UC Riverside

This dissertation seeks to contribute to earthquake hazard analyses and forecasting by conducting a detailed study of the processes controlling the occurrence, and particularly the clustering, of large earthquakes, the probabilities of these large events, and the dynamics of their ruptures. We use the multi-cycle earthquake simulator RSQSim to investigate several fundamental aspects of earthquake occurrence in order to improve the understanding of earthquake hazard. RSQSim, a 3D, boundary element code that incorporates rate- and state-friction to simulate earthquakes in fully interacting, complex fault systems has been successful at modeling several aspects of fault slip and earthquake occurrence. Multi-event earthquake models with time-dependent nucleation based on rate- and state-dependent friction, such as RSQSim, provide a viable physics-based method for modeling earthquake processes. These models can provide a better understanding of earthquake hazard by improving our knowledge of earthquake processes and probabilities. RSQSim is fast and efficient, and therefore is able to simulate very long sequences of earthquakes (from hundreds of thousands to millions of events). This makes RSQSim an ideal instrument for filling in the current gaps in earthquake data, from short and incomplete earthquake catalogs to unrealistic initial conditions used for dynamic rupture models. RSQSim catalogs include foreshocks, aftershocks, and occasional clusters of large earthquakes, the statistics of which are important for the estimation of earthquake probabilities. Additionally, RSQSim finds a near optimal nucleation location that enables ruptures to propagate at minimal stress conditions and thus can provide suites of heterogeneous initial conditions for dynamic rupture models that produce reduced ground motions compared to models with homogeneous initial stresses and arbitrary forced nucleation locations.