Validation of two new CyberShake studies in California

Xiaofeng Meng, Robert W. Graves, & Scott Callaghan

Submitted September 7, 2025, SCEC Contribution #14450, 2025 SCEC Annual Meeting Poster #TBD

The SCEC CyberShake platform is a 3D physics-based ground motion simulation platform developed for seismic hazard analysis. Previous CyberShake studies predict higher median ground motions relative to NGA-West2 Ground Motion Models (GMMs). Since 2022, two new CyberShake studies have been completed: 22.12 in Southern California and 24.8 in Northern California. Compared to previous CyberShake studies, there are several changes in the two new studies aimed at producing more realistic ground motions: 1) the rupture generator is updated to reduce the level of coherency in rupture propagation; 2) a taper is applied to the top of the 3D velocity models to replicate firm-rock conditions; 3) the minimum VS value in the simulation is reduced from 500 m/s to 400 m/s; and 4) VS30 are now obtained from Thompson (2022).

To evaluate the effects of the implemented changes, we develop partially non-ergodic GMMs based on the two new CyberShake studies and compare them to NGA-West2 GMMs. The median predictions of 22.12 show good agreement with the NGA-West2 GMMs at 2, 5 and 10s, supporting the implemented changes. However, the median predictions of CS24.8 are significantly larger than the NGA-West2 GMMs at 2s. Our analysis finds that very strong directivity effects in 24.8 play a crucial role for these large ground motions. Several factors contribute to the strong directivity in CS24.8: 1) All the major faults in 24.8 rupture toward the sites; 2) the rupture processes for events in 24.8 are much more coherent than 22.12, due to the geology-based velocity model. Moreover, differences in attenuation, crustal reflections and surface wave generation between the two 3D velocity models used in simulations are also important factors in affecting the ground motion level.

For the variance decomposition, we observe increasing variances of between-event residuals with periods, which are the results of stronger directivity with periods. We also find much smaller variances of site-to-site residuals for CyberShake GMMs, compared to NGA-West2 GMMs, which reflect the lack of near-surface structures in the velocity models. The variances of site-to-site residuals in 24.8 are consistently larger than that in 22.12, as the geology-based model is better at capturing some sharp boundaries, such as soil to bedrock or sediment layers.

These regional differences between the two CyberShake studies highlight the importance of continued non-ergodic GMM development.

Citation
Meng, X., Graves, R. W., & Callaghan, S. (2025, 09). Validation of two new CyberShake studies in California. Poster Presentation at 2025 SCEC Annual Meeting.

Related Projects & Working Groups
Ground Motions (GM)