Modeling Supershear Rupture Propagation on a Fault with Heterogeneous Surface

The formation of supershear rupture (with a slip front propagation velocity higher than S-wave velocity in the material) along a model fault with homogeneous and heterogeneous surface is analyzed based on the results of numerical simulation. Heterogeneity of the properties is specified by friction spots weakening at shear, interspersed with stable friction segments between them. This problem statement goes back to the well-known asperity model (Kanamori and Stewart, 1978). In this paper, we use seismological and geodetic data to estimate characteristic sizes of the fault-zone segments that are locked during interseismic period. Calculations show that the characteristic sizes of inhomogeneities on the slip plane largely determine the pattern of dynamic rupture propagation. A necessary condition for rupture to pass into supershear is a sufficiently rapid frictional weakening. The observed wavefield features (relatively weak attenuation, an increase in the amplitude of oscillations at some distance from the hypocenter, the predominant motion in the direction parallel to the fault, etc.) do not necessarily need nonlinear medium and are not the result of generation of a "shock wave," as assumed in some publications, but are only a result of wavefront interference. Interaction between regions with different frictional properties can cause rupture transition to supershear, induce dynamic slip pulses which re-fracture the fault segments previously displaced by the creep process, and slowdown rupture propagation. Judging by the results of calculations, rupture is more likely to accelerate into supershear on rough/undulating segments of contact surfaces with closely spaced frictional weakening zones. At the same time, propagation of such a rupture with a decaying displacement amplitude can be stable on locally smoother segments.