Inferring the thermochemical structure of the upper mantle from seismic data

Abstract

SUMMARY We test a mineral physics model of the upper mantle against seismic observations. The model is based on current knowledge of material properties at high temperatures and pressures. In particular, elastic properties are computed with a recent self-consistent thermodynamic model, based on a six oxides (NCFMAS) system. We focus on average structure between 250 and 800 km. We invert normal modes eigenfrequencies and traveltimes to obtain best-fitting average thermal structures for various compositional profiles. The thermochemical structures are then used to predict long-period waveforms, SS precursors waveforms and radial profiles of attenuation. These examples show the potential of our procedure to refine the interpretation combining different data sets. We found that a mixture of MORB and Harzburgite, with the MORB component increasing with depth, is able to reproduce well all the seismic data for realistic thermal structures. If the proportions of MORB with depth do not change, unrealistic negative thermal gradients below 250 km would be necessary to explain the data. Equilibrium assemblages, such as pyrolite, cannot fit the seismic data. The elastic velocities predicted by the reference mineral physics model tested are too low at the top of the lower mantle, even for the fastest (and most depleted) composition, that is, harzburgite. An increase in V P of 1 per cent and in VS of 2 per cent improves the data fit significantly and is required to find models that fit both traveltimes and normal modes, indicating the need for further experimental measurements of these properties at the simultaneously elevated pressure–temperature conditions of the lower mantle. Extending our procedure to other seismic and density data and interpreting the 3-D structure holds promise to further improve our knowledge of the thermochemical structure of the upper mantle. In addition, the same database of material properties can be used in dynamic models to test whether the thermochemical structure inferred from geophysical observations is consistent with the Earth’s evolution.