Why Do Two Very Close Stations in Central Japan Have Different Receiver Functions?

function (RF) analysis is one of the most powerful methods. In many cases, one draws spatial geometry of seismic velocity interface by using depth-migrated radial component RFs observed at several stations. On the other hand, many studies including Shiomi and Park (2008) have proposed the RF stacking method to estimate a depth of seismic velocity discontinuity and/or anisotropic features beneath each station. When we applied the RF stacking method to each station, we assumed the close stations have similar results. However, the depth and plunge azimuth of the Moho beneath two close stations, N.WATH and N.WTRF located in the Kii Peninsula, central Japan, were different as reported in Shiomi and Park (2008). These two stations are only 2.5 km apart from each other. Now, we focused on the data at these two stations and tried to understand why they have different RFs.

Although the distance between these stations is only 2.5km, the RFs are clearly different each other, especially transverse component. At N.WATH, the short-period (f_c = 1Hz) velocity seismograph is installed at the bottom of the borehole with depth of 110m. At N.WTRF, the STS-2 is equipped at the end of the tunnel. In the transverse RFs at station N. WATH, later phases with large amplitudes arrived 0~2s after the initial P wave for earthquake in SW direction. However, earthquakes in W~NW directions had large amplitude at N.WTRF. This characteristic is confirmed by checking the cross-correlation coefficients (CCs) of the teleseismograms used to estimate the RFs. The CCs of vertical, radial, and transverse components are about 0.8, 0.7 and 0.4, respectively. Thus, CCs of the teleseismograms from SW to NW were low for particularly transverse component.

This large discrepancy of transverse RFs might be due to the localized strong anisotropic crust and/or steep velocity discontinuity because these stations are located in the accretionary prism along the subducting Philippine Sea plate. In fact, we confirmed that the result could be affected by the crustal anisotropy by simple numerical check. We show the RF result in this region by considering the localized strong anisotropic crust and steep velocity discontinuity.