Towards an Improved Joint Inversion of Receiver Functions, SKS Splitting, and Surface Wave Dispersion Data for Layering in the North American Craton

Details of the formation and evolution of continents, especially in cratonic regions, remains poorly understood. Structural layering in the cratonic lithosphere is of particular importance due to competing hypotheses of craton formation, which include underplating by hot plumes or accretion by shallow subduction in continental or arc settings. One source of evidence that provides constraints on these hypotheses is seismic anisotropy, which is generally attributed to past and present rock deformation in the upper mantle. Anisotropic structure can be constrained from surface wave dispersion (SWD), which is sensitive to the uppermost portion of the mantle, and core refracted shear wave (SKS) splitting measurements, which add constraints on the integrated effect across the entire upper mantle.

Another source of evidence involves receiver function (RF) analysis which can detect sharp changes in seismic velocities. Fine scale layering in the lithosphere has recently become a topic of interest due to the detection of a sharp velocity reduction at 80-120 km depth across cratonic provinces in North America (NA). This has sparked debate on whether RF studies are in fact detecting the lithosphere-asthenosphere boundary or a mid-lithospheric discontinuity within the older cratonic provinces. The NA craton is of interest due to its rich tectonic history as well as data accumulation at broadband seismic networks including USArray, which provides a dense network coverage that allows for high resolution seismic velocity and structure models in the upper mantle.

We present a transdimensional Monte Carlo Markov chain inversion, in which the number of isotropic and anisotropic layers are considered to be unknowns, allowing the algorithm to have a flexible parameterization. The method performs a joint inversion that combines observations from SKS splitting, RF and anisotropic SWD in a consistent manner. The solution is a probabilistic shear velocity (Vs) model that accounts for azimuthal anisotropy. One advantage of this method is that it combines datasets of varying levels of noise and depth sensitivities, without the need to balance the contribution of each dataset in the inversion. Synthetic and real tests are done to resolve structure down to a depth of 350 km beneath seismic stations in North America.