Uncertainty and sensitivity analysis of coastal flood damage estimates in the west of the Netherlands

Abstract. Uncertainty analyses of flood damage assessments generally require a large amount of model evaluations. This is often hampered by the high computational costs necessary to calculate flood extents and depths using 2-dimensional flow models. In this paper we developed a new approach to estimate flood inundation depths that can be incorporated in a Monte Carlo uncertainty analysis. This allows estimation of the uncertainty in flood damage estimates and the determination of which parameters contribute the most to this uncertainty. The approach is applied on three breach locations on the west coast of the Netherlands. In total, uncertainties in 12 input parameters were considered in this study, related to the storm surge, breach growth and the damage calculation. We show that the uncertainty in flood damage estimates is substantial, with the bounds of the 95% confidence range being more than four times smaller or larger than the median. The most influential parameter is uncertainty in depth-damage curves, but five other parameters also contribute substantially. The contribution of uncertainty in parameters related to the damage calculation is about equal to the contribution of parameters related to the volume of the inflowing water. Given the emphasis of most risk assessments on the estimation of the hazard, this implies that the damage calculation aspect deserves more attention in flood risk research efforts. Given the large uncertainties found in this study, it is recommended to always perform multiple calculations in flood simulations and damage assessments to capture the full range of model outcomes.