Pressure-dependent large-scale seismic anisotropy induced by non-Newtonian mantle flow

Abstract

Summary Observations of large-scale seismic anisotropy can be used as a marker for past and current deformation in the Earth’s mantle. Nonetheless, global features such as the decrease of the strength of anisotropy between ∼150–410 km in the upper mantle and weaker anisotropy observations in the transition zone remain ill-understood. Here, we report a proof of concept method that can help understand anisotropy observations by integrating pressure-dependent microscopic flow properties in mantle minerals particularly olivine and wadsleyite into geodynamic simulations. The model is built against a plate-driven semi-analytical corner flow solution underneath the oceanic plate in a subduction setting spanning down to 660 km depth with a non-Newtonian n = 3 rheology. We then compute the crystallographic preferred orientation (CPO) of olivine aggregates in the upper mantle (UM), and wadsleyite aggregates in the upper transition zone (UTZ) using a viscoplastic self-consistent (VPSC) method, with the lower transition zone (LTZ, below 520 km) assumed isotropic. Finally, we apply a tomographic filter that accounts for finite-frequency seismic data using a fast-Fourier homogenization algorithm, with the aim of providing mantle models comparable with seismic tomography observations. Our results show that anisotropy observations in the UM can be well understood by introducing gradual shifts in strain accommodation mechanism with increasing depths induced by a pressure-dependent plasticity model in olivine, in contrast with simple A-type olivine fabric that fails to reproduce the decrease in anisotropy strength observed in the UM. Across the UTZ, recent mineral physics studies highlight the strong effect of water content on both wadsleyite plastic and elastic properties. Both dry and hydrous wadsleyite models predict reasonably low anisotropy in the UTZ, in agreement with observations, with a slightly better match for the dry wadsleyite models. Our calculations show that, despite the relatively primitive geodynamic setup, models of plate-driven corner flows can be sufficient in explaining first-order observations of mantle seismic anisotropy. This requires, however, incorporating the effect of pressure on mineralogy and mineral plasticity models.