A fault friction driven model of crustal stress in the Los Angeles region

We are using models of the dynamic friction on faults during earthquakes to create a model of the crustal stress in the Los Angeles region. The goal is to increment the displacement along the edges of the domain and allow earthquakes to control the evolution of the stress field within the volume. A wide variety of seismological and geodetic observations constrain the level of stress, including (1) rupture behavior, such as rupture speed and distribution of slip, inferred from kinematic source inversions of earthquakes, (2) the lack of heat flow found in exposed fault zones, (3) GPS and InSAR observations of plate motion, and (4) the need to support variations in topography and mass density. Comparison of synthetic strong ground motions with recorded motions will also provide feedback into the dynamic friction models. The model relies on two principle components. Dynamic rupture simulations of earthquakes provide a test bed for evaluating how well various friction models produce realistic rupture behavior and allow computation of the perturbations in the stress field caused by the slip on the fault. Static simulations produce estimates of the background stress field from gravity acting on topography and density variations, strain accumulation due to plate motion, and slip during prior earthquakes. Successive earthquakes within the region and subsequent earthquakes on a single fault will play an important role in discriminating between different friction models. Our preliminary results have focused on examining the behavior of ruptures with different friction models. We have found that (1) frictional sliding on fault surfaces is a highly nonlinear process that cannot be described by constant traction boundary conditions due to spatial and temporal variations in the magnitude and direction of the friction stress; (2) instantaneous healing of the fault upon termination of sliding accentuates the development of heterogeneity in the stress field and the distribution of slip; and (3) introducing rate-weakening behavior leads to pulse-like ruptures that tend to create stress fields and slip distributions with more heterogeneity for a given stress field than crack-like ruptures.