Uncertainties in the predicted number of life loss due to the dam breach floods

Abstract

Several empirical methods based on the relationship between historical dam failure events and the number of casualty are available for estimating the loss of human lives resulting from dam failure. However, none of the empirical methods can seamlessly predict the number of casualties caused by a dam failure, because the characteristics of dam breach flood, warning time, human response to the warning, understanding of flood severity, and other factors necessary for calculating the number of life loss are not understood clearly. Therefore, the number of life loss predicted by the empirical methods is subjective to uncertainty. Uncertainties in the predicted number of life loss due to dam breach flood are identified in this study by performing uncertainty analyses regarding the dam breach flood routing analysis results and life loss estimations. The uncertainty analyses are conducted using the Monte Carlo simulation based on the Latin Hypercube Sampling technique, and computational programs are developed to carry out the uncertainty analyses. Considerable degree of uncertainty in the predicted number of life loss as well as dam breach flood routing results is revealed through the uncertainty analyses and presented comparatively with the modified boxplots.