A new sensitivity analysis method is proposed for assessing uncertainties in 2D hydraulic model. It is based on perturbation of the probability distribution of the input variables. The relative sensitivity indices are calculated for each variable using the Gauss quadrature sampling. The approach was tested and validated over a 45 km reach of the Richelieu River, Canada. The finite-element 2D hydraulic model, H2D2, was used to solve the shallow water equations (SWE). The impacts of three input variables on the expected water depths were considered in the analyses: flow rate, Manning’s n coefficient, and topography. Three Gauss sampling points were used to handle the random input variables: mean value and mean ± standard deviation. Five flood scenarios were simulated with discharge rates of 759, 824, 936, 1113, and 1282 . The results revealed that the water depths were the most sensitive to the irregular topography of the shoal, especially in the downstream direction. The Manning’s n coefficient and the flow rate had comparatively less impact on the model results. The results reveal that the sensitivity indices of the predicted water depths are highest for the topography, particularly downstream of the shoal. The Manning’s n coefficient and the flow rate have comparatively lower Sr values.
This was displayed as a poster presentation