High-precision reservoir DEM extraction using integrated sky-space-land-water observation

In order to overcome the limitations of traditional reservoir Digital Elevation Model (DEM) extraction methods in terms of terrain acquisition in no-fly zones and the fusion of land-water boundary areas, this study proposed a coordinated "sky-space-land-water" high-precision DEM extraction method that integrated satellite stereometric mapping, Unmanned Aerial Vehicle (UAV) Light Detection and Ranging (LiDAR), unmanned vessel single-beam bathymetry, and Real-Time Kinematic (RTK) shoreline measurement. By combining RTK scatter data to construct an error correction model, the accuracy of satellite stereometric mapping-derived DEM in no-fly zones was improved. An exponential decay weight fusion algorithm based on shoreline constraints was proposed to achieve seamless splicing of land-water DEMs and mitigate the elevation step effect at the junction. The integration of multi-platform data from satellites, UAVs, RTK, and unmanned vessels overcame the challenges of terrain acquisition in no-fly zones and dynamic land-water transition zones. Taking the Tingzikou Hydropower Project as an example, a globally high-resolution fused DEM was constructed, and the reservoir storage capacity was calculated and compared with 2005 storage capacity data. The results show that the proposed method can effectively enhance the data accuracy and continuity of the DEM in no-fly zones and land-water transition zones. The storage capacity calculation based on this DEM revealed that by 2024, the dead storage capacity of the reservoir had decreased by 8.47% compared to 2005, while the flood control storage capacity and active storage capacity had decreased by 0.19% and 0.23%, respectively. The research provide a reliable DEM data foundation and storage capacity analysis tool for refined reservoir management and sedimentation assessment.