Numerous planetary bodies contain internal liquid layers in the form of either partially molten iron cores, buried water oceans or primitive magma oceans. Convection in these layers is usually driven by the combination of two buoyancy sources: a thermal source directly related to the planet’s secular cooling, the release of latent heat and possibly radioactive decay, and a compositional source due to some process of cristallisation or fusion, for example the growth of a solid inner core which releases light elements into the liquid outer core, or the melting/freezing of an ice layer which locally enriches or depletes the adjacent water ocean in salts. The dynamics of fusion/crystallization being dependent on the heat flux distribution, the thermochemical boundary conditions are locally coupled at a melting/crystallizing boundary which may affect the convection in various ways, particularly if heterogeneous conditions are imposed at one boundary. In addition, the thermal and compositional molecular diffusivities usually differ by at least 2 orders of magnitude. This can produce significant differences in the convective dynamics compared to pure thermal or compositional convection due to the potential occurence of double-diffusive phenomena. Traditionally, temperature and composition have been combined into one single variable called codensity under the assumption that turbulence mixes all physical properties at an "eddy-diffusion" rate. This description does not allow for a proper treatment of the coupling of the thermochemical boundary conditions and is probably incorrect inside stably stratified layers in which turbulence is diminished and double-diffusive phenomena can be expected. Temperature and composition should therefore be treated separately in simulations, but the weak diffusivity of the compositional field is technically difficult to handle in current geodynamo codes and requires the use of a semi-Lagrangian description to minimize numerical diffusion. During my PhD, I implemented and tested a semi-Lagrangian ”particle-in-cell” method into a geodynamo code (PARODY, E. Dormy, J. Aubert) to properly describe the compositional field. In this seminar, I will describe the general principles of this method and will discuss its advantages compared to classical field descriptions. I will then show first applications of this new tool to the formation of a chemically stratified layer below the CMB, and to geodynamo simulations including the coupling of the thermochemical boundary conditions.
[mehr]
