Dual effect of roughness during earthquake rupture sequences on faults with strongly rate-weakening friction

We numerically model earthquake rupture cycles on non-planar faults governed by rate and state friction with an enhanced dynamic weakening in the form of flash heating, investigating quantitatively the relationship between the fault geometry, stresses from the preceding earthquakes, and the rupture characteristics. On planar faults, earthquakes with similar magnitudes and stress drops periodically rupture the entire fault, propagating under a low background shear-to-normal stress ratio as self-healing slip pulses due to the large friction reduction, with a sub-Rayleigh rupture speed. Faults with low roughness levels generally show a similar pattern. Prior to some events, the stress ratio along the fault slightly increases, leading to ruptures with secondary slip pulses. As roughness increases, the stresses become more heterogeneous, leading to a more complex cycle with part of the ruptures arresting at restraining bends with a low stress ratio. Sequences of several partial ruptures lead to a significant stress accumulation on the locked fault segments. These are eventually released by crack-like earthquake ruptures with supershear propagation speed and significantly larger magnitudes, stress drops, and radiated energy than typical planar fault events. Therefore, while fault roughness provides a mechanism for rupture arrest, consistent with previous studies, it can also result in a substantial increase in earthquake magnitudes, which should be considered in hazard assessments.