HR: 1340h
AN: GP33A-0918    [Abstracts]
TI: A dynamic large eddy dynamo simulation in a rotating spherical shell
AU: * Matsui, H
EM: matsui@geosci.uchicago.edu
AF: Department of Geophysical Sciences, the University of Chicago, 5734 S Ellis Ave., Chicago, IL 60637, United States
AU: Buffett, B A
EM: buffett@geosci.uchicago.edu
AF: Department of Geophysical Sciences, the University of Chicago, 5734 S Ellis Ave., Chicago, IL 60637, United States
AB: Flow and magnetic field in the outer core are distributed over a vast range of length scale from the size of the outer core to the thickness of the boundary layers. Numerical simulations cannot include the full range of scales, so sub-grid scale (SGS) models are required to account for the effects of the unresolved fields on the large scale fields in geodynamo simulations. We previously performed a large-eddy simulation (LES) of a dynamo in a rotating plane layer model using the dynamic scale-similarity SGS model, In the present study, we implement the dynamic similarity model in a dynamo simulation in a rotating spherical shell. We include terms for the SGS momentum and heat flux, the SGS Lorentz force, and the SGS magnetic induction. We also correct the commutation error caused by interchanging the order of the spatial differentiations and filtering operation. The amplitudes of the SGS term and commutation error correction are adjusted using model coefficients, which are evaluated automatically with a dynamic scheme, based on the Germano identity. In the present study, we use a snapshot of the dynamo simulation without SGS model as an initial value. This simulation is performed using 5.2 million elements. The fluid region of the domain has 288 and 96 element in the zonal and radial direction. The initial solution is interpolated onto a coarser grid with half of the spatial resolution in each direction. We perform dynamo simulations with the SGS models (e.g. LES), and compare these results with the simulation on the original fine grid without the SGS model (e.g. resolved DNS). The predicted model coefficients for each SGS term increase gradually with the radius in the convective region, and decrease rapidly in the boundary layers, consistent with theoretical expectations. We also obtain preliminary estimates for the energetics of the simulations. We find that the buoyancy flux in the LES is significantly smaller than that in the resolved DNS case. This result suggests that the SGS heat flux reduces the variance in temperature perturbations in the LES. The SGS Lorentz force and SGS induction term also have an important role transferring the kinetic energy and magnetic energy from the resolved components to unresolved components. The SGS momentum flux also transfers the kinetic energy from the resolved components to unresolved components, but the amplitude of the energy flux of the SGS momentum flux is smaller than that of the SGS Lorentz force. Further tests and more quantitative investigations will be presented.
DE: 1510 Dynamo: theories and simulations
SC: Geomagnetism and Paleomagnetism [GP]
MN: 2007 Fall Meeting