Skip to Main content Skip to Navigation

Geodynamic Tomography

Abstract : Over the last three decades, seismologists have been successful in probing the Earth's internal structure using copious seismic records gathered at its very surface. Widely known as seismic tomography, one of its central goals is to construct three-dimensional elastic models of the Earth using various inversion strategies tailored to the type of seismic data used. A remaining challenge however is the interpretation of the inherent heterogeneities in its elastic structure in terms of several physical properties (e.g., density, chemical/mineralogical composition, temperature) and of the ubiquity of large-scale anisotropy that are crucial to understand plate tectonics and mantle dynamics. For instance, long-wavelength seismic anisotropy observed in tomographic models have often thought to have been caused by crystallographic preferred orientation (CPO) (i.e., the net alignment of intrinsically-anisotropic upper-mantle minerals) due to mantle deformation. Seismic tomography thus provides a great deal of information about the present-day flow in the mantle. We introduce Geodynamic Tomography, a novel approach to the tomographic problem that incorporates geodynamical and petrological constraints to reduce the number of Earth models down to a subset consistent with geodynamic predictions. This approach encompasses several methodologies: mantle flow modeling, texture evolution modeling, thermodynamic modeling, and seismic data inversion into a single self-consistent probabilistic inversion procedure. Ultimately, we aim to retrieve the complete pattern of upper-mantle deformation by inverting seismic data to better understand mantle dynamics from a seismological point of view. We focus on the inversion of surface wave measurements since they provide unique constraints to the large-scale anisotropy associated with convective flow in the mantle. Geodynamic tomography addresses one of the most pressing issues of conventional anisotropic surface wave tomography, that is, its inability to resolve the 21-component elastic tensor independently at every location. We formulate the tomographic problem by using geodynamical and petrological constraints to reduce this large number of model parameters. The full forward problem proceeds as follows: (1) calculation of instantaneous mantle flow using the temperature field, and the viscosity field as inputs, (2) from the flow field and deformation gradient of (1) as inputs, calculation of the induced CPO and seismic anisotropy using a micro-mechanical model for texture evolution, (3) modeling the pressure- and temperature-dependence of isotropic properties for a given bulk composition using a thermodynamic model, (4) construction of the full elastic tensor given (2) and (3), and finally estimation of azimuthally-varying surface wave dispersion curves. The non-linearity of the forward problem and the non-uniqueness of the solution warrant the need to cast the inverse problem in a Bayesian probabilistic framework. The formalism is an ensemble inference approach, where the solution is an ensemble of models representing a posterior probability distribution, accompanied by the uncertainties in each model parameter. In fact, any implicitly-computed variable (e.g. deformation and anisotropy) in the forward problem can be recast in terms of a posterior distribution in their respective model space. We efficiently explore the space of candidate Earth models (e.g., temperature field) by employing a Markov chain Monte Carlo (McMC) algorithm. The implementation of the method deemed to be a success as our results exemplified the implicit retrieval of the complete patterns of mantle deformation, and correspondingly, the 21-independent coefficients of the elastic tensor from the inversion of seismic data alone. Geodynamic tomography is therefore a potentially powerful technique to elucidate the structure of the upper mantle, and interpret seismic observations in terms of mantle deformation patterns
Document type :
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Tuesday, October 19, 2021 - 10:56:10 AM
Last modification on : Thursday, October 21, 2021 - 5:06:48 AM
Long-term archiving on: : Thursday, January 20, 2022 - 6:39:25 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03384871, version 1


John Keith Magali. Geodynamic Tomography. Earth Sciences. Université de Lyon, 2021. English. ⟨NNT : 2021LYSE1050⟩. ⟨tel-03384871⟩



Record views


Files downloads