Estimating Moho depth utilizing S-wave receiver functions
Abstract:H-k stacking method [Zhu and Kanamori, 2000] is a widely used grid search technique for estimating the Moho depth (H) and Vp/Vs (k) beneath a given station. The H-k surface reaches a maximum when the optimum H and k values are used, which is assumed to be the average crustal structure beneath the seismic station. In general, the method is employed in conjunction with P-wave receiver functions. Here, we investigate the usability of H-k stacking method with S-to-P (Sp) conversions and S-wave reverberations within the crust, employing an extended multi-taper deconvolution. We apply the method to southern California, using data recorded between 1990-2011. We compare results with those of prior studies that used P-to-S (Ps) conversions [Zhu and Kanamori, 2000; Yan and Clayton, 2007], applying a smoothing length of 0.5 degrees to reflect lateral Sp sensitivity.
P-waves have better potential to resolve lateral variations in Moho depth owing to the higher frequency content and the geometry of Ps ray path. Our results from Sp conversions are in broad agreement with those from Ps, affirming that S-wave receiver functions can be used in conjunction with the H-k stacking method. Consistent with the P-wave receiver function results, crust is thinner beneath the central Transverse Range (30 km) with respect to eastern Transverse Range (33 km) and Peninsular Region (35 km). Our Moho depth observations (35 km) are more compatible with those of Yan and Clayton  (~35 km) than Zhu and Kanamori  (~30 km) beneath Sierra Nevada, most probably due to a larger data set this study and Yan and Clayton  use. Also, results from this study are deeper than those from Ps for the Salton Trough (30-35 km vs. 25 km). In this case, broad receiver function waveform characteristics suggest a more gradual impedance change across the Moho discontinuity and/or a multi-layered crust. We suggest that a combination of P- and S-wave receiver functions can yield more robust crustal thickness estimates, providing additional constraints in areas where data coverage is not sufficient, Ps results become complicated due to multiple crustal layers, or velocity gradients are gradual.