Uncertainty Analysis of a Lumped Physically Based Numerical Model of Dam Breaching

Failures of dams and dikes often lead to devastating consequences in protected areas. Numerical models are crucial tools to assess flood risk and guide emergency plans, but numerous sources of uncertainty exist. Therefore, identifying input variables uncertainties that induce critical uncertainties in model outputs is essential. Using Monte Carlo procedure, we conducted a sensitivity analysis based on Sobol indices of total order on 28 reference configurations. In each case, it allowed ranking input variables as a function of the significance of their uncertainty on the output variability and, for the first time, the dependency between reference configurations and sensitivity analysis results was highlighted. Uncertainty propagation was performed based on our implementation of the lumped fully coupled hydro-morphodynamic model developed by Wu (2016). Also, two new methods were developed to generate samples of dependent input variables.