Slip rates and interseismic locking depths of southern California faults inferred from viscoelastic earthquake cycle models

Estimates of fault slip rates and interseismic fault locking in southern California are important for seismic hazard assessments. However, some model estimates of slip rates from geodetic data are lower than geologic rates on the Garlock Fault and Mojave and San Bernadino segment of the San Andreas Fault, and locking depth, which varies along the strike, is poorly resolved. We designed several classes of lithosphere block models to examine the effect of model assumptions on slip rate and locking depth estimates. We show that a viscoelastic earthquake cycle model constrained by GPS data predicts slip rates that are entirely consistent with geologic slip rate estimates an all major faults in southern California. We found that the discrepancy between the geological and geodetic rates depends on the model used to fit GPS data. Our viscoelastic earthquake cycle model consists of fault-bounded blocks in an elastic crust overlying a viscoelastic lower crust and uppermost mantle. Interseismic locking of faults and associated deformation is modeled with steady back-slip on faults and imposed periodic earthquakes. Based on comparisons of an elastic block model and our viscoelastic cycle model, we conclude that elastic block models tend to underpredict slip rates on the Mojave and Carrizo segments of the San Andreas Fault and the Garlock Fault because these faults are mid to late in the earthquake cycle and current strain rates across these faults are lower than average due to viscous relaxation of the lower crust. We conclude that elastic block models overpredict the composite slip rate across the southern Mojave ECSZ because these faults are in the early phase of the composite earthquake cycle and current deformation rates across this region are higher than average because of accelerated viscous flow in the lower crust. We consider the influence of model assumptions on locking depth estimates by simultaneously estimating the distribution of interseismic fault locking depths and fault slip rates. In one model, we assume deformation is steady with time, and faults are either locked or creeping at constant resistive shear stress. In inversions with this model, we solve for the distribution of locked and creeping patches on faults. In another class of lithosphere block models, we impose periodic earthquakes and consider time-variable viscous flow of the asthenosphere. In this model we assume that faults are locked above some depth during the interseismic period and creep below this depth. Preliminary results show that the steady model favors deep locking depths, while the earthquake cycle model prefers moderate locking depths for the San Andreas fault system.