Statistical Analysis of Seismic Motion Based on Hierarchical Bayesian Models

Abstract eng:
An attenuation relationship, i.e., a ground motion prediction equation, is a statistical equation, conventionally used to predict ground motion intensity measure in probabilistic seismic hazard analysis. Ground motion prediction consists of analysis of the source, propagation path and site amplification characteristics. The site amplification characteristics are considered to be important factors affecting precision of ground motion prediction, therefore a site-specific ground motion prediction is desirable for probabilistic seismic hazard analysis. A conventional attenuation relationship, however, is not a site-specific but a generic equation to predict ground motion intensity that is because of the assumption of ergodicity, because the number of records that are observed at specific site is limited. A Bayesian approach is considered efficient for constructing a site-specific attenuation relationship. This study proposes a methodology to develop a site-specific attenuation relationship by employing a hierarchical Bayesian model. Then, a site-specific attenuation relation for JMA (Japan Meteorological Agency) seismic intensity scale is developed for crustal earthquakes observed in Japan from 1997 to 2011. The developed attenuation relationship by the hierarchical Bayesian method is considered less biased and explicitly accounts for all the prevailing uncertainties that is because of the less assumption of ergodicity. The attenuation relationship by the least square method was considered to bias the mean value of the predicted JMA seismic intensity up to about one, and to overestimate the standard deviation of aleatory uncertainty around four-thirds.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket