A new analysis of experimental data on olivine rheology

We present a new statistical framework to analyze rock deformation data and determine a corresponding flow law and its uncertainty. All experimental uncertainties, including inter-run bias, are taken into account in the new formalism. Our approach is based on Bayesian statistics and is implemented by a Markov chain Monte Carlo method. We apply this approach to published data on the subsolidus deformation of synthetic olivine aggregates and try to establish experimental constraints on the rheology of Earth's upper mantle. Deformation data are interpreted on the basis of a composite flow law, which includes diffusion and dislocation creep mechanisms under both dry and wet conditions. We show that olivine rheology under wet conditions suffers major uncertainties suggesting the influence of poorly characterized parameters such as water content during deformation. Also, the pressure dependence of creep is still poorly constrained because of the lack of high-quality data under high pressures. However, our statistical analysis provides a solid framework to obtain a permissible range of flow laws constrained by the existing data, which will help geodynamic modeling with well-defined statistical bounds.