Brown Bag: Chhavi Jain
Abstract
The upper mantle rheology is usually modeled using the flow-law parameters for olivine. Olivine likely deforms under diffusion or dislocation creep under hot upper mantle conditions, and the flow-law parameters for these mechanisms are determined by the experimental data on the deformation of olivine aggregates. We recently reanalyzed experimental data on olivine to obtain more accurate estimates on these parameters [Jain et al., 2019]. In our analysis, experimental data were modeled with a composite flow law, including the parallel operation of diffusion and dislocation creep. The nonlinear inversion was conducted using a Markov chain Monte Carlo (MCMC) inversion scheme, with the statistics of flow-law parameters fully quantified including parameter covariance. We found that some flow-law parameters were only poorly constrained due to large data uncertainties and/or insufficient data. Here we propose a new way to further constrain the flow-law parameters, by conducting `probabilistic’ geodynamic modeling. Our approach also presents a means to connect experimental rock mechanics, theoretical mantle dynamics, and observational geophysics in a statistically satisfactory framework. As an example, we use steady-state plate-driven mantle flow beneath a mid-ocean ridge. First, from our MCMC inversion results, we generate an ensemble of permissible flow laws that are consistent with experimental data. Different flow laws result in different rheological predictions, such as the extent of dislocation creep dominated regions in the upper mantle, and such difference is further reflected in geophysical observables, such as the development of lattice-preferred orientation. In this study, we use the D-Rex code [Kaminski et al., 2004] to calculate the resultant radial anisotropy for the suboceanic upper mantle. By comparing these predictions with the radial anisotropy measured using surface wave velocities, we can identify the subset of flow laws that can reproduce the observed radial anisotropy profile and, thereby, tighten our constraints on each flow-law parameter.