Towards multi-observable thermochemical tomography of the lithosphere and sublithospheric upper mantle

Parametric tomography (i.e. focused on physical parameters such as wave velocity or electrical conductivity) is among the most common and powerful methods to study the internal structure of the Earth’s mantle. However, recent advancements in i) the acquisition and modelling of mineral physics data, ii) non-equilibrium and equilibrium thermodynamic formalisms, and iii) stochastic methods applied to inversion of multiple datasets, offer new possibilities for inverting geophysical data directly for the thermochemical structure of the Earth (i.e. thermochemical tomography). In this contribution we present results from a new 3D multi-observable non-linear inversion method based on a probabilistic (Bayesian) formalism. The method jointly inverts the following datasets: Rayleigh and Love wave dispersion curves, teleseismic arrival time residuals, gravity and geoid anomalies, surface heat flow, apparent resistivity and phase (magnetotelluric), and absolute elevation. We show that temperature anomalies of > 100-150 oC and large compositional anomalies (i.e. ΔMg# > 3 or bulk ΔAl2O3 > 1-1.5) can be expected to be (probabilistically) resolved simultaneously when high-quality geophysical data are available. Although the mapping of smaller anomalies seems outside the capabilities of any current method, the above resolution is sufficient to provide answers to some long-standing problems regarding the nature and evolution of the lithosphere (e.g. vertical stratification of cratonic mantle, compositional vs temperature signatures in seismic velocities, etc).