Updating proxy-based site amplification map with in-situ data in Osaka, Japan: A Bayesian scheme based on uncertainty projected mapping

Abstract

Site amplification maps are mostly proxy-based. Often due to the absence of in-situ data at the regional or local scale, a high level of confidence cannot be assigned to the site amplifications. It has often been observed that the in-situ amplifications differ from proxy-based estimates. So, whenever new in-situ data are made available, it is necessary to update the proxy-based estimates. Bayesian frameworks are recently gaining attention as model updating schemes. This study proposes a Bayesian scheme for updating proxy-based maps with in-situ data. This scheme is based on uncertainty projected mapping (UPM), where the significance of local in-situ data variability determines the posterior estimates. The study area is in Osaka, Japan, where discrepancies in proxy-based estimates and observed ground motions were documented during the 2018 Northern Osaka earthquake. Dense borehole data from the Kansai Geo-informatics Network are available in this area. Peak ground velocity (PGV) site amplification evaluations from this dense borehole network are used as likelihoods to update the prior proxy-based Japan seismic hazard information system (J-SHIS) site amplification map. As a result, the posterior map shows updated site amplification estimates which better represent the in-situ data. The updated site amplification map is then used to investigate the role of site amplification in explaining the building damage during the 2018 Northern Osaka earthquake.