Preliminary models of 3-D oceanic plateau subduction: influence of the width ratio between plateau and slab

Subduction is driven by the negative buoyancy of the slab, which depends on both the temperature and composition of the lithospheric plate. Before subduction, the oceanic crust and underlying harzburgite layer have positive buoyancy. As a result an oceanic plateau or an over-thickened oceanic crust introduces an excess of buoyancy and decreases proportionally the driving forces. Previous 2-D numerical models have shown that an oceanic plateau subduction, associated to a young plate (30 My), can modify the slab behavior during the subduction (Arrial & Billen, in review). In a kinematically-driven model (convergence rate of 5 cm/yr), a partially eclogitized plateau can lead to a flat slab. However, eclogitization limits the depth extent of buoyant material and all plateaus eventually subduct regardless of thickness or width. In contrast, in dynamic models, oceanic plateau subduction does not lead to slab flattening, but we observe a decrease in the convergence rate proportional to the plateau geometry, and slab detachment and cessation of subduction for large plateaus. Extrapolation of the 2-D results to 3-D geometry predicts that sufficient slab buoyancy is available to sustain subduction for narrow plateaus with slab widths 6-7 times the plateau width (depending on plateau thickness and length). Here we show the preliminary results for 3-D models of oceanic plateau subduction investigating how the width ratio between the plateau and the slab affect subduction dynamics. We have modified the code CitcomS to introduce an eclogitization reaction and a composition dependent viscosity. Using particles the code track the three main compositions: basalt, harzburgite and lherzolite. The basalt/eclogite reaction is modeled as a progressive increasing of basaltic density. Furthermore, in order to maintain a free subduction and slab retreat, we impose a weaker viscosity for the basaltic composition.