Reconcile Mantle Dynamic Models with Compositionally Distinct and Stable LLSVPs with the Observations of the Geoid and Dynamic Topography

Abstract:

The geoid has been well explained in mantle flow models with the buoyancy inferred from seismic models that in turn place constraints on mantle viscosity structure (e.g., Hager & Richards, 1989). These models often assume a whole-mantle convection with uniform composition and 1-D viscosity. However, seismic and geochemical observations suggest possible existence of chemically distinct piles under Africa and Pacific which extends hundreds of kilometers above the CMB (i.e., LLSVPs). As compositional heterogeneity would significantly alter the interpretation of seismic anomalies as buoyancy structure, important questions are whether a thermochemical mantle model based on seismic velocity anomalies can reconcile the geoid and how this may impact inference of mantle viscosity structure.

In this study, we formulate mantle flow models that use buoyancy derived from seismic model S40RTS (Ritsema et al., 2011), assuming that the LLSVPs are stable with negative buoyancy. The models use temperature-, depth- and composition-dependent viscosity and are computed for the geoid, dynamic topography and flow velocity using CitcomS. Seismic anomalies are converted to buoyancy using thermal conversion factor c_T for the whole mantle materials and composition conversion factor c_c for the chemical piles defined as the domains with seismic slow anomaly <-0.5% and a maximum height of 500 km. The temperature-dependence viscosity gives rise to 3 orders of magnitude variations in viscosity, and horizontally averaged viscosity profile is consistent with the inferred 1-D viscosity from the geoid. The viscosity in the chemical piles is further reduced by a factor of Cvisc to represent the compositional effect. We measure the stability of the chemical piles by the RMS vertical velocities on the piles boundary. Our preferred thermochemical models with stable chemical piles reach similar variance reduction of geoid at ~64% to that for the uniform composition models. In the preferred model, c_T is ~0.25 which is similar to that inverted from whole mantle models, c_C is about -0.7, and Cvisc=30. Compared with models with 1-D viscosity profile, the temperature- and composition-dependence of viscosity plays a key role in enhancing the stability of chemical piles and fits to the observed geoid and surface and CMB dynamic topography.