A Rapid Assessment Method for Flood Risk Mapping Integrating Aerial Point Clouds and Deep Learning

In recent years, floods have brought renewed attention and requirement for real-time and city-scaled flood forecasting due to climate change and urbanization. In this study, a rapid assessment method for flood risk mapping is proposed by integrating aerial point clouds and deep learning technique that is capable of superior modeling efficiency and analysis accuracy for flood risk mapping. The method includes four application modules, i.e., data acquisition and preprocessing by oblique photography, large-scale point clouds segmentation by RandLA-Net, high-precision digital elevation model (DEM) reconstruction by modified hierarchical smoothing filtering algorithm, and hydrodynamics simulation based on hydrodynamics. To demonstrate the advantages of the proposed rapid assessment method more clearly, a case study is conducted in a local area of the South-to-North Water Transfer Project in China. The proposed method achieved 70.85% in mean intersection over union (mIoU) and 88.70% in overall accuracy (OAcc), outperforming the PointNet and PointNet++ networks. For the case point cloud containing nearly 50 million points, the computation time is less than 9 min, while the computation times for PointNet and PointNet++ are both more than 24 h. Then, high-precision DEM reconstruction by proposed hierarchical smoothing method with topographic feature embedding. These results demonstrate the efficiency and accuracy of the proposed method in processing large-scale 3D point clouds and rapid assessment of flood risk, providing a new perspective and effective solution for flood risk mapping in the field of spatial information science.