Geological, tomographic, kinematic and geodynamic constraints on the dynamics of sinking slabs

We use geodynamic models with imposed plate velocities to test the forward-modeled history of subduction based on a particular plate motion model against alternative seismic tomography models. We utilize three alternative published reference frames: a hybrid moving hotspot-palaeomagnetic, a hybrid moving hotspot-true polar wander corrected-palaeomagnetic, and a Subduction Reference Frame, a plate model including longitudinal shifts of subduction zones by matching subduction volumes imaged by P-wave tomography, to assess which model best predicts present day mantle structure compared with seismic tomography and volumetrically derived subduction history. Geodynamic modeling suggests paleo-longitudinal corrections applied to the Subduction Reference Frame result in lower mantle slab material beneath North America and East Asia accumulating up to 10-15 degrees westward of that imaged by tomography, whereas the hybrid models develop material offset by 2-9 degrees. However, the Subduction Reference Frame geodynamic model produces slab material beneath the Tethyan Domain coinciding with slab volumes imaged by tomography, whereas the hybrid reference frame models do not, suggesting regional paleo-longitudinal corrections are required to constrain slab locations. We use our models to test inferred slab sinking rates in the mantle focusing on well-constrained regions. We derive a globally averaged slab-sinking rate of 13 +/- 3 mm/yr by combining the ages of onset and cessation of subduction from geological data and kinematic reconstructions with images of subducted slabs in the mantle. Our global average slab-sinking rate overlaps with the 15-20 mm/yr rate implied by mantle convection models using a lower mantle viscosity 100 times higher than the upper mantle. (C) 2013 Elsevier Ltd. All rights reserved.