Summary

SCEC hosted a workshop to accelerate the development and improvement of multi-scale seismic velocity models that are essential for accurate ground motion simulations and seismic hazard assessments in California and beyond. While significant progress has been made, challenges remain in integrating new data, improving resolution, merging models, and assessing uncertainties. This workshop focused on planning a coordinated research approach to develop multi-scale P- and S-wave velocity models for the San Andreas fault system in the Western US, targeting seismic wavefield simulations up to 5-10 Hz, a frequency range critical for engineering applications. Participants discussed tomography workflows, strategies for merging high-resolution local models and regional models, integrating diverse geophysical data, and developing robust uncertainty quantification methods. The workshop also addressed the development of essential IT tools for model management and access, ensuring practical implementation of workshop recommendations. Discussions at the workshop will guide the development of the next generation of SCEC CVMs, which will advance our understanding of crustal structures and seismic hazards.

Workshop Objectives

Community Velocity Models (CVMs) provide foundational information for many basic and applied topics including the determination of earthquake locations and other source properties, imaging of the subsurface, and simulations of ground motions for use in seismic hazard models. However, current large-scale velocity models within the state of California lack the deterministic information needed for ground motion simulations at the high frequencies required in engineering applications. To address this limitation, the next generation of CVMs should resolve fine-scale seismic structure, in particular in basins, the shallow crust, and around fault zones to allow simulations of ground motion in areas of high population density and critical infrastructure to frequencies of interest to engineers. The next generation of CVMs should be multi-scale while also covering the full spatial domain of the San Andreas fault system (SAFS) in the Western US. This workshop focused on advancing the development of such multi-scale velocity models that would be needed to uniformly develop a statewide seismic hazard model and for other applications. The workshop was divided into three main parts. The first session aimed at summarizing the state-of-the-art in current velocity (VP and VS) models available for the SAFS in the Western US with a focus on the different methodologies, resolutions, datasets, and spatial domains of these models. This session also covered the topic of critical data gaps and data needs such as underexplored geographic domains, full seismic characterization of crustal properties (seismic anisotropy and seismic attenuation), and model uncertainties. The second session discussed current approaches for combining velocity models to produce multi-scale models and the methods used to evaluate these models with an emphasis on improving future ground motion predictions across a broad range of frequencies. The third session introduced current IT tools available for interacting with, maintaining, querying, and visualizing models. Workshop discussions focused on future directions and the short- and near-term goals for critical data gathering, development of new methodologies, and resource needs.

Workshop Outcomes

The workshop saw a wide range of user needs and interests from the attendees including discussions on improving the spatial resolution of the CVMs; accurately representing the offshore regions, sedimentary basins and near-surface seismic velocities; implementing realistic seismic attenuation in ground motion simulations; using non-seismic datasets in model development and validation; and developing multiple velocity models that capture the epistemic uncertainty in the 3D seismic structure. In contrast, the preworkshop survey showed, fewer registrants had previously developed models, datasets, or merging algorithms to contribute to enhancing ground motion simulations and seismic hazard assessments in the Western US. This suggests an emerging interest in multi-scale CVMs, of which only a few examples currently exist. Many of the discussions pointed to a long-term research effort noting that recent improvements in the spatial density of datasets and novel techniques including machine learning based methods now facilitate underexplored approaches for building CVMs as well as new applications for the CVMs. Such applications ranged from near-surface to deep-crustal and lithospheric-scale studies. Based on the broad range of applications for multi-scale CVMs highlighted at the workshop including the seismic hazard applications, there is a need for cross-disciplinary discussions and collaborations in the early-stages of the development of these models that should also take into consideration how they are validated and maintained. Discussions at the workshop emphasized four essential areas listed below where a coordinated research effort in developing multi-scale P- and S-wave velocity models for the SAFS in the Western US should focus on making progress.