Blind identification of site effects and bedrock motion from surface response signals

Abstract

A method for blind identification of site effects from two nearby ground surface response signals is presented. The proposed approach makes use of ground surface accelerations from two nearby stations to back-calculate the transfer functions of both sites and their common bedrock motion. Seismic analysis of structures cannot be carried out accurately unless site effects are taken into account. Moreover, presently available empirical attenuation relationships for predicting ground surface motions are only useful if site effects are considered. While an extensive collection of analytical and numerical techniques is available to analyze ground responses induced by bedrock motions, their accuracy depends on a priori knowledge of site properties and the availability of bedrock motions. There are techniques based on direct/indirect measurements—such as spectral analysis of surface waves (SASW), and material testing of borehole samples—however, responses predicted by their output do not necessarily reflect site behavior during strong motions. As such, the estimation of site response from acceleration data recorded on the ground surface during real-life events is a key capability. In the method proposed herein, the site response is identified from recorded ground surface accelerations at two nearby stations through a blind identification technique, under the assumption that the unknown bedrock motion is identical for both stations and those two stations have different transfer functions. Most of the existing site response identification methods rely on a strategically chosen reference station, and the present approach obviates this limitation. We demonstrate the performance of this new approach using a synthetic, but adequately realistic, example.