Ground-motion variability in SCEC CyberShake

Morgan P. Moschetti, Scott Callaghan, Kyle B. Withers, Xiaofeng Meng, & Robert W. Graves

Submitted September 7, 2025, SCEC Contribution #14727, 2025 SCEC Annual Meeting Poster #TBD

We analyze the simulated ground motions from CyberShake version 22.12 to identify the controlling features of variability from three-dimensional (3D) ground-motion simulations. Low-frequency CyberShake results (T≥2 s) consist of pseudospectral accelerations from seismic sources in the Uniform California Earthquake Rupture Forecast, version 2 (UCERF2) simulated at 335 stations in southern California using 3D wave-propagation codes for distances of up to 200 km. These results comprise one of the most complete sets of simulated ground motions available for earthquakes, 6.15≤MW≤8.45. Ground-motion variability, which is used directly to compute probabilistic seismic hazard curves in CyberShake, is governed by variations in slip and hypocenter location, along with minor, stochastic variations in total moment, rupture speed, and rise-time parameters. Unlike empirical ground-motion models, which compute variability from the standard deviation of ground-motion residuals relative to a median model, CyberShake allows variability to be directly estimated at each site from the standard deviation of the (natural logarithm) ground motions from each seismic source. We extract the subset of 2,680 seismic sources from UCERF2 that are simulated at all stations and compute standard deviations for all seismic sources and stations. Standard deviations in this dataset are comparable to “single-station sigma” values, as they exclude site-to-site variability (ϕS2S).

We examine standard deviations from individual sources and find that rupture directivity strongly influences variability for most seismic sources. We fit the azimuthal variations in standard deviation for each source, using a constrained sinc function, then remove modeled directivity effects to produce directivity-corrected standard deviations. We present mean trends in the fitted directivity parameters and in the directivity-corrected standard deviations. We find evidence of period-dependent trends in the corrected standard deviations with distance, showing lower standard deviations at greater distances. Additionally, we observe positive correlations between corrected standard deviations and earthquake magnitude, with larger effects at longer periods (T≥5 s). These magnitude-dependent trends may result from increased rise-time values in larger-magnitude earthquakes. Work is ongoing to better understand the effects of kinematic rupture parameters on ground-motion variability.

Key Words
ground motion, simulations, uncertainty

Citation
Moschetti, M. P., Callaghan, S., Withers, K. B., Meng, X., & Graves, R. W. (2025, 09). Ground-motion variability in SCEC CyberShake. Poster Presentation at 2025 SCEC Annual Meeting.

Related Projects & Working Groups
Ground Motions (GM)