Automatic generation system of subsurface structure models by micro-tremor observation

为了预防或减轻未来的地震灾害，精确的地下结构模型是改进地震危险性分析和地震动估估的最有效手段。微震观测方便、廉价，可以用来建立地下结构模型，但微震是由各种波组成的，需要补充其他信息来构建合适的地下结构模型。首先，我们配备了7台便携式地震仪，在东京的东川和横滨的Kohoku，以及在新川地震中观测到的强烈地面运动的Ojiya和Tokamachi，进行了FK分析的阵列观测。与这些观测并行的是，我们用邻域算法开发了多目标的反演算法，并编制了一个计算机程序，可以快速稳定地计算高至瑞雷波正常模式的面波速度。我们在这四个地点进行了联合反演，并证明了将H/V谱中保留的深层界面信息与频散曲线中反映的精确浅层信息结合在一起，以改进深达数十米的地下模型，具有城市现实的阵列半径小于100米的效果。

In order to prevent or mitigate future earthquake disasters, precise subsurface structure models are most effective to improve earthquake risk analysis and earthquake ground motion estimates. Micro-tremor observation, which is handy and inexpensive, can be used to construct subsurface structure models ; however, micro-tremor is composed of various waves and should be supplemented by other information to construct a proper subsurface model. In this study, systematic search algorithm for an optimal subsurface model is sought, which utilizes an inverse analysis for H/V spectra and dispersion curves obtained by micro-tremor array observation.First, we were equipped with seven portable seismographs to carry out array observation with FK analysis at Higashi-shinagawa in Tokyo and Kohoku in Yokohama, and also at Ojiya and Tokamachi where strong ground motion was observed during Niigataken Chuetsu earthquake. Parallel to these observations, we have developed inverse algorithm for multiple targets with the Neighbourhood algorithm, as well as a computer program for calculating surface wave velocities fast and stable with generalized TR matrix method up to higher Rayleigh wave normal modes.We have,done joint inversion at those four sites and have shown the effectiveness of incorporating information of deeper layer interfaces retained in H/V spectra to that of precise shallower layering reflected in dispersion curves to improve subsurface models up to the depth of several tens of meters with urbanely realistic array radius of less than 100 meters.