Sandbox Simulations of the Evolution of a Subduction Wedge following Subduction Initiation

Subduction wedges at accreting subduction zones are bounded by a landward dipping pro-shear zone (= subduction thrust) and a seaward-dipping retro-shear zone in the overriding plate. For the Cascadia subduction zone, the surface trace of the retro-shear zone corresponds to the east side of the Coast Ranges of Oregon and Washington and the Insular Mountains of Vancouver Island. This coastal high or forearc high shows clear evidence of long-term uplift and erosion along its entire length, indicating that it is an active part of the Cascadia subduction wedge. The question addressed here is what controls the location of the retro-shear zone? In the popular double-sided wedge model of Willet et al (Geology 1993), the retro-shear zone remains pinned to the S point, which is interpreted to represent where the upper-plate Moho intersects the subduction zone. For this interpretation, the relatively strong mantle is considered to operate as a flat backstop. That model, however. is somewhat artificial in that the two plates collide in a symmetric fashion with equal crustal thicknesses on both sides. Using sandbox experiments, we explore a more realistic configuration where the upper and lower plate are separated by a gentle dipping (10 degree) pro-shear zone, to simulate the initial asymmetric geometry of the subduction thrust immediately after initiation of subduction. The entire lithosphere must fail along some plane for subduction to begin and this failure plane must dip in the direction of subduction. Thus, the initial geometry of the overriding plate is better approximated as a tapered wedge than as a layer of uniform thickness, as represented in the Willett et al models. We demonstrate this model using time-lapse movies of a sand wedge above a mylar subducting plate. We use particle image velocimetry (PIV) to show the evolution of strain and structure within the overriding plate. Material accreted to the tapered end of the overriding plate drives deformation and causes the retro-shear zone to propagate rearward with time. The main conclusion is that the rearward propagation will cease only when 1) the retro shear zone reaches the S point (i.e. the mantle cutoff in the upper plate) or 2) the erosion outflux from the subduction wedge matches the accretionary influx. Given the location of the upper plate Moho at Cascadia, it seems that erosion is the control factor in pinning the retro shear zone there.