HVSR fpeak-Based VS30 Estimation Model and Its Applicability to GMPE in Southwest China

Abstract We determined the first significant peak frequency (f0) and highest significant peak frequency (fpeak) from the horizontal-to-vertical spectral ratio curves for each strong motion station and proposed logarithmic bilinear models for VS30–f0 and VS30–fpeak. Compared to the standard deviation of VS30–f0 logarithmic bilinear, which is 1.3119 (0.1179 log10 units), the standard deviation of VS30–fpeak logarithmic bilinear model is 1.3020 (0.1146 log10 units), indicating lower uncertainty. Second, the ratio distribution, residuals, and their standard deviations for fpeak-based VS30 and topographic slope-based VS30 relative to the measured VS30 were compared and analyzed. Finally, using the VS30–fpeak logarithmic bilinear model and topographic slope, we estimated the site VS30-values for 228 stations with definite fpeak-values in the dataset of Zhang et al. (2022) (ZYLW22) ground-motion prediction equation (GMPE), which were then applied in the construction of GMPE, respectively. Compared with the site VS30 estimated using topographic slope, the ratio of fpeak-based VS30 to measured VS30 is centered around 1.016, with residuals predominantly distributed between −0.2 and 0.2. The residual standard deviation is 1.3071 (0.1163 log10 units), indicating smaller uncertainty. Moreover, using the VS30–fpeak logarithmic bilinear model to estimate site VS30 for the 228 stations with determined fpeak-values, the average standard deviations of the site-to-site, within-event, and total residuals in the GMPE for spectra acceleration at periods ranging from 0.2 to 3.0 s are reduced by 2.80%, 2.23%, and 1.57%, respectively, effectively reducing the variability of the GMPE in southwest China.