Empirical Ground Motion Prediction Model Considering Spatial Variation of Path and Source Effects

Abstract

In order to improve the accuracy of empirical ground motion prediction, we examined to make a ground motion prediction model considering spatial variation of path and source effects. A modeling using mixed geographically weighted regression was proposed to model spatial variations of that effects, and the effectiveness of the modeling was examined using data from crustal earthquakes around Fukushima Prefecture. It was confirmed that the proposed modeling could take into account the spatial variations according to the data, and the prediction model with smaller residuals than conventional modeling could be obtained. The proposed modeling is effective to make advanced ground motion prediction model by utilizing increasing data.