Chapter 4

Modelling mantle flow


Although some mantle rocks can be found at the surface of the Earth, they are rare. The composition of the mantle – making up more than half of the Earth by volume – is generally worked out from those of its bulk properties that can be determined remotely, especially density. It is denser than the overlying crust, made of denser minerals such as olivine and spinel. Seismic waves from earthquakes increase with depth in the Earth, suggesting increasing density; abrupt increases in density within the deep Earth suggest changes in mineral phases.

By this time, plate tectonics is established as a successful description of the movement of the Earth’s surface layer; the forces that drive it are less clear. In this 1974 paper, McKenzie, RobertsRobertsJean M. Hewitt (birth name)/Jean M. Roberts (married name) (b. 1940-) graduated with a Physics degree from Manchester University. Taught physics at Cambridge Technical College (now Anglia Ruskin University). Gained software experience at the Cambridge Institute of Theoretical Physics in 1968 then joined Dan McKenzie and Nigel Weiss at Madingley Rise Bullard labs. in 1969 to create the mantle convection model. This was one of the first applications of (the then) significant computing resources required for modelling the Earth’s mantle (a Boussinesq fluid of infinite Prandl number). More complex models were developed and key papers written during the subsequent 11 years. In 1983 Jean joined the company Shape Data, (subsequently EDS, then Siemens) to develop the first boundary representation solid modeller Parasolid for CAD, (based on the PhD thesis of Ian C. Braid in the early 1970s), now the kernel of many modern computer research and design applications. She was appointed Secretary of the Institute of Biotechnology, University of Cambridge, in 1996, a post held for 12 years. and WeissWeissNigel Weiss (1936-present) is a South African astronomer and mathematician specialising in astrophysical and geophysical fluid dynamics. He is currently Emeritus Professor of Mathematical Astrophysics at the University of Cambridge. examine simplified two-dimensional models of flow in the mantle and find unexpected complications, notably that few geophysical observations related directly to mantle flow. This paper demonstrates that mantle convection is possible, and even likely, but not in the ways envisaged at that time. Their modelling of mantle properties suggests that convection cells with the width of a plate are not geometrically feasible. A key conclusion is that gravity anomalies hold information about mantle flow, now a central idea in modern satellite geodesy.

READ THE PAPER — McKenzie, D., Roberts, J. & Weiss, N.O. 1974. Convection in the Earth’s mantle: towards a numerical simulation. Journal of Fluid Mechanics. 62(2): 465-538.


Graph showing results of variable viscosity by Jean Roberts, nd, from working file for “Convection in the Earth’s mantle: towards a numerical simulation” (1974).