Refined Three-Dimensional River Channel Reconstruction Method Based on Coarse DEMs for Flood Simulation

Flood forecasts commonly require reliable input data to accurately reflect the actual situation. Although widely used in the world, the coarse digital elevation models (DEMs) from remote sensing often provide poor representations of the real topography due to the effects of water and mountain shadows. Remote sensing methods cannot reliably capture riverbed elevations, and fine-scale DEMs are needed. Due to the high cost of labor and material resource limitations, complete fine-scale DEMs are difficult to obtain to support flood forecasting across long reaches at sufficiently high precision. This work presents a refined three-dimensional river channel reconstruction method by considering the longitudinal and lateral topographic features of rivers to provide realistic river terrain data. The performance of this method in flood simulation is confirmed by simulating extreme flood events in the lower-670-km reach of the Jinsha River at a 30-m resolution. The numerical simulations and field measurements are quantitatively compared in terms of flood peaks and flood propagation processes. Numerical experiments further confirm that uncertainties from terrain inputs are not amplified by the hydrodynamic model when producing the final flood forecasting outputs.