Uncertainty Analysis and Development of Probabilistic Inundation Mapping for Dam Break Induced Floods

A main function of a flood inundation mapping is to forecast flood hazard and assess potential loss. However, sources of uncertainties limited the reliability of inundation hazard assessments. It is identified that the major sources of uncertainties should be taken into consideration for supporting and guiding an optimal flood management strategy. In particular, this study focuses on a 20 km reach from the Shihmen Reservoir downstream. In this study, a dam break induced flood is utilized as the upstream boundary condition of flood routing. There are two major sources of uncertainties in the hydraulic model and the flood inundation mapping: uncertainties in the (1) dam break model and uncertainty of the (2) roughness coefficient. The pertubance moment method is applied to a dam break model and the hydro system model to develop the probabilistic flood inundation mapping. The different numbers of uncertain variables can be considered in these models and the variability of output uncertainty can be quantified. The probabilistic flood inundation mapping for dam break induced floods can be developed when the variability of output is considered. This can be achieved by a commonly used HEC-RAS model. The comparison between different probabilistic flood inundation mappings is discussed in this study. In general, a stochastic model could provide much more information for us to evaluate the likelihood of outputs. In other words, the probabilistic flood inundation mappings not only have the information of traditional flood inundation mapping but also provide an efficient way to demonstrate the likelihood of inundation for the flood-prone area. In conclusion, the decision-makers can weigh their decision when they have adequate information. It is hoped that the probabilistic flood inundation mappings can provide a new physical insight for us to evaluate the flooded areas