Anomalies of temperature and iron in the uppermost mantle inferred from gravity data and tomographic models

We propose a method to interpret seismic tomography in terms of thermal and compositional anomalies. In addition to the tomographic model, we use gravity data, which provide information on the density expressed as a relative density-to-shear wave velocity scaling factor (ζ=∂ ln ρ/∂ ln Vs). The inferred values of ζ are not consistent with the presence of thermal anomalies alone. However, simultaneous anomalies of temperature and composition explain the observations. Compositional anomalies can have several origins, but we find the most relevant parameter to be the global volumic fraction of iron (xFe=Fe/(Fe+Mg)). We invert the tomographic model S16RLBM (Woodhouse and Trampert, 1995) and the density anomalies correlated to Vs-anomalies (δρ/ρ0=ζδ Vs/V0) for anomalies of temperature (δT) and iron (δFe). The partial derivatives are provided by a numerical method that reconstructs density and seismic velocity for given temperatures and petrologic models (Vacher et al., 1998). Down to z=300 km depth, the distribution of temperature and iron anomalies strongly depends on the surface tectonics. The continental mantle below old cratons and stable platforms is colder than average and depleted in iron, whereas the oceanic mantle is mostly homogeneous. Due to uncertainties on the reference state of the mantle, error bars on δT and δFe reach 10% of the inverted values. Finally, we apply these results to the stability of continental roots and test the hypothesis that the negative buoyancy induced by lower than average temperatures is balanced by the positive buoyancy induced by the depletion in iron. We find that continental roots are stable only if the viscosity of the mantle is strongly temperature-dependent. However, some uncertainties remain on the real effects and importance of rheology.