Abstract
To image the subducting Pacific plate better, we have extended the treatment of Kawakatsu and Watada (2007, Science) in which they had corrected for the effect of a dipping interface on seismic receiver functions (RF) to image a low-velocity layer atop of the subducting slab. The dip angle of the Pacific plate estimated from seismicity is employed to correct the effect of the dipping interface. For each potential conversion point, only a P-S conversion which satisfies the Snell's law on the dipping interface is used for RF. Two horizontal component RFs are then rotated to the direction of expected polarization of P-S converted waves from the interface, and stacked at the conversion point in such a way that the amplitude corresponds to the possible S-wave velocity jump at the interface. This method is applied to Hi-net recording of teleseismic events from 2001 to the end of 2006. The total number of events analyzed is 681, and the number of RFs is more than 300,000. The results show a clear image of a bottom boundary of the subducting Pacific plate; the thickness of the plate is estimated to be ~80km (Tonegawa et al., 2006, EPSL). Below 350km right beneath central/southwestern Japan, there also exist signatures inside of the slab which we attribute to those originated from the postulated meta-stable olivine wedge (MOW; Iidaka and Suetsugu, 1992, Nature). We observe both velocity decrease (from shallow to deep) and increase corresponding respectively to the upper and lower edge of the MOW which is expected to have several percent slower seismic velocity relative to the surrounding normal slab (Kaneshima et al., 2007, EPSL). The catalogue seismicity by JMA indicates that deep earthquakes are located along the lower edge of the MOW. The detailed investigation of the relative locations of these features should give a tight constraint on the origin of deep earthquakes. The existence of the MOW indicates insignificant amount of water present in the subducting slab in the region (Hosoya et al., 2005, GRL); together with the observed deep depression (~40km) of the 660-km discontinuity in the same area, the effective Clapeylon slope of dry slab for the 660km discontinuity should be significantly steeper than those predicted by recent high-pressure experiments (e.g., Katsura et al., 2003, PEPI).