Evolving mantle convection from bottom up to top down

Abstract

When it comes to convection, what goes up must come down. Or is it, what goes down must come up? The truth is it depends. Although convection must be mass balanced, there is no reason that it must be force balanced: the positive and negative buoyancy forces driving convection up and down, respectively, do not necessarily need to be balanced. The balance, or imbalance, all depends on the top and bottom boundary layers. Thus, convection in Earth's mantle depends on the temperature differences across the core-mantle boundary below and the lithosphere-asthenosphere boundary above. Convective asymmetry predominated by positive buoyancy, or bottom-up convection, would be driven by plume ascent, whereas if it were predominated by negative buoyancy, or top-down convection, it would be driven by plate subduction. Symmetric convection would balance plume ascent and plate subduction. Is mantle convection on Earth balanced, dominantly top down or bottom up, or time dependent?