Converted Ps amplitude variations on the dipping slab Moho beneath the Kii Peninsula, central Japan

Receiver function (RF) analysis is a very useful method to detect seismic velocity discontinuities beneath a seismic station. One can estimate the depth of interfaces using the delay time of Ps converted phases in RFs. One can also evaluate elastic properties at an interface from changes of Ps polarity and Ps amplitude. Ps amplitude depends primarily on the impedance contrast at an interface, but the variation of Ps amplitude on back azimuth (BAZ) of the incoming P wave is affected if the interface is dipping and/or anisotropic rock surrounds the interface. Moreover, variation in the incidence angle of incoming P waves also causes variation in Ps amplitude. Shiomi and Park (2009; AGU FM) defined 'standard amplitude' of a converted phase at a dipping interface beneath a station, based on back azimuth dependence of the Ps amplitude. They did not consider ray parameter dependence to the Ps amplitude evaluation, but it affects the accuracy of the standard amplitude estimate. In this study, we check ray parameter dependence, and apply the method to the stations in the Kii Peninsula, central Japan. We select earthquakes with high signal-to-noise ratio observed from October 2000 to August 2010 with magnitudes 6.0 or greater. Checking distribution of BAZs and incidence angles of teleseismic waveforms observed at each station, we confirm that 80% of the selected events are located to the south of stations. Ray parameters of 10% of the events are larger than 0.077. In the case of dipping interface, Ps amplitude variation with BAZ is larger for incoming P waves with larger ray parameters. Since events located in the west or northeast of stations are fewer than other directions, the contribution of events with large ray parameter is not small in these directions. These directions correspond to the dipping direction of the subducting Philippine Sea slab, thus the Ps amplitudes tend to become large. This means the Ps amplitude may be overestimated when we do not take ray parameter dependence into account. In order to avoid this contamination, we select limited events with ray parameters in the range 0.055-0.077. We also apply amplitude corrections. We confirmed that the Ps amplitudes decrease from 11% of the primary P wave to 7% as the oceanic Moho deepens to 40km. Deeper than 40 km, the Ps amplitudes are roughly constant, at 5-7% of the primary P wave. According to the P-T diagram of the Kii Peninsula region, we say that the Ps amplitude decrease likely reflects a phase transition from lawsonite blueschist to lawsonite-amphibole eclogite as water is released to the overlying layer, and this metamorphic fluids likely influence the occurrence of low-frequency nonvolcanic tremor, as we already reported. On the one hand, the regionality of standard amplitude distribution within the Kii Peninsula becomes unclear. To understand what happens along the subducting slab interface, it is important to construct models to explain the observed amplitude change as the next step.