Receiver function images beneath Kii Peninsula, southwest Japan with an improved procedure
Abstract
Teleseismic P waves can be theoretically represented by convolving the impulse response of recording instruments, seismic source functions and the impulse response of earth structures. And the convolution of the instrument response and the source time function can be approximated by the vertical component waveform because an incidence angle of the teleseismic P wave is small (Langston, 1979). Hence, receiver functions (RFs) are conventionally calculated by deconvolving the vertical component from the radial and transverse components of the P wave part of teleseismic events. However, noises uncorrelated among the components cause the deconvolution unstable (Langston and Hammer, 2000). Therefore, it is necessary to reduce the noises in the vertical component to obtain accurate RFs. In addition, a radial RF includes PS converted waves at S wave velocity discontinuities and scattered waves caused by heterogeneous structures as well as a direct P wave which does not include structure information. The direct Pp phase in the radial RF may mask smaller Ps phases generated at shallower depths or prevent observed and synthetic waveforms from fitting smaller phases in RF waveform inversions. In this study, we tried to estimate better source time functions and remove the direct Pp phase from RFs according to the procedure of Bostock and Rondenay (1999). We used wave data from linear seismic arrays in Kii Peninsula (Shibutani et al, 2009). In the RFs obtained by this method, the direct Pp phase with a large amplitude is removed and PS converted waves near t = 0 s clearly appear. The pre-signal noise is also reduced. For the part after the direct P waves, the RFs obtained by this method are mostly consistent with the corresponding RFs by the conventional method although some phases are not fitted in their amplitude and phase. For an example, some peaks between 2 s and 10 s become sharper in the revised RFs than in the conventional RFs. The conventional RFs are probably more influenced by the noises in the vertical component. By the stabilization effects in this method, we can detect converted waves which were difficult to detect in the conventional method and emphasize them. Then, we migrated the improved RFs with a realistic velocity model which included dipping interfaces corresponding the subducting Philippine Sea slab. The migration technique is newly developed by Abe et al. (submitted to GJI). In the resulting RF images, we can clearly find a pair of interfaces which correspond to the upper surface of the low velocity oceanic crust and that of the high velocity oceanic mantle (the slab Moho). In this presentation, we will show revised RF images along three profile lines in the Kii Peninsula with comparison to the previous images and discuss improved features. We will also discuss structure around the subducting Philippine Sea slab based on the RF images.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2010
- Bibcode:
- 2010AGUFMDI31A1950N
- Keywords:
-
- 0900 EXPLORATION GEOPHYSICS;
- 7240 SEISMOLOGY / Subduction zones