Crustal deformation occurs both as localized slip along faults and distributed deformation off of faults. While there are few robust estimates of off‐fault deformation in nature, scaled physical experiments simulating crustal strike‐slip faulting allow direct measurement of the ratio of fault slip to regional deformation, quantified as kinematic efficiency (KE). We offer an approach to predict KE using a 2D convolutional neural network (CNN) trained directly on fault maps produced by physical experiments. Experiments with different loading rates and basal boundary conditions generate the fault maps throughout the evolution of strike‐slip faults. Strain maps allow us to directly calculate KE and its uncertainty, utilized in the loss function and performance metric. The trained CNN achieves 91% custom accuracy in the KE prediction of an unseen data set. Although the CNN model is trained on scaled experiments, it can predict off‐fault deformation of crustal faults that matches available geologic estimates.

Earthquake hazard assessments rely on observations from the field and geophysical data that provide fault slip rate estimates at specific sites and inform the geometry of active faults; however, uncertainty remains for both slip rate and geometry. Furthermore, incompatibilities between inferred fault geometry and geologic slip rates arise within crustal deformation models where model and geologic slip rates disagree. The impact of these incompatibilities may be local to sites or have wider effect on the fault system deformation. Here, we investigate the roles of structural position of sites and uncertainty of slip rates using three-dimensional mechanical models that simulate deformation across many earthquake cycles along southern San Andreas fault near the San Gorgonio Pass in California. Within the models, the impact of strike-slip rate sites on the fault system depends on their structural positions. Slip rates at sites along short and segmented faults has lesser impact on the slip along the fault system than either slip rates at sites along longer faults or at sites within fault branches. Consequently, inaccuracies in the slip rate estimates used for seismic hazard assessment may have differing impacts on the fault system depending on location and structural position of the slip rates. Fault branches along strike-slip faults warrant detailed investigation not only because these areas have high spatial variability of slip rate and accrue nearby off-fault deformation but also because changes in slip rates along branches has larger impact on deformation along fault system than other sites. Lack of data or large uncertainty in slip rate data from fault branches can affect our ability to accurately assess seismic hazard of the region.

Recent field studies provide evidence of fault slip-rate variability over time periods of 10–100 k.y., yet researchers do not know how processes internal to the fault system (e.g., fault reorganization) impact records of fault slip rates. In this study, we directly observed fault-system evolution and measured slip-rate histories within a scaled physical experiment of a dextral strike-slip 15° restraining bend representative of a gentle crustal restraining bend. To assess the degree of slip-rate variability at particular sites along the experimental faults, such as would be revealed in a field study, we tracked fault slip rates at specific locations that advected throughout the experiment with accrued fault slip. Slip rates increased or decreased (5%–25% of the applied velocity) both during fault reorganization (e.g., fault growth and abandonment) and as sites migrated to new structural positions. Sites that advected into the restraining bend showed decreased slip rate. While we expect new fault growth to reduce slip rates along nearby fault segments, we document that the growth of new oblique-slip faults can increase strike-slip rates on nearby fault segments. New oblique-slip thrust faults within the experiment accommodated off-fault convergence and unclamped nearby strike-slip segments. The experimental results show that even under a constant loading rate, slip rates at sites located on stable fault segments can vary due to either reorganization elsewhere in the fault system or site advection.

Crustal deformation models show incompatibility between inferred fault geometry and geologic slip rates where model and geologic slip rates disagree. We do not know if the impact of these incompatibilities is limited to near sites or have wider effect on the fault system deformation. Here, we investigate the roles of structural position of sites and uncertainty of slip rates using one suite of mechanical models that limits the dextral slip rates to within the range of observed slip rates at the sites of geologic investigations and another suite that explores the impact of each slip rate site on each other and on the nearby fault system. The suites of models employ two viable configurations for the southern San Andreas fault: with and without an active northern slip pathway around the San Gorgonio Pass. While the Active Northern Pathway model has greater mismatch to geologic slip rates < 16ka, it produces lesser off-fault deformation than the Inactive Northern Pathway model. The impact of strike-slip rate sites on the system depends on their structural positions. Sites along segmented faults may have lesser impact than either sites along the same fault segment or sites at fault branches. Consequently, inaccuracies in the slip rates used for seismic hazard assessment may have differing impacts depending on location of the slip rates. Fault branches along strike-slip faults warrant detailed investigation because these areas have high spatial variability of slip rate and accrue nearby off-fault deformation, affecting our ability to accurately assess seismic hazard of the region.

Inversions of interseismic geodetic surface velocities often cannot uniquely resolve the three-dimensional slip-rate distribution along closely spaced faults. Microseismic focal mechanisms reveal stress information at depth and may provide additional constraints for inversions that estimate slip rates. Here, we present a new inverse approach that utilizes both surface velocities and subsurface stressing-rate tensors to constrain interseismic slip rates and activity of closely spaced faults. We assess the ability of the inverse approach to recover slip rate distributions from stressing-rate tensors and surface velocities generated by two forward models: 1) a single strike-slip fault model and 2) a complex southern San Andreas fault system (SAFS) model. The single fault model inversions reveal that a sparse array of regularly spaced stressing-rate tensors can recover the forward model slip distribution better than surface velocity inversions alone. Because focal mechanism inversions currently provide normalized deviatoric stress tensors, we perform inversions for slip rate using full, deviatoric or normalized deviatoric forward-model-generated stressing-rate tensors to assess the impact of removing stress magnitude from the constraining data. All the inversions, except for those that use normalized deviatoric stressing-rate tensors, recover the forward model slip-rate distribution well, even for the SAFS model. Jointly inverting stressing rate and velocity data best recovers the forward model slip-rate distribution and may improve estimates of interseismic deep slip rates in regions of complex faulting, such as the southern SAFS; however, successful inversions of crustal data will require methods to estimate stressing-rate magnitudes.