Firm mantle plumes and the nature of the core–mantle boundary region

Recent tomographic imaging of thick plume conduits in the lower mantle, when combined with plume buoyancy flux based on hotspot swell topography, indicates a very high plume viscosity of 1021–1023 Pa s. This estimated plume viscosity is comparable or may even be greater than the viscosity of the bulk lower mantle, the estimate of which ranges from 2×1021 to 1022 Pa s. Here I show that both very high viscosity and large radii of lower-mantle plumes can be simultaneously explained if the temperature dependency of lower-mantle rheology is dominated by the grain size-dependent part of diffusion creep, i.e., hotter mantle has higher viscosity. Fluid-dynamical scaling laws of a thermal boundary layer suggest that the thickness and topography of the D″ discontinuity are consistent with such mantle rheology. This new kind of plume dynamics may also explain why plumes appear to be fixed in space despite background mantle flow and why plume excess temperature is only up to 200–300 K whereas the temperature difference at the core–mantle boundary is likely to exceed 1000 K.