Mark R. Petersen
Los Alamos National Laboratory
in collaboration with Matthew W. Hecht, Beth A. Wingate, Darryl D. Holm
The LANS-alpha Turbulence Parameterization in Ocean Modeling
The Lagrangian-averaged Navier-Stokes-alpha (LANS-alpha) model is a turbulence parameterization that increases eddy activity near the grid-scale. We will review the implementation and results of LANS-alpha in the POP ocean model. Simulations with LANS-alpha in an idealized channel model domain produce turbulence statistics like a doubling of resolution, including kinetic energy, eddy kinetic energy, and temperature profiles. The cost of adding LANS-alpha is only 30%, versus a factor of ten for doubling the horizontal resolution.
LANS-alpha, which is derived using Hamiltonís principle, modifies the momentum equation with an extra nonlinear term and a smoothed advecting velocity. The alpha parameter controls the strength of smoothing in a Helmholtz inversion operator, and thus the strength of the turbulence model. We show that efficient filters can replace the Helmholtz inversion, with similar results.
A linear dispersion analysis of LANS-alpha shows that it increases the effective Rossby Radius, thus allowing more eddy activity near the grid-scale in low resolution ocean simulations. Preliminary simulations in a North Atlantic domain produce deeper and more realistic penetration of eddy kinetic energy in the North Atlantic Current.