Some effects of anisotropy on velocity contrasts between subducting lithosphere and overriding mantle

We compute velocity anisotropy for three models of preferred crystal alignment due to finite strain in the mantle which, when juxtaposed with computed anisotropy in subducted lithosphere and neglechng other eflects, give velocity contrasts that we compare with those obscrvcd. We find a strong dip dependence for computed velocity contrasts due to slow axis alignment in the overriding mantle perpendicular to the slab- mantle interface. Moreover, for an isotropic mantle wedge or convergence-parallel flow, it is difficult to produce greater velo- city in the subducting slab than the overlying mantle for near- vertical pmpagation even with high strain rates. We do. how- ever, obtain positive contrasts using a dip axis-parallel flow model, which is the only one that yields contrasts that approash observations fmm the Japan Trench. chemical discontinuities, and deflection of the a-p olivine phase boundary, they also suggesled thal ulivinc crystal orientation might be partially responsible. We test this latter mechanism using models of lattice preferred orientations in subducting slabs and idealized flaw pattcrns in the overlying manlle. From the juxiaposifion of slab and mantle crystal orientations, we cal- culate theoretical velocity contrasts. Wc assume a lattice preferred orientation in the upper litho- sphere that becomes "frozen" in as it cools, thereafter remain- ing constant, in the local coordinate system, even iI the local stress field changes. We also assume. following previous experimental, field, and theoretical considerations, that the fast axes of olivine crystals tend to align with the axis of greatest finite strain during induced mantle flow at spreading centers (e.g., Hess. 1964: Raht et al., 1971; McKenzie, 1979; Ribe, 1989al. Such alignment suggests that olivine fast axes in the ocean floor mughly parallel the paleospreading direction, an assumption which appears to be consistent with seismic obser- vations (e.g.. Forsylh, 1975; Shearer and Orcurl. 1986; Nishimura and Forsyth. 19891. Crystal axis orientations in sub- ducting slabs are computed by mapping these surface orienta- tions Qwndip thmugh a rotation about the dip axis.