Identifying the Mantle Wedge Signature in Shear-wave Splitting Parameters in Oblique Subduction Zones

This study examines the local S-wave splitting in the creeping part of the mantle wedge in generic subduction systems with varying amounts of subduction obliquity. Previous studies using numerical models have shown that mantle wedge flow above an obliquely subducting slab is complex as a result of coupling between the subducting slab and the overriding mantle [e.g., Kneller and van Keken, 2008; Wada et al., 2015]. We develop a series of 3-D coupled kinematic-dynamic models of generic subduction systems with obliquity ranging from 30° to 60°. We calculate the olivine crystal preferred orientation for various olivine fabric types using DRex [Kaminski et al., 2004]. The calculated flow velocities and the average fast axis align with the inflow but not in the outflow or where the flow transitions from inflow to outflow. Using tensors from DRex and the MSAT toolbox [Walker and Wookey 2012], we calculate the expected splitting parameters for a full range of initial polarizations and an incidence angle of zero. The fast direction with the maximum delay time (MDT) is approximately perpendicular to the margin for A- and E-type olivine fabrics in all models. Fast directions with a smaller secondary delay time peak are predicted in the forearc and sub-arc areas and are perpendicular to the MDT fast direction and parallel to the margin. The fast directions for the forearc and subarc not included in the MDT and secondary peaks have a delay time that is too small to explain the typical range of delay times that are observed in subduction zones. These results indicate that the observed fast directions in the forearc and arc regions are those with either the MDT or secondary peak delay times and that there are optimal initial polarization directions relative to the anisotropic media that produce the observed delay times. In the backarc, the secondary peak in time delay is absent, and more notably, the range of predicted fast directions increases from ~20° at 30° obliquity to ~80° at 60° obliquity. The range of possible fast directions is limited to the margin-perpendicular direction for small obliquities whereas the fast directions can vary from margin-perpendicular to highly oblique to the margin for high obliquities. We plan to qualitatively compare these model predictions with observed splitting parameters in real oblique subduction zones.