Seismic tomography and deformation modeling of the junction of the San Andreas and Calaveras faults

Local earthquake P traveltime data is inverted to obtain a three-dimensional tomographic image of the region centered on the junction of the San Andreas and Calaveras faults. The resulting velocity model is then used to relocate more than 17,000 earthquakes and to produce a model of fault structure in the region. These faults serve as the basis for modeling the topography using elastic dislocation methods. The region is of interest because active faults join, it marks the transition zone from creeping to locked fault behavior on the San Andreas fault, it exhibits young topography, and it has a good spatial distribution of seismicity. The tomographic data set is extensive, consisting of 1445 events, 96 stations, and nearly 95,000 travel time readings. Tomographic images are resolvable to depths of 12 km and show significant velocity contrasts across the San Andreas and Calaveras faults, a low-velocity zone associated with the creeping section of the San Andreas fault, and shallow low-velocity sediments in the southern Santa Clara valley and northern Salinas valley. Relocated earthquakes only occur where vp>5 km/s and indicate that portions of the San Andreas and Calaveras faults are non vertical, although we cannot completely exclude the possibility that all or part of this results from ray tracing problems. The new dips are more consistent with geological observations that dipping faults intersect the surface where surface traces have been mapped. The topographic modeling predicts extensive subsidence in regions characterized by shallow low-velocity material, presumably the result of recent sedimentation. Some details of the topography at the junction of the San Andreas and Calaveras faults are not consistent with the modeling results, suggesting that the current position of this "triple junction" has changed with time. The model also predicts those parts of the fault subject to contraction or extension perpendicular to the fault strike and hence the sense of any dip-slip component. In each locality the relative vertical motion accross the fault is consistent with the fault dips found with the new hypocentral locations.