A data-driven, multi-scale sediment velocity model for Southern California
Yi Liu, Grigorios Lavrentiadis, & Domniki AsimakiSubmitted September 7, 2025, SCEC Contribution #14749, 2025 SCEC Annual Meeting Poster #TBD
We present a non-parametric, data-driven near-surface velocity model for Southern California that can be used to populate the basin structures of SCEC CVM-S4.26. The model is developed as a conditional random field of uncertain trend function and uncertain fluctuations about the trend of the residuals relative to the SCEC CVM-S4.26, using Gaussian process regression with the superposition of stationary and spatially varying kernels. Results indicate that the kernel function that integrates geological information and uses sonic log data for conditional constraints at depth has optimal predictive ability among the structured kernel functions we tested. By integrating statistical modeling with geophysical priors, our model provides a robust, flexible, and interpretable solution for modeling shallow velocity structures in regions with broad engineering and scientific applications such as risk assessment of infrastructure systems and ground motion simulations of basin effects. In the future, we plan to integrate datasets with higher spatial resolution such as DAS inversion data to better capture the spatially varying components of the model, and work in collaboration with SCEC IT to implement the model on UCVM for testing in deterministic high frequency ground motion simulations.
Citation
Liu, Y., Lavrentiadis, G., & Asimaki, D. (2025, 09). A data-driven, multi-scale sediment velocity model for Southern California. Poster Presentation at 2025 SCEC Annual Meeting.
Related Projects & Working Groups
Community Earth Models (CEM)
