Subduction, back-arc spreading and global mantle flow

Abstract

The shapes and orientations of Benioff zones beneath island arcs, interpreted as marking the location of subducted lithosphere, provide the best presently available constraints on the global convective flow pattern associated with plate motions. This global flow influences the dynamics of subduction. Subduction zone phenomena therefore provide powerful tests for models of mantle flow. We compute global flow models which, while simple, include those features which are best constrained, namely the observed plate velocities, applied as boundary conditions, and the density contrasts given by thermal models of the lithosphere and subducted slabs. Two viscosity structures are used; for one, flow is confined to the upper mantle, while for the other, flow extends throughout the mantle. Instantaneous flow velocity vectors match observed Benioff zone dips and shapes for the model which allows mantle-wide flow but not for the upper mantle model, which has a highly contorted flow pattern. The effect of trench migration on particle trajectories is calculated; it is not important if subduction velocities are greater than migration rates. Two-dimensional finite element models show that including a coherent high viscosity slab does not change these conclusions. A coherent high viscosity slab extending deep into the upper mantle would significantly slow subduction if flow were confined to the upper mantle. The maximum earthquake magnitude, M_w, for island arcs correlates well with the age of the subducted slab and pressure gradient between the trench and back-arc region for the whole mantle, but not the upper mantle, flow model. The correlations with orientations of Benioff zones and seismic coupling strongly suggest that the global return flow associated with plate motions extends below 700 km. For both models, regions of back-arc spreading have asthenospheric shear pulling the back-arc toward the trench; regions without back-arc spreading have the opposite sense of shear, suggesting global flow strongly influences back-arc spreading.