Applicability of the site fundamental frequency as a VS30 proxy for Central and Eastern North America

Abstract:

One of the most important issues in developing accurate ground-motion prediction equations (GMPEs) is the effective use of limited regional site information in developing a site effects model. In modern empirical GMPE models site effects are usually characterized by simplified parameters that describe the overall near-surface effects on input ground-motion shaking. The most common site effects parameter is the time-averaged shear-wave velocity in the upper 30 m (V_S30), which has been used in the Next Generation Attenuation-West (NGA-West) and NGA-East GMPEs, and is widely used in building code applications. For the NGA-East GMPE database, only 6% of the stations have measured V_S30 values, while the rest have proxy-based V_S30 values. Proxy-based V_S30 values are derived from a weighted average of different proxies’ estimates such as topographic slope and surface geology proxies. For the proxy-based approaches, the uncertainty in the estimation of Vs30 is significantly higher (~0.25, log10 units) than that for stations with measured V_S30(0.04, log10 units); this translates into error in site amplification and hence increased ground motion variability.

We introduce a new V_S30 proxy as a function of the site fundamental frequency (f_peak) using the NGA-East database, and show that f_peak is a particularly effective proxy for sites in central and eastern North America We first use horizontal to vertical spectra ratios (H/V) of 5%-damped pseudo spectral acceleration (PSA) to find the f_peak values for the recording stations. We develop an f_peak-based V_S30 proxy by correlating the measured V_S30 values with the corresponding f_peak value. The uncertainty of the V_S30 estimate using the f_peak-based model is much lower (0.14, log10 units) than that for the proxy-based methods used in the NGA-East database (0.25 log10 units). The results of this study can be used to recalculate the V_S30 values more accurately for stations with known f_peak values (23% of the stations), and potentially reduce the overall variability of the developed NGA-East GMPE models.