Non-Linear Soil Amplification Identified Empirically from Strong Earthquake Ground Motions

Abstract

Strong earthquake ground motions are influenced remarkably by local soil conditions. Though various factors of local soil conditions contribute to the effect, the effect owing mainly to the non-linear stress-strain relationship of surface soils is discussed in this paper with the aid of statistical analyses. An empirical regression model is presented first to obtain separately the effects due to seismic source, propagation path of waves and local site conditions from strong-motion spectra observed at the ground surface. The model is extended so as to derive soil amplification due to the material non-linearity of surface soils using the proportionality of soil strain to particle velocity and to lead to site-specific soil amplification dependent on the motion level which is called "non-linear soil amplification" in this paper. The extended regression model is applied to 228 strong-motion earthquake records observed at 26 sites in Japan. The non-linear soil amplifications identified by the regression analysis show quite site-dependent and period-dependent characteristics. Although being derived purely based on a statistical technique, the non-linear soil amplifications have a compatible relation with the physical response function resulting from the S-wave theory. The comparison between the non-linear soil amplification and their corresponding soil profiles reveals that the non-linearity of local site response is well correlated with the softness and formation of soils, particularly, with the N value distribution at each observation site.