Faults are complex structures: they are geometrically complex, and have variable stress conditions and frictional behaviors along their length. In addition, faults exist in heterogeneous settings, in terms of surrounding geology, and in terms of regional and local stresses. These individual types of heterogeneity all contribute to complex dynamic rupture behaviors and ground motion distributions, as inferred from observational data and supported by previous modeling studies. In this study, we investigate the effects of individual types of complexity, and we combine different types of heterogeneity in order to enhance the realism of models of real-world faults.

We use the finite element method to conduct dynamic rupture models of earthquakes on faults with complex geometry, initial stresses, frictional parameters, and surrounding geology, and with combinations of these factors, in order to investigate the effects of this complexity on fault interactions, rupture extent, and ground motion. In particular, we investigate the effect of critical weakening distance on the ability of rupture to propagate through a discontinuity in the fault trace, the effect of a small fault between the larger strands of a stepover on the ability of rupture to jump the stepover, and how zones of aseismic creep affect rupture through locked portions of the same fault. We also construct realistically complex models of the northern San Jacinto Fault, California, incorporating realistic geometry, velocity structure, and combined regional and stochastic stress fields.

We find that the distribution of complexity of any type on the fault, which leads to heterogeneous fault strength, has the primary controlling effect on rupture behavior. The relative strength or weakness of the fault, rather than the actual value, is most important. We also find that the balance of the energy budget is crucial; if too much energy is redirected into fracture, rupture stops. Lastly, we find that each type of complexity affects rupture in its own way, but it is difficult to separate out the effects of individual factors in a model that incorporates many types of heterogeneity. We therefore emphasize the importance of including as many types of realistic complexity as feasible when modeling real faults.