Estimation of local site effects by a generalized inversion scheme using observed records of ‘Small-Titan’

A simple-extended array of observation for strong motions, named ‘Small-Titan’, was installed in Sendai City, Japan in 1997. This array system consists of 20 observation sites, which have an average of about 4 km for the station-to-station distance. Each observation station was located at various kinds of soils to effectively obtain local site effects. Since its completion, the system has obtained about 160 motion records. As is well known, ground motions are influenced by the three main factors such as source, path and local site effects. An accurate prediction of ground motions should thus be performed based on the three factors. In this context, theoretical methods used commonly lack enough information related with source and path effects. A generalized inversion scheme to separate the effects of source, propagation path and local site from S-wave spectra was applied to the records of Small-Titan obtained in the Sendai region, Japan. The objective of this paper is to estimate local site effects in Sendai City and to obtain meaningful information of input motions available to earthquake-resistant design, aimed at setting in terms of local soil conditions. This paper also describes the applicability of the generalized inversion scheme to seismic zoning.

Site amplification factors estimated from the generalized inversion scheme vary significantly, reflecting the effects of soil characteristics at each site. Strong site-dependency was found in the site amplification factors. In particular, the local site amplification factors in the western side are smaller than those in the eastern side, across the Nagamachi–Rifu Fault that exists in the central part of the Sendai area. Results from the generalized inversion scheme are compared with those obtained by the direct spectral ratio scheme, the H/V method and the bootstrap inversion. There is a good agreement in the amplification factors between the generalized inversion scheme, the direct spectral ratio scheme and the bootstrap inversion scheme. The H/V method for estimating site amplification is found to extract the same dominant periods as the other methods. This paper especially emphasizes a good agreement in not only dominant period, but also amplification factor.

The generalized inversion scheme is a useful method in separating the effects of source, propagation path and local site and in revealing the site amplification. In addition, the generalized inversion scheme provides meaningful information of input motions available to earthquake-resistant design, aimed at setting in terms of local soil conditions. Moreover, this paper indicates that the generalized inversion scheme is applicable to the seismic zoning toward disaster mitigation.