Comparative Validation of Physics-Based Ground Motion Simulations in Northern and Southern California

Chukwuebuka C. Nweke, Sajan K C, Robert W. Graves, & Evan T. Hirakawa

Submitted August 30, 2026, SCEC Contribution #15629, 2026 SCEC Annual Meeting Poster #TBD

This study evaluates and compares physics-based ground-motion simulations for moderate-magnitude earthquakes in Northern and Southern California. The dataset includes 26 San Francisco Bay Area events and 30 Southern California events with magnitudes M4.0–5.5. Simulations are performed with a minimum shear-wave velocity of 200 m/s and evaluated up to 1 Hz using effective amplitude spectra (EAS). The Northern California simulations use SFCVM v26.4, whereas the Southern California simulations use CVMSi velocity model.

Simulation predictions and empirical ground-motion model predictions are independently compared with recorded ground motions. A Bayesian linear mixed-effects regression framework is used to quantify systematic bias and partition residual variability into event, site, and remaining within-event components. The resulting total residuals, event terms, site terms, and variability components are compared between the two regions to determine whether simulation performance differs systematically.

To investigate the causes of regional differences and station-specific biases, new site-proxy parameters are extracted directly from the three-dimensional velocity models used in the simulations. These proxies characterize features of the modeled subsurface structure, including shear-wave velocities over different depth intervals, depths to selected velocity horizons, vertical velocity gradients, impedance contrasts, and lateral variations in velocity structure. Their relationships with regression-derived site terms and simulation residuals are evaluated to determine whether they explain systematic amplification or de-amplification patterns not captured by conventional site parameters.

This comparison provides a consistent framework for assessing whether differences in simulation bias and variability between Northern and Southern California arise from regional wave-propagation effects, velocity-model characteristics, or modeled site response. The findings will support improved validation, interpretation, and development of physics-based ground-motion simulations for seismic hazard applications.

Key Words
ground motion simulation, validation, site response, velocity models, residual analysis

Citation
Nweke, C. C., K C, S., Graves, R. W., & Hirakawa, E. T. (2026, 08). Comparative Validation of Physics-Based Ground Motion Simulations in Northern and Southern California. Poster Presentation at 2026 SCEC Annual Meeting.

Related Projects & Working Groups
Ground Motions (GM)