STATISTICAL INFERENCES OF FLOOD PEAK DISCHARGE INDUCED BY LANDSLIDE DAM OUTBURSTS

Abstract

 In this study, data of the flood peak discharges induced by landslide dam outbursts were collected from the literature. Statistical inferences, including multiple regression analysis and Bayesian interference using Markov chain Monte Carlo (MCMC) method, were performed to obtain the statistical equations for the flood peak discharges induced by landslide dam outbursts. For Bayesian interference, prior probability and likelihood function were constructed and posterior probability was determined using MCMC method. Residual analysis was performed to detect outliers, and statistical inferences were carried out again to obtain the statistical equations after the abnormal values were deleted. The results of the statistical inferences were compared with each other to check their reliability. Moreover, a scheme to predict the flood peak discharge with a certain probability induced by a future landslide dam outburst was described, which makes it possible to more accurately determine mitigation measures such as evacuation areas with appropriate reliability in preparation for the landslide dam outburst.