T53E-02:
Circum-Slab Mantle Deformation: Insights from Finite Strain and Seismic Anisotropy
Friday, 19 December 2014: 1:55 PM
Jeanette F Di Leo1, Zhong-Hai Li2, Neil M Ribe3, Andrew Walker1,4, James M Wookey5 and J Michael Kendall5, (1)University of Bristol, Bristol, United Kingdom, (2)Chinese Academy Geological Sciences, Beijing, China, (3)University of Paris-Sud 11, Orsay, France, (4)University of Leeds, School of Earth and Environment, Leeds, United Kingdom, (5)University of Bristol, Bristol, BS8, United Kingdom
Abstract:
Recent numerical modeling studies of the time-dependent development of texture and seismic anisotropy during subduction have shed light on how the mantle deforms as a slab subducts. It is thus becoming more and more clear that the term “mantle flow” may be too ambiguous in the context of subduction. For instance, it has been suggested that trench-parallel shear wave splitting fast directions (from SKS and source-side S splitting) on the seaward side need not necessarily be the result of trench-parallel movement (i.e., “flow”) of mantle material, but are rather due to pure shear deformation in the sub-slab mantle. Here we present results of a numerical modeling study where we have systematically investigated how mantle propagation, finite strain, olivine lattice-preferred orientation (LPO), and SKS splitting vary with slab width in a fully dynamic 3-D subduction model. We find that even in the complex circum-slab flow field, the finite strain ellipsoid (FSE) is a good proxy for LPO. However, it does not necessarily align with the instantaneous mantle flow velocity vector. We identify two distinct domains with different deformation types in the central sub-slab upper mantle: simple shear induced by plate advance dominates at shallow depths and results in trench-normal fast splitting, while pure shear due to slab rollback dominates in the deeper mantle (above 410 km) and produces trench-parallel fast orientations. In our models, the SKS splitting pattern strongly depends on these two competing effects as well as the subduction partition ratio, γ = Xp/Xt, where Xp and Xt are the lengths of plate advance and trench retreat, respectively. If γ < 1 (narrow slabs, < 1000 km), trench-parallel fast directions are produced. In contrast, γ > 1 (wide slabs, > 1000 km), results in trench-normal fast splitting. This may explain the observed dichotomy in natural subduction zones of sub-slab fast splitting patterns (away from slab edges) usually being either trench-parallel or –perpendicular.
