Plumes: The Overview

Abstract

The plume hypothesis has not yet been cast in a form where it can be tested, or falsified. Currently it has low empiric content and high probability (i.e., it predicts everything). Many of the phenomena which have been attributed to plumes appear only to require access to the mantle, extensional stresses in the plate, lithospheric boundaries or normal mantle convection. About 50% of the upper mantle has such low seismic velocities that the presence of a fluid phase is implied, probably a melt phase, and all that is required to generate magmatism is appropriate plate conditions. Even the coldest, highest seismic velocity, sublithospheric upper mantle will melt upon adiabatic ascent, if the plate is removed or broken. The large-scale lateral variation of seismic velocity in the mantle can almost entirely be explained by continental insulation, cooling by past subduction, volatile fluxing by subduction and lithospheric thickness variations. For example, the non-Pacific hemisphere mantle has been repeatedly cooled by slabs, over several supercontinental cycles, and the shallow mantle has been repeatedly fluxed by volatiles. This mantle cannot mix readily with the Pacific mantle and large chemical, physical and thermal domains are established and maintained. In addition, thick cratonic lithosphere has both a geometric and thermal effect on convection, dictating, to a large extent, the locations of downwellings and upwellings. Previous subduction tiles over the base of the system which is therefore cooled from below in places. This does not occur in Rayleigh-Bénard convection with temperature independent properties, because thermal boundary layers are only marginally buoyant and do not decouple from background flow.