Refining Public DEMs for Urban Waterlogging Simulation via Vector–Raster Integration

The Digital Elevation Model (DEM), a crucial data source for waterlogging simulations, significantly influences the accuracy of the results. In complex urban environments, low-resolution DEMs cannot accurately capture the depressional characteristics of city roads or water levels during river floods, leading to distorted urban flooding simulations. To this end, this study developed a novel technique to refine the public 30 m resolution DEM to 1 m resolution for the urban area. The method establishes a zero-flood-depth baseline by correcting the elevations of key elements to improve the accuracy of urban inundation simulations. This is achieved through a semi-automated vector–raster integration workflow, which includes (1) road elevation correction that classifies road vectors, samples elevation at end points, and applies linear interpolation to depict roads as depressions and (2) waterway elevation correction that raises riverbed levels to match adjacent banks, simulating a pre-flood critical state. Polk County in Florida, USA, and the Central Business District (CBD) in Beijing, China, were selected as the research areas. In Polk County, we directly verified its accuracy using the official 1m LiDAR DEM. The results show that the mean error (ME), the root mean square error (RMSE), and the Standard Deviation (SD) improved by approximately 9%, 20%, and 65%, respectively, compared with previous methods. In Beijing, we used a volume matching algorithm to simulate urban flood depths under different rainfall scenarios, indirectly validating the results by comparing the simulated inundation volumes with the theoretical rainfall amounts. The refinement of the DEM significantly improved the topological accuracy of the river channels and the reliability of flood depths, and we analyzed two types of water accumulation behavior patterns. Overall, this study innovatively integrates public raster and vector data, utilizing known attribute information to refine public datasets and construct a highly precise water accumulation model.