[S-CG49] Integrative seismic and secondary hazard/risk assessment

[SCG49-P01] Regional site-amplification effects in the San Francisco Bay region, California, USA from ground-motion analysis and modeling of regional seismic velocity structure

*Morgan Moschetti¹, Brad Aagaard¹, Oliver Boyd¹ (1.US Geological Survey)

Keywords:earthquake ground motions, site response

We explore the relationship between regional site amplification and seismic velocity profiles in the San Francisco Bay region of California through ground-motion analysis and modeling of one-dimensional profiles from three-dimensional seismic velocity models. We compute Fourier amplitude ground motions using a dataset of 34,000 records from 200 M>3.3 earthquakes. The site terms are computed from mixed-effects regressions of the ground-motion residuals, relative to the Bayless and Abrahamson (2019; BA19) ground-motion model (GMM), and partitioned into bias, event, and station-specific components. We first evaluate the effect on data misfit from the use of site parameters—time-averaged shear-wave speed to 30 m (V_S30) and depth to the 1-km/s shear-wave horizon (Z₁). The impact of the V_S30 and Z₁ parameters on fits from the BA19 GMM is assessed by comparing sets of site terms that result from the use of uniform reference V_S30, V_S30-only, and from V_S30-Z₁-pairs at each site when evaluating the GMM. We identify regions of persistent long-period (T>1 s) misfit, which indicate site response that differs from that of the GMM. For all sites, we extract profiles from the U.S. Geological Survey San Francisco Bay region seismic velocity model (Aagaard and Hirakawa, 2021) and the National Crustal Model (Boyd, 2020). Using Thompson-Haskell propagator matrix methods, we compute amplifications from the profiles and compare the period-dependent trends in amplification with the trends in the ground-motion residuals. With a primary focus on the sedimentary basins of the region, we explore the effect of seismic velocity structure on the ground-motion misfits. Understanding the causes of varying site amplifications will be an important step towards integrating nonergodic GMMs in regional seismic hazards assessments and may lead to improved models to predict site response that extend beyond the San Francisco Bay region.