| Earthquake ground motions can cause severer damages to many types of buildings and structures.Site effects are one of the main factors for structural damages as site effect tends to amplify the ground motions in a wide period range around the site period.The ground motion at a particular site consists two horizontal components and one vertical component and the frequency contents of the vertical component can differ significantly from the horizontal ones.Earthquake structural damage can be caused by either or both the horizontal and the vertical components.A soil site with a limited soil depth and a large impedance ratio can usually amplify the ground motions in a moderately large period range around the site periods and the amplification ratios for the horizontal components may differ from those of the vertical component.The site effect is,therefore,a complicated engineering problem in the design process.Most recent GMPEs for the vertical component use S-wave site parameter,such as VS30 or site classes based on site periods for shear-waves.Some studies show that the peak site amplification ratios of the vertical component for a group of sites appear to be close to the average P-wave site period of these sites.This means that the site amplification ratios for the vertical component should be modeled by P-wave site period even though the response spectrum,the peak response of a single-degree-of-freedom oscillator for the vertical component appears to be in the S-wave window.This means that the site term based on the S-wave parameter to model the vertical component in the GMPEs assume that P-and S-wave are perfectly correlated.Using S/B ratios these effects are investigated.The strong-motion network KiK-net provides a large set strong-motion records at the surface and the borehole at the bedrock.The most valuable information from the KiK-net is the measured soil properties down to the bedrock,such as S-and P-wave velocities.In this study,7305 pairs of strong-motion records were used to construct an empirical model for the S/B ratios.The geometric mean values of the spectra from two horizontal components were used.The empirical models include a number of parameters based on previous studies,magnitude,fault depth,source distance,and P-and S-wave velocity parameters.The S-wave site parameters include VS30,site period,impedance ratios based on S-wave velocity and the P-wave site parameters include VP30,travel time averaged P-wave velocity of 30m near-surface layers,site period,impedance ratios based on P-wave velocity.The empirical models are derived by using a random effects model and all model parameters were tested statistically.The results show that,indeed,magnitude and fault depth and source distance significantly affect the site response spectral amplification ratios,consistent with the results from the other existing studies for both horizontal and vertical components.One unexpected result is that at very short periods,P-wave velocity parameter significantly affects the S/B ratios at short spectral periods up to 0.3s for both horizontal and vertical components and the S-wave parameters affect the site amplification ratios for spectral periods over 0.3s also for both horizontal and vertical components.A surprising result is that amplification ratios for both horizontal and vertical components increase within increasing the impedance ratios from the S-wave velocities at short periods up to about0.4s.The impedance ratio from the P-wave velocities affects the amplification ratios of the horizontal and vertical components in a period range of 0.05-0.5s but the coefficients for the P-wave impedance are negative at many periods,suggesting that the amplification ratios decrease with increasing impedance ratios.For a horizontal layered one dimensional(1D)site,the Fourier spectral ratios between the site surface and the bottom of the soil layers are functions of the material properties soil layers and impedance ratio does not affect the transfer functions.Response spectral amplification ratios differ from the Fourier spectral ratios but for the elastic response,they should be similar except for that response spectral amplification ratios from different input motions.The reason for the impedance ratio effect may come from 2 or 3D effects because there only a small number of KiK-net stations have 1-D characteristics in the transfer functions between the surface and borehole records.This aspect was investigated by 2-D basin modeling.4 models were established with different width/depth ratios and modeled as elastic.Vertical incident seismic waves were assumed and the horizontal and vertical motions uncoupled that can be modeled separately.Three strong-motion records were used as the input excitation.The modeling results show that the amplification ratios indeed depend on impedance ratios at some spectral periods for both P-and S-wave excitations.For the P-wave excitation,at the center of the 2-D basin with a width/depth ratio of 4.0,response spectral ratios increase with increasing impedance ratios in a reasonably wide spectral period range,at very short periods,the amplification ratios decrease with increasing impedance ratios.The frequency content of the strong-motion records appears to be a factor.At long periods,for example,4.0s,the amplification ratios increase with increasing impedance ratio for one record and the amplification ratios from the other two records are very close to 1.0.For the basin with a large width/depth,the effect of the impedance ratio at moderately large spectral periods is generally small compared with those basins that have small width/depth ratio.For the S-wave incidence,the effects of impedance ratio are close to those for the case of P-wave incidence.The amplification ratios at short periods generally decrease with increasing impedance ratios;at moderate spectral periods,the amplification ratios at the center of the basin generally increase with increasing impedance ratios.At long periods,the effects of impedance ratio become smaller than those from other spectral periods. |