A Reduced Order Approach for Joint Probabilistic Inversions of 3D Magnetotelluric and Surface Wave Data

Abstract:

Multi-observable probabilistic inversions are gaining popularity for imaging the Earth's interior and elucidating the physicochemical structure of the lithosphere [1]. Of particular interest is the joint inversion of magnetotelluric (MT) with seismic data as they are inherently sensitive to different physical properties, viz. electrical conductivity and seismic velocity and they, therefore, provide complementary information on the thermal structure, fluid pathways and water content. As both data sets can strongly constrain the first-order thermal structure of the lithosphere, this background effect can be ‘filtered out’ from the MT data to isolate the contribution of anomalous features such as fluids and melt content. Information about these anomalies is critical for understanding the complex fluid-rock interactions responsible for mineralization events and water-assisted tectonism.

Joint probabilistic inversions of MT and seismic data have been successfully implemented in the context of 1D MT data only. In the case of 2D and 3D MT data, however, joint probabilistic approaches have, up until now, been impractical due to the large computational cost of the full MT forward solutions. We have recently presented a novel strategy [2], called RB+MCMC, that combines i) an efficient parallel-in-parallel structure to solve the 3D forward problem, ii) a Reduced Basis Method to create fast and accurate surrogate models of the forward problem, and iii) adaptive strategies for both the MCMC algorithm and the surrogate model. This strategy reduces the computational cost of the 3D MT forward solver and makes it possible to perform full probabilistic 3D MT inversions.

In this contribution, we adopt the RB+MCMC approach to include MT data into joint probabilistic inversions for the 3D imaging of deep thermochemical anomalies and fluid pathways and present the first joint probabilistic inversion of 3D MT and surface waves. These results illustrate the capabilities of our conceptual and numerical framework for 3D joint probabilistic inversions of MT with other geophysical data sets and open up exciting opportunities for elucidating the Earth’s interior.

[1] Afonso, J.C. et al., (2016), J. Geophys. Res., 121, doi:10.1002/2016JB013049

[2] Manassero, M.C et al., (in review), doi:10.31223/osf.io/7n5mv