New Inferences of Mantle Composition in Terms of Three End-member Mixing via Markov Chain Monte Carlo Inversion of the GyPSuM Tomography Model

Advances in global seismic mantle tomography over the past two decades reflect improved methods and increasing data coverage. However, interpreting these velocity distributions as thermochemical distributions remains challenging, and is the focus of our work here. Several groups [e.g., Forte & Mitrovica 2001, Cobden et al. 2009] have calculated the sensitivity of density and seismic velocities to temperature and compositional variations to estimate thermochemical heterogeneity in the lower mantle. Here we employ a Markov chain Monte Carlo (MCMC) approach to derive the best fitting major oxide bulk composition from a mixture of three end-members: primitive mantle, depleted mantle, and the lower mantle component of a fractionated early Earth magma ocean. Each end-member composition was defined in terms of five major oxides: SiO₂, MgO, FeO, Al₂O₃, and CaO. Physical material properties were calculated using the mineral physics software package PerpleX [Connolly 1990, 2009] for a range of mantle temperatures (T) and pressures (P). Proportions of end-members in the model compositions were inferred by fitting density and seismic velocities at a given T-P. Future work will consider other end-member compositions (e.g. pyroxenite). We applied this new approach to the 3-D GyPSuM tomography model [Simmons et al. 2010] to infer global mantle heterogeneity in terms of large-scale temperature and major element variations. GyPSuM is ideally suited to this task, because it models anomalies in P and S seismic velocities and mantle density, via joint seismic-geodynamic inversions. The importance of this analysis is underscored by the ongoing debate concerning the interpretation of the Large Low Shear Velocity Provinces (LLSVP) and their dynamic relationship to the convecting mantle, as well as the possible existence of isolated chemical reservoirs. In addition to gaining insight into major-element chemical heterogeneity, this work provides a basis for reinterpreting lateral viscosity variations in the mantle [e.g. Glisovic et al. 2015]. We anticipate that these results will shed a new light on the importance of lateral heterogeneity in composition and rheology for the thermal, chemical, and dynamic evolution of the mantle.