Item

Abstract

Pseudo-dynamic source modeling is an efficient tool to simulate near-source ground motions and investigate their characteristics. It combines the advantages of both kinematic and dynamic source modeling. In other words, it keeps the computational efficiency of the kinematic source modeling, while the input values of the pseudo-dynamic source models are constrained by analyzing a set of dynamic source models. In this study, we investigated the non-stationary characteristics of the dynamic source models. Although the pre-existing pseudo-dynamic source modeling method assumes the stationarity of source parameters, it is reasonable to expect the non-stationary distributions of kinematic source parameters such as asperities. We carefully analyzed a set of dynamic source parameters, focusing on their non-stationary distributions. Then the non-stationary components (trend part) were extracted by using elliptical Gaussian surfaces. After extracting the trend part, we confirmed that the source statistics of the remaining distribution (fluctuation part) show a reasonable level of stationarity. We further plan to simulate near-source ground motions by using the separated source models into trend and fluctuation parts. Then we investigate the contribution of each source component to the frequency contents of near-source ground motions. We expect to improve the pseudo-dynamic source model significantly by considering the non-stationarity of source parameters and dealing with the trend and fluctuation parts separately.