Performance of the flood models in different topographies.

• A guideline for a proper model, inputs, and resolutions based on physical features. • This work evaluates the performance of flood models in different topographies. • Sensitivity of flood models to DEMs, modeling techniques and roughness is examined. • Main factors to get the best results with minimum computational costs are determined. • Best setting is selected for floodplain zones simulations in different topographies. Climate change leads to more severe floods, which needs to be modeled properly. This study uses two hydraulic models including Hydrologic Engineering Center River Analysis System (HEC-RAS) and LISFLOOD-FP to simulate floods in two basins (Kashkan and Missouri) with different topographies (mountain and plain terrain). The impacts of different parameters like input resolutions, Digital Elevation Models (DEMs), roughness coefficient, and modeling techniques on flood simulation are examined. Results indicate that the performance of the models varies based on topography, input resolutions, and modeling approach. Indeed, the 2D modeling of both models in different topography types leads to better results compared to 1D/2D modeling. In Kashkan basin, HEC-RAS and LISFLOOD-FP models simulate floodplain flow with accuracies of 74% and 70%, respectively, and the accuracies of the models in Missouri basin are 80% and 84%, respectively. Although both models, particularly HEC-RAS with ALOS 30 m, can provide reasonable results, with an accuracy of 74% for Kashkan and 63% for Missouri basin, LISFLOOD-FP can perform better (about 4% higher accuracy) than HEC-RAS with a higher resolution of DEM (3DEP) in Missouri basin with significantly lower computational costs. In addition, the performance of the two models significantly decreases using ASTER DEM in plain topography. Results confirm that HEC-RAS outperforms LISFLOOD-FP in coarser resolution DEMs, while LISFLOOD-FP performs better in plain topography, which requires higher resolution DEMs. Despite the low sensitivity of the two models to roughness coefficients, the variability of the LISFLOOD-FP simulations is smaller than HEC-RAS. This work uses the best setting and factors to simulate floodplain zones, which is valuable for flood risk and hazard assessment to reduce and manage damages and losses. [ABSTRACT FROM AUTHOR]