BAYESIAN ESTIMATION OF TSUNAMI FRAGILITY CURVE OF BUILDING DAMAGE USING ABIC BY A FEW DATA SAMPLES

Abstract

 Hierarchical Bayesian estimation of fragility curve of specific building damage is investigated by using the prior distribution of fragility curve of similar building damage and the likelihood inputting a few data samples and by minimizing the Akaike's Bayesian Information Criterion (ABIC). It is applied to building damage by the 2011 Tohoku Earthquake Tsunami. Bayesian estimation results of tsunami fragility curve of building damage in Minami Souma using the prior distribution of that in all inundation area are agree with estimation results by the minimum square method of corresponding observation data. Bayesian estimation results of tsunami fragility curve of RC damage with more than the third floor using the prior distribution of that of all RC damage are also agree with estimation results by the minimum square method of corresponding observation data. Bayesian estimation of tsunami fragility curve of building damage of thermal power plant using the prior distribution of that of similar building damage can be properly performed without effect of input order of used observation data on the likelihood.