Whole Mantle Convection with Two Structures and Timescales of Flow

Mantle convection has often been debated to be either a mode of ‘top-to-bottom’ whole mantle convection, or flow within separated geochemical ‘reservoirs’ such as a denser layer often proposed to be the origin of lower mantle LLSVPs. Here we propose a straightforward resolution in which plate tectonic downwelling is linked to a ~3000 km-broad N-S circumglobal ‘ring’ of higher-than-average seismic wavespeeds in the lower mantle that has been recognized since the first global models of non-radial seismic structure. In the high-viscosity lower mantle, subduction-linked downwelling occurs at speeds of <~1.3 mm/yr, which is the origin of the long-known ~1.7Ga ‘isochrons’ seen in both hotspot and mid-ocean ridge volcanism.  This ~3000 km-wide great-circle ring of slow downward flow is associated with two antipodal axial spokes of twice-as-fast but still very slow largescale upward flow in the ‘LLSVP’ regions. In addition to this background pattern of large-scale lower mantle circulation, upward counterflow to plate subduction preferentially takes material from a warmer D’’ thermal boundary layer at the core-mantle boundary through ~10-20 mantle plumes that feed a sublithospheric plume-fed asthenosphere. In the lower mantle, the relatively warmer and lower viscosity plumes preferentially rise through and are slowly attracted towards the LLSVP regions by the low-order mode of slow lower mantle flow, with plume-conduits further warming their surrounding LLSVP lower mantle.

In this contribution we review the seismological and geochemical observations that support this scenario of two interlocking modes of whole mantle convection with very slow flow in the lower mantle that is linked to and pierced by much faster flow in a D’’-plume-asthenosphere upward flow circuit. We then present 3-D thermomechanical models designed to elucidate under what conditions this mode of flow can arise from a highly variable viscosity mantle with both internal heating and significant heatflow across the core-mantle boundary. Finally we briefly touch on some further implications of this scenario for Earth’s radial mantle structure, supercontinent evolution, the geoid, and the geodynamo.