Anisotropic feature inferred from receiver functions and S-wave splitting in and around the high strain rate zone, central Japan

By the recent dense GPS observation, the high strain rate zone (HSRZ) crossing the central Japan was discovered. In the HSRZ, E-W compressive stress field is observed, and large earthquakes with M>6 are frequently occurred. In this study, we try to reveal depth-dependent anisotropic feature in this region by using teleseismic receiver functions (RFs) and S-wave splitting information. As a target, we select NIED Hi-net stations N.TGWH and N.TSTH, which are located inside and outside of the HSRZ respectively. For RF analysis, we choose M>5.5 teleseismic events from October 2000 to November 2011. Low-pass filters with fc = 1 and 2 Hz are applied to estimate RFs. In the radial RFs, we find clear positive phase arrivals at 4 to 4.5 s in delay time for both stations. Since this time delay corresponds to 35 km-depth velocity discontinuity existence, these phases may be the converted phases generated at the Moho discontinuity. Seeing the back-azimuth paste-ups of the transverse RFs, we can find polarity changes of later phases at 4 to 4.5 s in delay time at the N.TSTH station. This polarity change occurs for direction of N0E (north), N180E (south), and N270E (west). Although we have no data in N90E (east) direction, this feature implies that anisotropic rocks may exist around the Moho. In order to check this feature, we consider 6-layered subsurface model and compare synthetic RFs with the observation. The first three layers are for thick sediments and upper crust including a dipping velocity interface. The fourth, fifth and sixth layer corresponds to the mid crust, lower crust and uppermost mantle, respectively. The best model infers that the mid- and lower-crust beneath the N.TSTH station should have strong anisotropy whose fast axis directs to the N-S, though the fast axis in the uppermost mantle seems to show E-W direction. Moreover, to explain the observation, the symmetric axes in the lower crust and the uppermost mantle should be dipping about 20 degrees. To check anisotropic feature of this station, we also apply S-wave splitting analysis to the local events. In order to avoid contaminations of scattered phases, we select seismic waveforms with incident angle less than 35 degrees. We select good S/N records and apply 2-8 Hz butter-worth type band-pass filter to the waveforms. Then, we estimate the leading S wave polarization direction (LSPD) and delay time of each event. At the N.TSTH station, we can select crustal earthquakes (< 30 km in depth) and the deep earthquakes (> 90 km) which occur along the subducting Pacific slab. For deeper events, LSPD shows two possibilities: ENE-WSW or NNW-SSE. On the other hand, only for shallow events, LSPD indicates NS. This result is consistent with the feature of RFs. We can conclude that the crustal anisotropic feature beneath the station N.TSTH corresponds to the lineament on the ground surface, not to the E-W compressive stress field. The LSPD in the uppermost mantle reflects to the lattice-preferred orientation of anisotropic minerals beneath this station.