Comparison of measurement- and proxy-based Vs30 values in California

This study was prompted by the recent availability of a significant amount of openly accessible measured V_S30 values and the desire to investigate the trend of using proxy-based models to predict V_S30 in the absence of measurements. Comparisons between measured and model-based values were performed. The measured data included 503 V_S30 values collected from various projects for 482 seismographic station sites in California. Six proxy-based models—employing geologic mapping, topographic slope, and terrain classification—were also considered. Included was a new terrain class model based on the Yong et al. (2012) approach but recalibrated with updated measured V_S30 values. Using the measured V_S30 data as the metric for performance, the predictive capabilities of the six models were determined to be statistically indistinguishable. This study also found three models that tend to underpredict V_S30 at lower velocities (NEHRP Site Classes D–E) and overpredict at higher velocities (Site Classes B–C).