New Inferences of Earth’s Mantle Viscosity Structure and Implications for Long-wavelength Structure in the Lower Mantle

Abstract:

The viscosity structure of Earth's deep mantle affects the thermal evolution of Earth, the ascent of mantle plumes, settling of subducted oceanic lithosphere, and the mixing of compositional heterogeneities in the mantle. Modeling the long wavelength non-hydrostatic geoid provides a constraint on the radial viscosity structure of Earth’s mantle. We carried out inversions for the radial mantle viscosity structure using a transdimensional, hierarchical Bayesian technique that allows us to obtain solutions without specifying at the outset the number or locations of viscosity changes within the mantle. We obtained a posterior probability distribution of mantle viscosity structures, which allowed us to assess our confidence in our inferences of the viscosity structure. We find robust evidence for an increase in viscosity at 800-1200 km depth, significantly deeper than the mineral phase transformations which define the mantle transition zone. The viscosity increase is coincident in depth with regions where tomographic models image slab stagnation, plume deflection, and changes in large-scale structure, manifested in the mantle radial correlation function for the lowest spherical harmonic degrees. Here, we present new results from 3D, spherical-shell geometry thermal and thermochemical mantle convection simulations with prescribed plate motions based on paleogeographic reconstructions. These simulations employ a range of admissible mantle viscosity structures from our geoid inversions. We find that by including the inferred increase in viscosity at 1000 km depth, we can better reproduce the long wavelength mantle radial correlation function observed in the latest tomographic models GAP-P4 and SEMUCB-WM1. The similarity of the modeled and observed radial correlation functions is sensitive to the choice of lower mantle viscosity and the inclusion of phase changes in the transition zone and the mid-mantle. We will also discuss the effect of these viscosity structures on predictions of evolving long-wavelength thermochemical structures as well as the core-mantle boundary heat flow.