On-Fault Parameters and Fault Connectivity Control Synthetic Earthquake Catalogs and 30-Year Rupture Probabilities in RSQSim Simulations of the Eastern San Francisco Bay Area

Evan Marschall , Alice-Agnes Gabriel, Bruce Shaw, Jeena Yun, Albert Kottke, Jennifer Donahue, Michael Eugene Oskin

The faults in the Eastern San Francisco Bay Area in California have long been known as a source of seismic hazard. This potential for large earthquakes, paired with high population density and economic activity leads to high seismic risk in the Bay Area. In this study we utilize the physics-based earthquake simulator RSQSim to generate synthetic earthquake catalogs for the San Francisco East Bay. We run a suite of models with different fault connectivity and on-fault parameters. We compare the model catalogs with UCERF3 probabilities and paleoseismic data. We find that initial on-fault assumptions of friction, weakening and connectivity shape the resultant catalogs. Firstly, we find that all our catalogs have b-values greater than 1, with models with uniform velocity-weakening and strong coseismic weakening having the lowest b-values (~1.4). The b-values are higher (1.5~1.8) for models with velocity-neutral and velocity-strengthening patches. However, if only fitting Mw 5 and greater, all model b-values fall within ~1.0-1.2. Secondly, we observe connected faults leading to more multi-segment slip events for the Hayward and Calaveras faults. We also observe good agreement between the RSQSim models and the UCERF3 Time-Independent and Time-Dependent probabilities of Mw 6.7 or greater events on the Hayward and Calaveras faults, within the maximum and minimum probability bounds. Models without velocity-strengthening yield time-independent 30-year probabilities of 22.4-28.9% for the Hayward and 13.3-20.5% for the Calaveras and time-dependent probabilities of 17.3-42.2% and 12.5-25.3% respectively. Finally, we compare model results to the mean recurrence intervals at three paleoseismic sites using a Welch’s t-test. For all but three models (all of which have uniform velocity-weakening friction) we can reject the equality of the mean recurrence rate between the paleoseismic data at Tyson Lagoon and the model catalog at the 95% confidence level, motivating future work exploring the full spectrum of fault slip.