Terrestrial Heat Flow and the Mantle Convection Hypothesis

Abstract

Summary The paper aims to show that the distribution of terrestrial heat flow is consistent with the mantle convection hypothesis provided that certain restrictions are placed on allowable patterns of convection. Some objections to convection are first discussed. Numerical experiments have been carried out on the loss of heat from the upper surface of a mantle convection cell, assuming a simplified flow pattern. The results are compared with observed heat flow. Subject to the assumptions of the models, the results suggest that sub-oceanic mantle convection currents, if they exist, are overlain by a layer 50–100 km thick which is stationary or moves much less rapidly. This appears to rule out the mechanism of continental drift by ocean floor spreading as suggested by Dietz; continental drift affecting the overlying layer must occur at a much lower velocity than the convection current. If thermal diffusivity remains approximately constant with depth the velocity near the upper surface of the convection cell needs to be at least about 20 cm/y to explain the uniformity of heat flow, but if radiative conduction becomes dominant a lower velocity would be acceptable. The experiments have been extended to a convection cell flowing beneath a continental margin. The results suggest that the approximate equivalence of oceanic and continental mean heat flow can best be explained if convection currents are generally present beneath oceans but absent beneath continents, unless the continental crust has a much lower radioactivity than is normally supposed. The anomalous low heat flow of Pre-Cambrian shields suggests absence of convection in the mantle beneath since the Pre-Cambrian. Conditions particularly favourable to large-scale partial fusion of the upper mantle occur in the topmost section of a rising convection current. This should result in abundant igneous activity and local areas of anomalously high heat flow. This supports the view that ocean ridges overlie uprising convection currents. Belts of relatively low heat flow adjacent to the ocean ridges are difficult to explain if the convecting mantle behaves as a Newtonian viscous fluid, but are explicable on the non-Newtonian model of convection recently suggested by Orowan.