Crustal Imaging With Bayesian Inversion of Teleseismic P Wave Coda Autocorrelation

Abstract

The autocorrelation of the seismic transmission response of a layered medium (autocorrelogram), in the presence of a free surface, corresponds to the reflection response. Despite many studies on the imaging of local structures through retrieval and forward modeling of stacked autocorrelograms, there is limited work on the inversion of these data. In this study, we demonstrate that the probabilistic inversion of autocorrelograms is efficient and can be used as an alternative imaging tool when other approaches are not applicable. Here, we calculate autocorrelograms of teleseismic P wave coda recorded on more than 1,200 permanent and temporary seismic stations across Australia and utilize a Bayesian framework to invert these data for crustal imaging. The results show patterns of structures consistent with those seen in previous crustal models constructed from receiver function, seismic reflection, and refraction methods. The new approach can therefore image large‐scale crustal structures comparable to those from other seismological methods.

New Insights Into Crustal Properties of Anatolia and Its Surroundings Inferred From P‐Coda Autocorrelation Inversions

“…To suppress the sidelobe effects that stems from the zero‐lag autocorrelation peak on the inversion results, high amplitudes of the autocorrelation's side lobes for both data and predicted autocorrelations are damped by using an exponential weighting function given in Tork Qashqai et al. (2019).…”

“…Inverse modeling of P‐coda autocorrelograms within a Bayesian framework was first presented in a case study performed by Tork Qashqai et al. (2019) that aimed at crustal‐scale imaging of Vp and Moho structures across Australia. In their study, the authors validated the potential usage of this novel approach for crustal imaging by revealing very consistent crustal features compared with the Australian Seismological Reference Model (AuSREM; Salmon et al., 2013).…”

Areal parameter estimates from multiple datasets

“…The physical model used to extract these results enforces a sharp discontinuity (Qashqai et al, 2019), and the differences from the interpreted surface, where present, provide an indication of the thickness of the crust-mantle transition.…”

“…Since 2011 the number of Moho estimates in Australia has grown significantly, with a strong continuing program of full-crustal reflection profiling and the advent of new methods such as the exploitation of stacked station autocorrelograms Qashqai et al, 2019). In many places this means that there are multiple values for the depth to Moho based on different physical assumptions.…”

Probabilistic Cover‐Basement Interface Map in Cloncurry, Australia, Using Magnetotelluric Soundings

“…But thorough model space search requires a significant amount of computing resources, which depends on the computation time of the forward solver, which increases with the complexity of the physics involved and the number of model and data parameters. Probabilistic inversion of geophysical data is receiving increasing attention, for applications to the shallow subsurface and lithosphere using electromagnetics (Ardid et al, 2021;Blatter et al, 2019;Brodie & Sambridge, 2006;Hauser et al, 2015;Manassero et al, 2020;Minsley, 2011), at crustal scale using seismic methods (Bodin et al, 2012;Guo et al, 2020;Tork Qashqai et al, 2019). Chen et al (2012) applied a sharp boundary parametrization to a 2D probabilistic inversion of MT data to recover depth to interfaces and associated uncertainty.…”