Testing mantle dynamics : Constraining high resolution numerical spherical convection models with geochemistry and geophysics

Mantle convection is important since it drives (i) plate tectonics (the ultimate process behind seismicity and mountain building); and (ii) melting, (critical for volcanism and producing crust, cryo/hydrosphere and atmosphere) but we do not know how it 'works'. In convection, for example water heated in a saucepan, the movement of the light (hot) material from the base to the surface where it cools and sinks back down to restart the cycle again provides a very efficient heat transfer mechanism. The differences in buoyancy that drive flow, with lighter material rising and denser material sinking, can be due to differences in temperature and/or composition. Amazingly the solid mantle deforms by creep on the geological time-scale allowing Earth's mantle to also lose its heat by convection. While we know that the ocean plate is the manifestation of the surface element of this cycle on Earth, and we have incomplete knowledge from seismic imaging for the present-day geometry of this process, we have no direct evidence of the geometry in the past. The field of mantle convection is now ready to yield a significant advance using the combination of the improvements in mantle convection modelling, the maturity of geophysics and geochemical observables, and mineral physics constraints. Convection in the mantle is more complex than convection in simple systems, such as water in a saucepan, since as hot mantle reaches the surface it melts. The melt rises to the surface forming a crust, and degasses to give an atmosphere and hydrosphere, and leaves behind a residue. The combination of these processes make the modelling more interesting since the crust and residue have a different buoyancy to the starting material. Significantly it also gives us the means to constrain the process. For example the rate of melting and degassing is related to the vigour of convection. The known amount of Argon40 that has collected in the atmosphere, produced at a known rate in the mantle from Potassium40, provides an integrated constraint on the rate of degassing. We will also use the flux of primordial Helium3 and alpha particle produced He4 as further constraints. We will also look at the isotopes of lead which are the stable daughters of radioactive U and Th parents. These are further useful stopwatches on mantle convection, but are different to the inert gases since they are not degassed but are recycled. They are returned to the mantle where the convection stirs the crust, residue and starting material together. When they are melted again their Lead isotopic signature is dependent on the proportion of the various components, the stirring and the time that has elapsed since it last melted. To understand mantle stirring one needs models in the right geometry (we will model it correctly as a spherical shell) and at the right vigour (we can reach Earth-like vigour even for early Earth). The geophysics evidence suggests that present-day the mantle convects as a whole body, while geochemical evidence requires ancient isolated reservoirs. There are a large number of hypotheses in play (usually motivated by one discipline alone) trying to reconcile these constraints. We will test these hypotheses. The geochemical data-sets we will use have been collected over very many decades, by countless research teams across the globe, utilizing data whose value at collection easily exceeds £1 billion (>2000*500k). Understanding mantle convection is a zeroth order problem for solid Earth science and the project proposed will allow us to make a significant long-lasting advance. The numerical geodynamic approach allows the broadest range of constraints to be brought to bear in a quantitative manner - the basic conservation laws of physics, geophysics (including integrative ones such as size of inner core - and very high spatial resolution seismic tomography) and geochemistry observables; providing the meanest test of this proliferation of hypotheses.