SELF-HEALING SLIP PULSES IN DYNAMIC RUPTURE MODELS DUE TO VELOCITY-DEPENDENT STRENGTH

Seismological observations of short slip duration on faults (short ris e time on seismograms) during earthquakes are not consistent with conv entional crack models of dynamic rupture and fault slip. In these mode ls, the leading edge of rupture stops only when a strong region is enc ountered, and slip at an interior point ceases only when waves from th e stopped edge of slip propagate back to that point. In contrast, some seismological evidence suggests that the duration of slip is too shor t for waves to propagate from the nearest edge of the ruptured surface , perhaps even if the distance used is an asperity size instead of the entire rupture dimension. What controls slip duration, if not dimensi ons of the fault or of asperities? In this study, dynamic earthquake r upture and slip are represented by a propagating shear crack. For all propagating shear cracks, slip velocity is highest near the rupture fr ont, and at a small distance behind the rupture front, the slip veloci ty decreases. As pointed out by Heaton (1990), if the crack obeys a ne gative slip-rate-dependent strength relation, the lower slip velocity behind the rupture front will lead to strengthening that further reduc es the velocity, and under certain circumstances, healing of slip can occur. The boundary element method of Hamano (1974) is used in a progr am adapted from Andrews (1985) for numerical simulations of mode II ru pture with two different velocity-dependent strength functions. For th e first function, after a slip-weakening displacement, the crack follo ws an exponential velocity-weakening relation. The characteristic velo city V-0 of the exponential determines the magnitude of the velocity-d ependence at dynamic velocities. The velocity-dependence at high veloc ity is essentially zero when V-0 is small and the resulting slip veloc ity distribution is similar to slip weakening. If V-0 is larger, ruptu re propagation initially resembles slip-weakening, but spontaneous hea ling occurs behind the rupture front. The rise time and rupture propag ation velocity depend on the choice of constitutive parameters. The se cond strength function is a natural log velocity-dependent form simila r to constitutive laws that fit experimental rock friction data at low er velocities. Slip pulses also arise with this function. For a reason able choice of constitutive parameters, slip pulses with this function do not propagate at speeds greater than the Raleigh-wave velocity. Th e calculated slip pulses are similar in many aspects to seismic observ ations of short rise time. In all cases of self-healing slip pulses, t he residual stress increases with distance behind the trailing edge of the pulse so that the final stress drop is much less than the dynamic stress drop, in agreement with the model of Brune (1976) and some rec ent seismological observations of rupture.