[S-SS11] Crustal Structure

[SSS11-02] Beyond Receiver Functions: Crustal Properties from Bayesian Joint Inversion of Vertical and Radial Teleseismic P-wave Coda Autocorrelations

*Mehdi Tork Qashqai¹, Erdinc Saygin^2,1 (1.Deep Earth Imaging Future Science Platform, CSIRO, Perth, WA, Australia, 2.Department of Physics, School of Physics, Mathematics & Computing, Faculty of Engineering and Mathematical Sciences, University of Western Australia, Perth, WA, Australia)

Keywords:Teleseismic P-wave Coda Autocorrelation, Receiver functions, Crustal structure, Probabilistic joint inversion

P-to-S receiver functions (RFs) are a class of signals that contain converted P to S-waves at seismic discontinuities of the subsurface. They have been widely used to obtain receiver-side Green's functions, where they are primarily sensitive to the shear-wave velocity (S-wave) at seismic discontinuities as the reflected and transmitted P-waves are cancelled by the deconvolution process during the calculation of the RFs. In addition to the RFs, the autocorrelation of teleseismic P-wave coda recorded at vertical and radial components of the seismic sensors include P-wave reflection responses as well as phases associated with P-to-S-wave conversions (as in RFs). Therefore, compared to the classical RFs, the autocorrelations of teleseismic waveforms recorded on both vertical and radial components contain more information about the receiver-side crustal properties. Thus, joint inversion of autocorrelations of vertical and radial components alone or with RFs offers a framework which has a great potential to provide more robust estimates of the medium properties (V_p, V_s and V_p/V_s) and their associated uncertainties. There have been many studies about the inversion of RFs, but, to our best knowledge, there are no published studies on the joint inversion of both radial and vertical components autocorrelations (or with RFs). In this study, we present a probabilistic joint inversion approach to better constrain both the receiver-side V_p and V_s structures simultaneously, and consequently V_p/V_s crustal structure, which is commonly used to make inferences about the composition of the rocks. We first show the feasibility of using this approach by comparing to single component inversions through a series of synthetic inversion tests. The synthetic tests show that probabilistic joint inversion of autocorrelations of both radial and vertical components provides a robust estimate of the crustal properties and the inclusion of the receiver function into this framework does not provide a better constraint. We then apply the approach on real field passive seismic data, recorded on a series of broadband seismic sensors deployed across a north-south profile in central Australia.