Oral Presentation

In practice, seismic hazard analysis and structural engineering draw heavily on empirical relations derived from recorded shaking (i.e., ground motion prediction equations). However, these ground motion models are limited in that they usually only account for very simple 1D features in the subsurface structure (Vs30 or bedrock depth), and ground motion measurements very close to the source are scarce. Alternatively, physics-based models can simulate realistic seismic wavefields that result from the combination of complex source, path, and site properties, and can synthesize ground motion time series at any chosen point in the volume. With continuing advances in computational power, development of realistic 3D Earth models, and increasingly sophisticated earthquake source models, physics-based modeling to inform seismic hazard analysis and earthquake-safe infrastructure design has the potential to become increasingly valuable. Still, a key step to gain confidence that these models are predicting realistic earthquake ground motions is to validate them with existing data.

In this talk, I will first describe recent efforts to validate synthetic motions with the USGS San Francisco Bay region 3D seismic velocity model (SFVM). We recently performed a suite of simulations for moderate (Mw3.5-4.5) earthquakes in the SF Bay region and compared synthetic and observed ground motions (Hirakawa and Aagaard, 2022). We found that by making some simple adjustments to velocity-depth relations, synthetic motions more accurately reproduce observed waveforms than with the previous version of the SFVM. These adjustments are included in the recently released update to the SFVM (v21.1.).

Next, I will highlight other advances from the computational seismology community, with emphasis on ground motion simulations and model validation for California earthquakes. This will include studies that mostly focus on effects from 3D velocity structure, as well as studies that focus more on earthquake source complexity. An underlying goal for the physics-based modeling community is to work towards fully-deterministic, end-to-end fault-to-structure simulations.