A Bayesian Analysis of Lithosphere-Asthenosphere Boundary Depth Constraints From Different Proxies

Surface wave dispersion data are routinely employed to model seismic velocities and anisotropy, which can be used to put constraints on the lithosphere-asthenosphere boundary (LAB). However, the choice of the proxy used to define the depth of the LAB can strongly affect the results. It can yield to inconsistent LAB depths and therefore affects our understanding of the nature of the LAB. In this study, we investigated how well constrained these different proxies are using a Bayesian approach. We applied a model space search technique to global fundamental mode surface wave phase velocity maps to determine posterior probability density functions for 3-D radial anisotropy and velocity models. This enables us to obtain quantitative uncertainties on the different proxies. In oceanic regions, we find that the LAB depth defined as the largest negative vertical gradient in shear-wave velocity increases with the age of the ocean floor, consistent with a thermal origin of the LAB. However, vertical changes in radial anisotropy yield a LAB depth that is almost flat across oceans at around 80 km. We also note that the depth of the LAB obtained from horizontally polarized shear-waves is, however, deeper by a few tens of kilometers than that resulting from vertically polarized shear-waves. These results are consistent across different oceans. Similar results have been reported in previous studies, and some authors have suggested that radial anisotropy and velocities are sensitive to different physical and/or compositional properties of the upper mantle. The Bayesian approach we adopted here will provide statistics that will enable us determine whether the observed discrepancies in LAB depth are significant. Future work will also analyze the constraints provided by higher mode surface wave data and group velocities.