Estimation of site effects using strong ground motion network data

首先，建立了关西地区强震数据库。资料由Kyoshin-Net、CEORKA和神户时间强震观测组提供。此外，我还收集了遗址的地质和地球物理地下结构信息。利用这些数据库，我们试图用反演法分离场地和震源因子以及Q值(岩田和池仓，1988)。所得Q值与前人的研究结果相吻合。对于Q值的估计，我们可以仅使用岩石和硬土场地的数据集来获得稳定的Q值。S波垂直分量的Q值与水平S波的Q值一致，说明S波的垂直分量在地壳中以S波的形式传播。S波垂直分量的场地因子与P波垂向分量的场地因子相似，说明S波的垂直分量主要由P波构成。其次，对1995年日本南方大地震期间神户大地震灾区及周边场地因子进行了估算。即使在大灾区，不同场地的场地因素也不同，这表明仅有表层放大效应并不能解释神户的大灾害。利用K网数据，我们还利用小震数据作为经验格林函数，获得了1997年鹿儿岛-北部省地震的震源破裂过程。

First, I constructed strong motion data base in Kansai area. Data were provided by Kyoshin-net, CEORKA, and temporal strong motion observation group for Kobe. Also, I collected geological and geophysical underground structure information of the sites. Using these data base, we tried to separate site and source factors and Q-values by the inversion method (Iwata and Irikura, 1988). Obtained Q-values coincide with those by the previous studies. For estimating Q-values, we can obtain the stable Q-values with using data set of only rock and hard-soil sites. Q-values from vertical component of S-waves are consistent with those from horizontal S-wave component and this shows that vertical component of S-waves travel as S-waves in the crust. Site factors of vertical component of S-waves are similar to those of vertical component of P-wave part and this concludes that the vertical component of S-waves is mainly composed by P-waves.Secondly, we estimated site factors in and around large earthquake disaster area in Kobe during the 1995 Hyogoken Nanbu earthquake. Even in the large disaster area, site factors are different among the sites and this shows only superficial layer amplification effects can not explain the large disaster in Kobe.Using K-net data, we also obtained source rupture processes of the 1997 Kagoshimaken-Hokuseibu earthquakes using a small shock data as a empirical Green's function.