Examining the Impact of Prandtl Number and Surface Convection Models on Deep Solar Convection

Abstract:

Turbulent motions within the solar convection zone play a central role in the generation and maintenance of the Sun's magnetic field. This magnetic field reverses its polarity every 11 years and serves as the source of powerful space weather events, such as solar flares and coronal mass ejections, which can affect artificial satellites and power grids. The structure and inductive properties are linked to the amplitude (i.e. speed) of convective motion. Using the NASA Pleiades supercomputer, a 3D fluids code simulates these processes by evolving the Navier-Stokes equations in time and under an anelastic constraint. This code simulates the fluxes describing heat transport in the sun in a global spherical-shell geometry. Such global models can explicitly capture the large-scale motions in the deep convection zone but heat transport from unresolved small-scale convection in the surface layers must be parameterized. Here we consider two models for heat transport by surface convection, including a conventional turbulent thermal diffusion as well as an imposed flux that carries heat through the surface in a manner that is independent of the deep convection and the entropy stratification it establishes. For both models, we investigate the scaling of convective amplitude with decreasing diffusion (increasing Rayleigh number). If the Prandtl number is fixed, we find that the amplitude of convective motions increases with decreasing diffusion, possibly reaching an asymptotic value in the low diffusion limit. However, if only the thermal diffusion is decreased (keeping the viscosity fixed), we find that the amplitude of convection decreases with decreasing diffusion. Such a high-Prandtl-number, high-Peclet-number limit may be relevant for the Sun if magnetic fields mix momentum, effectively acting as an enhanced viscosity. In this case, our results suggest that the amplitude of large-scale convection in the Sun may be substantially less than in current models that employ an effective turbulent Prandtl number of order unity. These results are found to be insensitive to the nature of the surface convection model.