Geodynamic inversion: Constraining rheological parameters of the lithosphere
Tobias Baumann
What is the strength of the lithosphere for a particular region of the Earth? What are the dominating deformation mechanisms, viscous creep or plastic yielding, and, more important, are these deformation processes consistent with what we observe at the surface? In this project we address these questions by combining geodynamic forward models with a Bayesian inversion approach.
Our approach is complementary to laboratory experiments, where rheology is constrained for small rock samples and extrapolated to geological conditions.
We use these laboratory results as a priori information for numerical end-member models of continental collision or subduction. In a second step, we employ these models to directly invert geophysical data such as topography, gravity or surface (GPS-) velocities and constrain rheological model parameters in a probabilistic sense.
Techniques
We closely follow Sambridge (1999a,b) in tackling the inverse problem. The approach consists of a direct search and an appraisal stage:
(1) Direct-search
We adapted the Neighbourhood algorithm by M. Sambridge (1999a) and rewrote the parallel implementation to perform an optimized direct-search with (parallel) 2D or 3D geodynamic models (Link to publication).
(2) Appraisal stage
The resulting collection of models is interpreted as an approximation of the likelihood function. The marginal posterior distribution is estimated with respect to the prior information
