Deep slab flattening and temporary stagnation within the lower mantle transition zone

Temporary slab stagnation is a common morphological phenomenon of the mid-mantle with almost half the slabs world-wide flattening between 660 and 1000 km depth (Goes et al., 2017, Fukao and Obayashi, 2013 and Fukao et al., 2009). While many mechanisms have been proposed for temporary stagnation around 660 km depth, including buoyancy changes due to the phase changes of ringwoodite to bridgmanite and periclase, deeper slab flattening is harder to explain. This is due to a lack of obvious physical mechanisms and/or mineralogical phase changes beyond 660 km depth. However, tomography models show a clear change in mantle characteristics at 1000 km depth. This is observed both through flattening slab morphologies and the ponding of slow material below and around a 1000 km, interpreted by French and Romanowicz (2015) as deflected plume material below the transition zone. A radially varying mantle viscosity profile is also proposed by Rudolph et al., (2015) who infer a significant mantle viscosity increase at a 1000 km depth. In the work presented here, we use numerical simulations to revisit slab flattening at 1000 km. We attempt to understand the relationship between ambient mantle strength, radial mantle viscosity variations, slab strength at depth and the occurrences of deep slab flattening. Our models of 2-D self-consistent, one sided subduction are solved by the finite difference/volume multigrid code StagYY (Tackley, 2008). Our initial condition assumes active subduction, and a low viscosity layer between 660 and 1000 km depth based on the work of Rudolph et al., (2015) and Kido and Cadek, (1997). Our models exhibit slab flattening at a 1000 km depth without the added stagnating effects of phase changes or compositional variations. We thus, propose that deep slab flattening at 1000 km is the result of the relative strength ratios of the slab and the ambient mantle, as defined by the mantle's radial viscosity profile.