A Study Of Site Nolinear Model And Its Ground Motion Attenuation Relationship Based On V_S30 And Bedrock Depth

A ground motion prediction equation?GMPE?including source,path and site effect terms is an important tool and means for compiling seismic hazard zonation maps and aseismic engineering designs.Site effect has been considered to be one of the most important part in the GMPE and has been a research focus recently.On the one hand,in most modern GMPEs,V_S30?travel time averaged shear-wave velocity of the top 30m?or site classes based on site period TS?four times the shear-wave travel time from the bedrock to the ground surface?have been the conventional parameters to represent site effect.However,some studies suggest that V_S30 cannot represent the dynamic characteristics of deep soil site well.Though TS leads to better median amplification ratio and has been suggested as a better site parameter than V_S30 by some studies,obtaining the soil shear-wave velocity and the engineering bedrock depth for a deep soil site may result in a high cost in engineering applications.On the other hand,it is difficult to derive nonlinear site terms directly from the strong-motion records,because few records contain the effect of strong nonlinear soil response.Considering these two aspects,this study proposes to use TH-4Hrock/V_S30?where Hrock is the depth of the engineering bedrock?to replace TS and V_S30 as the parameter for the site effect in a GMPE when Hrock>30m and TH=120/V_S30 when Hrock? 30m.One dimensional?1D?models were constructed based on the shear-wave velocity profiles of 293 sites from the KiK-net strong-motion stations.529 rock-site horizontal components of strong-motion records from a world-wide dataset were selected as the bedrock input motions.A random effects regression model was fitted to the spectral amplification ratios according to the following steps:?1?VS?760m/s was selected as the shear wave velocity of the bedrock and the bedrock depth of 293 sites from KiK-net stations was determined.For each soil layer,a nonlinear model was selected based on the soil depth and its shear wave velocity randomly but within a set of given parameter ranges.An equivalent linear analysis was carried out to calculate amplification ratios of the ground surface motion and the outcrop rock motion by using the SHAKE program;?2?An empirical model was fitted to the calculated response spectral ratios.Then,residual analyses were conducted to test the model parameters.The distributions of residuals were used to test if a model parameter is required and statistical tests were used to determine if a parameter is statistically significant.To repeat the residual analysis until all statistically significant terms were derived.Finally,the 1D nonlinear model is defined a function of the bedrock response spectrum,TH,magnitude,source distance and impedance ratios;?3?In this paper,a random effects regression method was used to obtain the coefficients of the model parameters for the nonlinear site amplification ratios at each spectral period.The values of all parameters were smoothed across all spectral periods under a condition that the maximum log-likelihood does not decrease by more than 5%;In order to overcome the limitations of the 1D models,the nonlinear model derived in this study should be adjusted so that the amplification ratios from the model in this study equals the elastic amplification ratio in a GMPE when the rock site spectrum is small and the cross-over rock site spectrum for the final model is similar to those derived in this study when the rock site spectrum is large.