118 Testing Mesoscale Eddy Parameterizations in the `Neverland' Configuration

Tuesday, 27 June 2017
Salon A-E (Marriott Portland Downtown Waterfront)
Sina Khani, Princeton University, Princeton, NJ; and A. Adcroft and M. F. Jansen

We introduce a new idealized set-up as a testbed to evaluate mesoscale eddy parameterizations. This new geometrical set-up is called `Neverland' and is implemented in the latest generation of the Modular Ocean Model (MOM6). The geometry, which is limited to the Southern hemisphere, consists of a Southern Ocean along with two basins, and can be readily set at any desirable resolution. The model is entirely adiabatic, allowing for integration to full equilibrium in only $100$ years of simulation time, which is small enough that we can afford to run at fine resolution.

A fine resolution simulation at $1/16$ degree resolution is performed, and used as a reference solution. We also perform simulations at $1/4$ and $1$ degree resolutions with different eddy parameterizations. These parameterizations include three Gent and McWilliams based formulations: a constant eddy diffusivity coefficient, a variable eddy diffusivity following Visbeck/Green, and an eddy diffusivity based on an explicit mesoscale eddy kinetic energy (MEKE) budget. A negative viscosity (backscatter) parameterization based on a subgrid eddy kinetic energy budget is also tested. We analyze results by comparing the mean flow characteristics, isopycnal displacements, energy spectra and eddy diffusivity coefficients between these simulations. Preliminary results indicate only moderate differences between different formulations of the Gent and McWilliams parameterization. It is also shown that at $1/4$ degree resolution, the backscatter scheme outperforms all other eddy parameterizations.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner