Bayesian Uncertainty Quantification of Subduction Zone Rheology from the Geoid

Observations of the Earths geoid near subduction zones have the potential to provide insight into the structure and rheology of the subducting plate and surrounding mantle. We generate synthetic geoid measurements from an instantaneous 2D model of intra-oceanic subduction using ASPECT. This model extends to the core-mantle boundary, includes phase transitions at 410 km and 660 km, a compositionally- and rheologically-distinct crust, and depth-, temperature-, and stress-dependent nonlinear rheology for the mantle, overriding plate, and subducting plate. We use a hierarchical Bayesian inversion procedure to calculate the posterior probability density functions for rheological parameters (viscosity, stress exponent, temperature-dependence) constrained by the synthetic observations plus varying amounts of noise. From the ensemble solutions, we assess tradeoffs between parameters and non-uniqueness. We discuss the implications of these inversions of synthetic geoid measurements for the resolution of rheological properties of the slab and mantle in natural settings.