Inundation Simulation Model with Local Grid Refinement

The meteorological and geographical conditions in Taiwan frequently cause inundation in the midstream and downstream floodplain when storm occurs. It usually takes losses of people's wealth and life. Owing to the large amount computation requirements on the two-dimensional inundation simulation in large areas with fine resolution in dense population areas, the down-scale inundation simulation using local refinement grid at the dense population areas could save the time of model calculation. The study aims to find a reasonable boundary treatment for the local regions where fine grid resolution is adopted. The boundary conditions on the interfaces between the coarse and the fine grids are calculated from the coarse grid simulation, and then redistributed to the fine grids based on the weighting factors. The weighting factors can be determined by different flux or velocity distribution methods, including the uniform flux distribution, the uniform velocity distribution, the Chezy's velocity distribution and the Manning's velocity distribution. We evaluated the performance of the four boundary treatments methods according to the accuracy, comparing to the benchmark modelling results using a global uniform fine grid, and the computation time.