Slab Stagnation in the Lower Mantle: A Multidisciplinary Investigation

Recent tomographic models show that while many slabs seem to deflect or stagnate at the 660 km discontinuity, some slabs continue to subduct deeper and pond at 1000 km below the earth's surface (Fukao and Obayashi, 2013). Only one slab is observed to penetrate significantly deeper into the mantle. Furthermore, some mantle upwellings also appear to be deflected at 1000 km in depth. The radial correlation functions for the low-order spherical harmonics of most tomographic inversions show that while seismic wave velocities are correlated for all depths below ~1000 km, velocities at depths between 400-1000 km are uncorrelated with velocities at any other depth. This implies that there are large scale velocity features coherent from 1000 km to the core-mantle boundary, but no large scale features coherent from the top of the transition zone down to 1000 km. Seismic studies using precursors and receiver functions find evidence for numerous reflectors in the mid-mantle, ranging from 900 km in depth beneath the southern Pacific and southeast Asia to 1200 km beneath Europe and Japan. This range of depths could indicate topography along a single laterally continuous discontinuity or result from multiple unconnected features. Some reflectors are geographically near, and therefore may be associated with, subducted slabs, however the origin of the others is unclear. The 1000 km 'discontinuity' could potentially be explained by an increase in viscosity or density, such as a compositional difference in the mantle below this depth. We use an interdisciplinary approach to investigate the diversity in apparent slab stagnation behavior and which geophysical mechanisms prevent subduction into the lower mantle. The controlling factor may be a function of the slab itself, including subduction rate, trench rollback, composition, or temperature. Alternatively, bulk mantle properties may control slab penetration. We perform 2D and 3D numerical simulations to determine the influence of viscosity on stab stagnation and flow. We also explore the influence of a density jump at 1000 km. Finally, we generate theoretical seismograms for various velocity and density jumps in the mid mantle, and compare to seismic observations. This work was initiated at the CIDER 2014 program.