Aseismic creep in steady-state rate-weakening interfaces

Abstract:

We investigate aseismic creep on model faults. In particular, its propagation to large distances along the fault. This study considers a sliding rate and state dependence of the fault’s frictional strength, with steady-state rate-weakening interfacial properties. The surrounding medium deforms elastically. We find that rate-weakening interfaces can exhibit a stable aseismic sliding over long distances without any stabilizing mechanisms (e.g., dilatant strengthening) or choice of frictional properties that force a large elasto-frictional lengthscale. We propose that creep penetration is due to the fault's tendency, when driven, to reach a state of steady sliding from a locked state. Here, we define the locked state as the state of slip when the velocity is arbitrarily less than the steady-state sliding rate for a given pre-stress.

The slip is driven by an imposed dislocation accrued at a constant rate at one end with the other end either at (1) the free surface of an elastic half-space or (2) strictly locked (buried) in the elastic full space. We find that when fault size is large, a critical run-out distance (its half for the case 2) exists at which creep transitions to a localized instability. The creep run-out distances, however, are not relatable to the usual elasto-frictional nucleation lengthscales (e.g., Rubin & Ampuero, 2005). We show that the system of sliding rate and state during creep propagation exhibits slow-fast dynamics when expressed in similarity coordinates. At the creep run-out distances (and times), loss of stability of the creep leads to the transition to a localized fast sliding. We also explore a non-linear system of ODEs (equation of motion of an equivalent singular crack) that governs the evolution of the creep front’s speed and rupture length.

When fault size is small, with the buried end, the penetrating creep does not provoke any instability. We can consider that the creep-front tries to smooth any velocity gradient whereas a faster 'lock-front' (case 2) from the other end steepens a velocity-gradient. For small faults, the creep could not reach the critical distance to nucleate any localized instability. However, the free end leads to the nucleation of the first and all subsequent dynamic events of the emerging cycle at/near the free surface after the creep traversed the entire length of the fault.