Reconciling laboratory and observational models of mantle rheology in geodynamic modelling

Abstract

Experimental and geophysical observations constraining mantle rheology are reviewed with an emphasis on their impact on mantle geodynamic modelling. For olivine, the most studied and best-constrained mantle mineral, the tradeoffs associated with the uncertainties in the activation energy, activation volume, grain-size and water content allow the construction of upper mantle rheology models ranging from nearly uniform with depth to linearly increasing from the base of the lithosphere to the top of the transition zone. Radial rheology models derived from geophysical observations allow for either a weak upper mantle or a weak transition zone. Experimental constraints show that wadsleyite and ringwoodite are stronger than olivine at the top of the transition zone; however the uncertainty in the concentration of water in the transition zone precludes ruling out a weak transition zone. Both observational and experimental constraints allow for strong or weak slabs and the most promising constraints on slab rheology may come from comparing inferred slab geometry from seismic tomography with systematic studies of slab morphology from dynamic models. Experimental constraints on perovskite and ferropericlase strength are consistent with general feature of rheology models derived from geophysical observations and suggest that the increase in viscosity through the top of the upper mantle could be due to the increase in the strength of ferropericlase from 20–65GPa. The decrease in viscosity in the bottom half of the lower mantle could be the result of approaching the melting temperature of perovskite. Both lines of research are consistent with a high-viscosity lithosphere, a low viscosity either in the upper mantle or transition zone, and high viscosity in the lower mantle, increasing through the upper half of the lower mantle and decreasing in the bottom half of the lower mantle, with a low viscosity above the core. Significant regions of the mantle, including high-stress regions of the lower mantle, may be in the dislocation creep (power-law) regime. Due to our limited knowledge of mantle grain size, the best hope to resolve the question of whether a region is in diffusion creep (Newtonian rheology) or dislocation or grain-boundary creep (power-law rheology), may be the presence of absence of seismic anisotropy, because there is no mechanism to rotate crystals in diffusion creep which would be necessary to develop anisotropy from lattice preferred orientation. While non-intuitive, the presence or absence of a weak region in the upper mantle has a profound effect on lower mantle flow. With an asthenosphere, the lower mantle organizes into a long-wavelength plan form with one or two (degree 1 or degree 2) large downwellings and updrafts, which may contain a cluster of plumes. The boundary between the long-wavelength lower mantle flow and upper region flow may be deeper, likely 800–1200km, than the usually assumed base of the transition zone. There are competing hypotheses as to whether this change in flow pattern is caused by a change in rheology, composition, or phase.