Laboratory and numerical studies of subduction zone anisotropy and the structure of subducted lithosphere

The deep structure and fate of subducted slabs, as well as the development of upper mantle anisotropy in the vicinity of a subducting slab, are studied using laboratory experiments. Rigid plexiglas plates, whisker particles, and a fluid with temperature-dependent viscosity are used. Upper mantle anisotropy is also studied numerically.;I determine experimentally the behavior of a cold, negatively buoyant, highly viscous slab incident on a fluid interface with a viscosity and a density increase. The experiments are scaled to mantle conditions. A wide range of slab deformation styles are observed when slight changes in the parameters are imposed. Sinking slab, stagnant slab, spreading slab, sinking pile and stagnant pile represent the five different modes identified. Slab penetration into the lower layer takes the form of folded slab piles, and stagnant behavior is promoted by rapid retrograde trench motion. The sensitivity of slab behavior to transition zone density and viscosity changes and to trench migration can account for the coexistence of different modes of deformation in subducted lithosphere.;I use laboratory and 2-D numerical experiments to determine the orientation of the olivine a-axis in the upper mantle in the vicinity of a subducting slab. A laboratory analog for olivine a-axis motion during creep deformation is developed using small cylinders (whiskers) suspended in a viscous fluid. Subduction beneath a stationary overriding plate, variable dip angles and realistic rollback and down-sip plate kinematices are examined. For all dip angles examined, rollback produces strong plate-parallel whisker alignments. Plate-parallel alignment is confined near the symmetry axis for a vertical plate and is distributed along the plate length for non-vertical dips. Alignment in the wedge occurs in three layers for leading edge depths equivalent to 800 km to 1600 km; a layer of dip-parallel alignment above the downgoing plate, a layer of subhorizontal alignment beneath the overriding plate, and a core of intermediate orientations in the wedge. The amount of alignment beneath the overriding plate and the wedge core decrease with increasing dip angle. My results indicate that olivine a-axis orientation in the seaward-side mantle is controlled by the amount of slab rollback, and that orientations in the mantle wedge depend upon slab dip angle.;Numerical results for the rollback of a vertical plate show that olivine a-axis preferred orientation in the seaward side mantle is plate-parallel for rollback of 1500 km to 2000 km as in the laboratory experiments. Down-dip numerical results produce trench-perpendicular olivine a-axis orientations beneath the overriding plate, dip-parallel alignment above the downgoing plate and a core of random orientations for leading edge depth of 800 km to 1600 km. For the down-sip numerical model, calculated shear wave delay times for a layer depth of 400 km depend on leading edge depth and vary with distance from the trench. The range of delay times predicted in the numerical experiments are similar to those determined from seismic observations. (Abstract shortened by UMI.).