Evaluating Intra-slab Anisotropy using the Backazimuth Dependence of Shear Wave Splitting Patterns at West Pacific Subduction Zones

It has been shown that many deep subduction earthquakes (whose depth > 60 km) show large non-double couple components (ndcc) in the results of moment tensor analyses. These findings are used as evidence to argue that deep earthquake mechanisms were different from shallow earthquakes. Recently, it was shown that there is strong evidence of high seismic anisotropy in the vicinity of deep earthquakes (subducting slabs) which can cause the observed apparent ndcc. If this hypothesis is correct, an important consequence is that the strong anisotropy in the dipping slabs can cause the observed different shear wave splitting (SWS) patterns which critically depend on the back-azimuths of earthquakes as well as the slab anisotropy. It is in sharp contrast with many previous SWS studies which did not consider the source locations in the interpretation. We will evaluate this hypothesis in the Japan subduction zone, using Hi-net stations (>760) for global earthquakes. The preliminary results obtained from 21 stations located in Japan and 230 teleseismic earthquakes located mainly in Java and Tonga, suggest a variation of the delay time between the fast and slow S waves from about 0 s to ~3 s. Both the fast S polarization and the delay time have a complex relation with respect to the source location and epicentral distance. These measurements will be further used to test if a tilted transverse isotropy slab can cause the SWS observations using 3D-anisotropic elastic finite difference modeling and a propagator matrix method. It is expected that an intra-slab anisotropic model can simultaneously explain both the earthquake ndcc radiation patterns and the observed SWS patterns.