Three-dimensional mechanical simulations of the San Andreas fault system within the Coachella Valley in southern California produce deformation that matches geologic observations and demonstrate the firstorder impact of fault geometry on uplift patterns. To date, most models that include the Coachella Valley segment of the San Andreas fault have assumed a vertical orientation for this fault, but recent studies of seismicity and geodetically observed strain suggest that this segment of the fault may dip 60°-70° to the northeast. We compare models with varied geometry along this segment of the fault and evaluate how well they reproduce observed uplift patterns in the Mecca Hills and Coachella Valley. Incorporating wellconstrained fault geometry in regional models will provide a more accurate understanding of active faulting in southern California, which is critical for rupture and hazard modeling that is used to identify regions most susceptible to earthquake damage.We have tested three boundary-element method models for the active geometry of the Coachella Valley segment of the San Andreas fault: one contains a vertical Coachella segment, the second contains a northeast ~65° dipping Coachella segment, and the fi nal alternative contains a vertical Coachella segment plus a subparallel northeast-dipping fault at depth. This fi nal model honors the geometric interpretation of seismicity from the Southern California Earthquake Center Community Fault Model version 4.0. The models containing vertical Coachella Valley segments both produce uplift between the San Andreas and San Jacinto faults that is more uniformly distributed than geologic observations suggest, and these models fail to produce uplift in the Mecca Hills. The dipping model produces tilting of the Coachella Valley consistent with geologic observations of tilting between the San Jacinto and San Andreas faults. The dipping model also produces relative subsidence southwest of the fault and localized uplift in the Mecca Hills that better match the geologic observations. These results suggest that the active Coachella Valley segment of the San Andreas fault dips 60°-70° to the northeast.

Between ~1.5 and 1.1 Ma, the southern San Andreas fault system underwent a major reorganization that included initiation of the San Jacinto fault zone and termination of slip on the extensional West Salton detachment fault. The southern San Andreas fault itself has also evolved since this time, with several shifts in activity among fault strands within San Gorgonio Pass. We use three-dimensional mechanical Boundary Element Method models to investigate the impact of these changes to the fault network on deformation patterns. A series of snapshot models of the succession of active fault geometries explore the role of fault interaction and tectonic loading in abandonment of the West Salton detachment fault, initiation of the San Jacinto fault zone, and shifts in activity of the San Andreas fault. Interpreted changes to uplift patterns are well matched by model results. These results support the idea that initiation and growth of the San Jacinto fault zone led to increased uplift rates in the San Gabriel Mountains and decreased uplift rates in the San Bernardino Mountains. Comparison of model results for vertical-axis rotation to data from paleomagnetic studies reveals a good match to local rotation patterns in the Mecca Hills and Borrego Badlands. We explore the mechanical efficiency at each step in the modeled fault evolution, and find an overall trend toward increased efficiency through time. Strain energy density patterns are used to identify regions of incipient faulting, and support the notion of north-to-south propagation of the San Jacinto fault during its initiation.

No abstract

Unraveling the details of fracture propagation leading to catastrophic rock failure is critical for understanding the precursors to earthquakes. Here we present numerical simulations of fracture growth using a work optimization criterion. These simulations apply work optimization to fracture propagation by finding the propagation orientation that minimizes the external work at each increment of fracture growth, repeating this process for each growing fracture tip in the model. We simulate published uniaxial compression experiments performed on a cylinder of marble with pre-cut fractures of varied lengths, orientations, and positions. This suite of experiments provides an ideal benchmark for the numerical simulations because of the relatively simple boundary conditions and the range of pre-cut fracture geometries that focus deformation. We compare the results of homogeneous, isotropic model material to results that incorporate hundreds of small randomly oriented and distributed microcracks representing internal weaknesses, such as grain boundaries. From these numerical models, we find that slip on and propagation of microcracks governs the non-linear stress-strain response observed before failure under axial compression. We use a suite of Monte Carlo realizations incorporating different initial seeding of microcracks to explore the range of fracture propagation paths that might result from inherent variation between rock samples. We find that while models that include microcracks begin to propagate fractures at smaller cumulative axial strains than an equivalent homogeneous isotropic model, ultimately, models including heterogeneity require more energy to reach failure than the homogeneous model. These results highlight the critical role of heterogeneity, such as microcracks, within the processes leading up to failure.

No abstract