統計的グリーン関数法による強震動予測法の検証と仮想福岡地震への適用

Abstract

In our previous work we extracted statistical characteristics of seismic ground motions from K-net records of 25 earthquakes observed at 142 stations in the Kyushu region in order to establish the synthesis method of statistical Green's functions. In this paper we study the strong motion prediction method using these statistical Green's functions. First, we apply this method to the Kagoshima-ken Hokuseibu earthquake of March 26, 1997 and compare the synthetics with the observed. Second, we simulate strong motions at the bedrock level during the Hyogo-ken Nanbu earthquake of 1995. Synthetic waveforms based on this method match quite well with the observed. Thus it is proved that we can predict the bedrock motions using our statistical Green's function if we properly express the source model. Finally we applied this method to predict strong motions in Fukuoka City for a hypothesized Fukuoka earthquake. First strong motions at the bedrock level are predicted and then the strong motions at the ground surface are obtained by the 1-D wave propagation theory. The peak ground velocity thus estimated reaches 100 cm/s, which is less than Kobe.