The effects of the eddy-induced advection coefficient in a coarse-resolution coupled climate model

The role of the eddy-induced advection coefficient kappa, used in the Gent and McWilliams (1990) parameterisation (GM), is analysed in terms of the response to idealised wind stress perturbation experiments in the GFDL global coupled climate model CM2.1, and compared to solutions with an eddy-permitting version of the same coupled model, CM2.4. The closure implemented in CM2.1 for kappa is flow-dependent and includes a maximum limit that caps its value. In this paper, we present simulations with a modified version of CM2.1, where the upper limit for kappa is doubled to 1200 m(2) s(-1) and the cap to the isopycnal slope S(max) in GM is also increased to 1/100 from 1/500. These changes allow their product, kappa S(max), which is the upper limit to the effect of parameterised eddies, to be an order of magnitude higher than in the original CM2.1 version. Modifications to both GM parameters result in changes in the mean circulation and overall climatology that are non-negligible, which shows that attention has to be paid to the GM implementation during model development. Increasing the value of kappa does produce a stronger compensation between mean and eddy-induced meridional overturning circulations under stronger wind stress forcing, but the residual circulation response is still stronger than in the eddy-permitting model CM2.4. We show that spatially varying kappa, both in the horizontal and vertical directions, is necessary for a correct simulation of the response to changes in the wind stress. New and improved closures for kappa are needed, and should be tested in coupled climate models. (C) 2011 Elsevier Ltd. All rights reserved.