A Hierarchical Bayesian Model-Based Uncertainty Analysis for Tsunami Heights along Shorelines in Korea

Abstract

Uncertainties in estimation of tsunami inundation risk are mainly caused by limited information of tsunami characteristics associated with locations and propagation paths. It is common to use probability distributions in risk assessment so that a selection of the probability distributions and good estimates of the parameters are especially important to reduce the uncertainties in the assessment. The uncertainty in existing studies, however, has not been properly addressed. In this study, a new probabilistic tsunami-inundation risk assessment approach is proposed to characterize the uncertainties. A main objective of this study is to combine different sources of the uncertainties related to the attributes of earthquake (i.e., location and magnitude) and the estimation of the parameters of the distribution (i.e., Gamma distribution) in a Hierarchical Bayesian Model (HBM) framework. This study estimated the tsunami inundation risk with a Bayesian credible interval by combining the data of three historical tsunamis and 11 virtual tsunamis. The results indicate that the HBM well represented the underlying distribution and the associated uncertainties. In addition, the results confirmed that the proposed model was more relevant in quantitatively combining and estimating the uncertainties. The inundation risk information incorporating uncertainty could be used to better understand and manage tsunami-related hazards.