Deconvolution of Teleseismic P-Waves Using the SV Autocorrelation Method with Application to the P-Wave Structure Beneath the Hi-CLIMB Array in Tibet

The analysis of seismic receiver functions has become an effective approach for determining crust and upper mantle structure. In the traditional receiver function method, the vertical component is used to deconvolve the radial component, where the vertical component is assumed to be equivalent to the source wavelet. However, in the approach of Dasgupta and Nowack (2006), the deconvolution of three-component teleseismic P-waves is performed by using the autocorrelation of P to SV scattered waves (SVA method). In this approach, three-component seismic data (in the vertical-radial-transverse frame) is transformed into P- SV-SH frame (Kennett, 1991) with the effects of the free surface also taken into account. The P to SV scattering is assumed to be random and white so that the autocorrelation of the SV component is equivalent to the autocorrelation of the source wavelet. This is similar to the assumption used for predictive deconvolution using P to P scattering in exploration geophysics. A minimum-phase source wavelet is estimated from the autocorrelation of the SV component and this wavelet can be used to deconvolve the teleseismic P-wave (unrotated radial and vertical components). However, the radial and vertical components must be transformed to minimum phase before the deconvolution. This can be done since under conditions typical of a teleseismic incident wave, the P-wave impulse response of receiver-side stratification will be minimum phase (Bostock, 2004). In this way we can get both deconvolved P to P and P to SV scattered components. The approach is first tested with synthetic data in which random velocity fluctuations are added to a 1-D deterministic crust and upper mantle structure. The approach is then applied to deconvolve teleseismic P-wave components (radial and vertical) for data from stations of the Hi-CLIMB seismic array which were deployed across the Himalayan-Tibetan collision zone. The SV-autocorrelation method is used for different earthquake events for each station and then the results of a group of events are stacked. The Moho depths obtained for the PpPmp phase on the deconvolved vertical component are then compared to those obtained from the Ps phases for selected stations beneath the Hi-CLIMB array.