Global Adjoint Reconstructions of Earth’s Mantle

Mantle convection is the driving mechanism for plate tectonics and associated geological activities, including earthquakes, surface dynamic uplift and subsidence, and volcanoes. Mantle convection can be regarded as the central framework for linking the sub-disciplines of solid Earth science, e.g., geochemistry, seismology, mineral physics, geodesy and geology.

In theory, it is possible to model mantle convection by integrating the principial conservation equations in time, given a past mantle-state as the starting point. Nonetheless, there remains a fundamental lack of knowledge on any past mantle-states. Without such knowledge any direct comparison of convection models and solid Earth observations is challenging and often impractical. One can, however, pose the problem differently, and obtain a past flow history by minimising ‘a misfit’ functional between observations and models of Earth’s mantle. The recent applications of adjoint method in geodynamics, together with the ever-increasing computational power, has facilitated solutions to such minimisation problems, where a unique flow history in Earth’s mantle can be generated, subject to assumed geodynamic modelling parameters.

Here, we build on previously published adjoint models and present a suite of eight high resolution (11 kms) reconstruction models going back to 50 Ma ago. These models incorporate many improvements. First, we take advantage of the recent advances in surface and body waveform tomography to obtain high resolution images of present-day structures in Earth’s mantle. Our thermodynamic modelling of mantle structures rely on the most recent datasets of mantle mineralogy and account for effects of anelasticity. Furthermore, we assume a wide range of viscosity profiles, including published models consistent with observations of geoid, mantle mineralogy, and post-glacial rebound studies. Finally, we verify these models by comparisons against a range of different geologic observations.