Evaluation of 3D Seismic Velocity Models in the San Francisco Bay Area using Observed Ground Motions

Tara A. Nye, Grace A. Parker, Evan T. Hirakawa, Annemarie S. Baltay, Kyle B. Withers, & Morgan P. Moschetti

Submitted September 7, 2025, SCEC Contribution #14569, 2025 SCEC Annual Meeting Poster #TBD

We compare structural differences between two 3D velocity models and evaluate how well they model observed ground motions in the San Francisco Bay Area (SFBA). Seismic velocity models are a necessary input for simulating ground motions, affecting both the temporal evolution and amplitude of shaking. The uncertainty in the prescribed velocity model contributes to the reliability of simulated ground motions and confidence in seismic hazard assessments that depend on simulations. In this study, we compare the tomographic model of Guo et al. (2025) with the geology-based USGS velocity model (Aagaard and Hirakawa, 2021) to assess how well they capture travel time and path attenuation of observed ground motions. The tomography model has a coarse resolution with smoothly varying velocity gradients. Conversely, the geology model has a fine resolution with sharp velocity contrasts across geologic blocks and nearly uniform velocity within blocks. The observational dataset used in this study consists of ground motions from 1126 M3–6 earthquakes recorded at 822 stations across the SFBA. For each of these records, we compute the theoretical raypath of the fastest arriving ray through both models using the raytracing software Pyekfmm. We evaluate the time domain accuracy of the models by comparing raypath computed travel times with catalog phase picks. To assess the path attenuation, we examine correlations between the integrated change in velocity along each raypath (dVI) and the observed ground-motion path effects computed using mixed-effects regression of ground-motion model residuals. Differences in model resolution and velocity gradient lead to variable 3D paths, travel times, and dVI. Neither model comprehensively outperforms the other at capturing features of the observed ground motions. This comparison allows us to develop best practices for merging multiple velocity models to improve future ground-motion simulations.

Key Words
velocity model, validation, ground motions

Citation
Nye, T. A., Parker, G. A., Hirakawa, E. T., Baltay, A. S., Withers, K. B., & Moschetti, M. P. (2025, 09). Evaluation of 3D Seismic Velocity Models in the San Francisco Bay Area using Observed Ground Motions. Poster Presentation at 2025 SCEC Annual Meeting.

Related Projects & Working Groups
Community Earth Models (CEM)