Structures of the continental shear zone beneath the San Andreas Fault inferred from two-layer modeling of SKS splitting

How continental plate boundary faults develop with depth has been under debate. We inverted SKS shear wave splitting data along the San Andreas Fault (SAF) into two layers of anisotropy using a Bayesian inversion. While the two layers are statistically required, the fast polarization directions of the upper layer do not match the strike of the SAF as previously reported. To capture the lithospheric shear zone, we progressively decrease the upper limit of the delay time of the upper layer. In northern California where SAF strikes 140-150, the upper layer fast directions get close to these azimuths when the delay time is reduced to ~0.5 s. In southern California where the SAF strikes 120-130, the upper layer fast directions capture the SAF with delay times of a similar magnitude. For olivine LPO with vertical shear plane, these delay times translate to a anisotropy layer of 40 km thickness, or a depth of 70 km from the surface, assuming the seismogenic zone in the crust is too localized to influence the SKS splitting. This depth coincides with the depth of lithosphere-asthenosphere boundary independently estimated. The lower-layer fast directions are in between the absolute plate motion directions of the American and the Pacific plates, or at least agree with that predicted from surface wave study in northern California. We picture a vertical continental shear zone widening to at least 200 km at the bottom of the lithosphere, transitioning to a horizontal shear regime in the asthenosphere driven by plate motions. This architecture of continental shear zone is consistent with our understanding of the rheology of crust and mantle.