Geometry of crustal faults: Identification from seismicity and implications for slip and stress transfer models

Abstract

Geometric complexities of faults and fault systems are first‐order effects that complicate the mechanics of faulting and physics of earthquakes. We investigate the importance of this complexity using relocated seismicity from a catalog of events for the Joshua Tree–Landers earthquake sequence and present a new method to locate faults or fault systems in clouds of seismicity. The abundance and ever improved accuracy of earthquake and microseismic location of such events permits a better understanding of the geometric intricacies of fault systems. The proposed method assumes that seismicity is most abundant in the central fault zone and the spatial density of seismicity is used to locate finite width fault zones and construct fault surfaces from the seismicity. The method is different from statistical fits (e.g., L1‐ and L2‐norm fits) in that it does not suppose a decay of seismicity from the central fault zone and that it identifies the tiplines of faults from the fault zone seismicity directly. In the Joshua Tree–Landers earthquake catalog, the method identifies 10 separate fault segments ranging in average strike from north‐south to east‐west that compare well with surface trace fault maps. These faults exhibit significant nonplanarity with the Joshua Tree fault departing from a planar approximation by more than 2000 m. The mechanical effects of the geometrically complex fault surfaces are illustrated by inverting for coseismic slip using surface displacements and when compared to slip inversions on planar faults reveal a more complicated pattern of slip on the fault. The low RMS error in surface displacement, the good match to geodetic moment, and robust estimates of maximum slip compare well to the results for planar faults. The inverted slip distributions are used to solve the quasi‐static fault stress transfer problem and estimate the tractions changes by slip on the Joshua Tree fault on the fault segments involved in the Landers earthquake. We find that the propensity for slip on the Landers faults increased in regions of initiation and largest slip during the subsequent event. The geometrically complex models predict greater likelihood for slip along the northern faults involved in the Landers earthquake than the commonly used planar and vertical four‐fault models.