Oral Presentation

Seismic hazard and risk analyses—which influence building codes, insurance models, and disaster preparedness—depend on Earthquake Rupture Forecasts (ERFs), which quantify future earthquake rupture probabilities. Unfortunately, many assumptions within current ERFs are difficult to test because large earthquakes are rare and historical records are limited. Multicycle, Physics-Based Earthquake Simulators (MP-BES, pronounced “MP-Bees”) are computational models designed to address this limitation by simulating fault systems over thousands of earthquake cycles. By explicitly modeling the physics of stress accumulation, rupture nucleation, friction evolution, and multi-fault interactions, these simulators generate long synthetic earthquake catalogs that can help evaluate current assumptions and overall predictability. Of course, these models will always be limited approximations as well, so the relevant question is not whether they are right or wrong, but whether any robust and useful inferences can be made from them. This presentation will illustrate this with two currently relevant questions: 1) does elastic-rebound predictability depend on magnitude and/or fault maturity; and 2) if one large rupture triggers another large overlapping rupture on the same fault, can the latter nucleate from well inside the rupture surface of the first event? These are just two of many consequential questions, and until more physics-based models become reliable, we will be forced to continue to rely on expert judgement.