two coastal channels (washouts); (2) generation of the Digital Terrain Models with the dense cloud classification; and (3) accuracy assessment of DTMs when comparing them with GNSS-RTK checkpoints. The two study areas comprise just over 14 hectares. After the mosaic alignment composed of 533 images, the GCPs were inserted (projection error less than half-pixel). The dense cloud alignment was classified, and IDW interpolated the ground features. The DTMs were confronted with 50 checkpoints, obtaining the RMSE elevation of 0.0475 for channel 1 and 0.05 for channel 2. With the morphometric goals achieved, flood simulations were performed using the bathtub model for three vertical levels of sea-level rise based on scenario RCP 8.5 of the Intergovernmental Panel on Climate Change. The results demonstrate the consolidation of the UAV-SfM approach, including when used to assess flooding in coastal areas. We emphasize that, even though it is robust and increasingly accessible, the use of UAV-SfM is limited to relatively small areas, meteorological conditions, and legal regulations.