S21B-4446:
The effect of seismogenic zone depth on the likelihood of fault stepover jump 

Tuesday, 16 December 2014
Kangchen Bai and Jean Paul Ampuero, California Institute of Technology, Pasadena, CA, United States
Abstract:
Earthquakes can involve rupture of geometrically complex fault systems (e.g. 1992 Landers earthquake, 2012 Indian Ocean earthquake), including fault stepovers, i.e. offsets in the fault trace or gaps between fault segments. Quantifying the likelihood of multi-segment ruptures is important to evaluate earthquake hazard. Empirical observations suggest that earthquakes rarely jump across stepovers that have a gap larger than about 5 km. Although the data cannot distinguish between a random or a causal relation between gaps and rupture arrest, a maximum stepover distance for rupture jump has been previously shown to emerge in 2D dynamic rupture simulations and 2.5D earthquake cycle models. The mechanical origin of this critical stepover distance (Rc) is not fully understood yet. Here, we examine one possible mechanism that has not been explored extensively: the role of the finite seismogenic zone depth, which places a limit on the stress intensity factor and hence on the distance reached by off-fault large stresses.

Here we use a 3D spectral element method coupled with a dynamic rupture solver to investigate the possibility of rupture jump across fault stepovers. We conduct spontaneous rupture simulations under slip-weakening friction incorporating the finite seismogenic zone depth (W) and we calculate the dynamic Coulomb stress perturbation surrounding the primary fault. Simulation results confirmed the depth (W) control over the stress intensity factor while providing three new insights. (1) The normalized critical stepover distance Rc/W is determined by the ratio Lc/W (where Lc=(shear modulus)*(critical slip distance)/(strength drop) is the characteristic size of the slip-weakening process zone), which controls the amplitude of the strong seismic phase radiated at the end point of the primary fault. (2) The stopping phase is more efficient at promoting rupture across compressional stepovers than dilatational ones, which suggests the possibility of a larger Rc in compressional stepovers than in dilatational ones. (3) The Lorentz contraction at the front of a steady rupture substantially distorts the angular pattern of the Coulomb stress distribution, leading to an increase of Coulomb stress at a certain azimuth in the dilatational side.