Constraining anisotropy in the US continental lithosphere using a joint inversion of receiver function and ambient noise data

Multiple seismic observables are increasingly often combined to image the Earth's deep interior due to their complementary sensitivities to subsurface structure. We use a reversible jump Markov chain Monte Carlo (rjMcMC) algorithm to incorporate different seismic observables including surface wave dispersion, particle motion Ellipticity (ZH ratio), and receiver functions into a transdimensional, hierarchical Bayesian (THB) inversion for the profiles of shear velocity (Vsh and Vsv), compressional velocity (Vp), and density beneath a seismic station. In addition, we apply a Parallel Tempering sampling method to improve the sampling efficiency of rjMcMC for the joint inversion. Compared to traditional inversion approaches, Bayesian approaches yield an ensemble of models instead of a single best-fit model, enabling us to fully quantify uncertainty and trade-offs between model parameters. We perform tests on idealized data in which all three data types are analyzed individually and together. We demonstrate that by treating the number of model parameters as an unknown to be estimated from the data, we can both eliminate the need for a fixed parameterization based on prior information, and obtain better model estimates with reduced trade-offs. By analyzing the inversion results obtained using different combinations of data types, we show that while an individual data type is able to retrieve certain features of the profiles of Vp, Vs, and density, a joint inversion can leverage their complementary sensitivities to recover more accurate profiles. We then apply the THB approach to analyze the ability of surface wave dispersion constraining the 1D radial anisotropy. The ensemble results of Vp, Vsv and Vsh allow us to estimate not only the radial anisotropy, but also the trade-off between the radial anisotropy and Vp. We also use the actual phase velocity measurement from USArray ambient tomography to invert for 1D velocity structure and radial anisotropy. We analyze the distributions of the retrieved velocity profiles with depth, and compare our inferred results with those from other studies.