Comparison of a Coupled Ocean–Atmosphere Model with and without Oceanic Eddy-Induced Advection. Part I: Ocean Spinup and Control Integrations

The Gent and McWilliams (GM) parameterization for large-scale water transport caused by mesoscale oceanic eddies is introduced into the oceanic component of the Commonwealth Scientific and Industrial Research Organisation global coupled ocean–atmosphere model. Parallel simulations with and without the GM scheme are performed to examine the effect of this parameterization on the model behavior for integrations lasting several centuries under conditions of constant atmospheric CO2. The solution of the version with GM shows several significant improvements over that of the earlier version. First, the generally beneficial effects of the GM scheme found previously in studies of stand-alone ocean models, including more realistic deep water properties, increased stratification, reduced high-latitude convection, elimination of fictitious horizontal diffusive heat transport, and more realistic surface fluxes in some regions, are all maintained during the coupled integration. These improvements are especially pronounced in the high-latitude Southern Ocean. Second, the magnitude of flux adjustment is reduced in the GM version, mainly because of smaller surface fluxes at high southern latitudes in the GM ocean spinup. Third, the GM version displays markedly reduced climate drift in comparison to the earlier version. Analysis in the present study verifies previous indications that changes in the pattern of convective heat flux are central to the drift in the earlier version, supporting the view that reduced convective behavior in the GM version contributes to the reduction in drift. Based on the satisfactory behavior of the GM model version, the GM coupled integration is continued for a full 1000 yr. Key aspects of the model behavior during this longer period are also presented. Interannual variability of surface air temperature in the two model versions is briefly examined using some simple measures of magnitude. The variability differs between the two versions regionally, but is little changed on the global scale. In particular, the magnitude of variability in the tropical Pacific is little changed between the versions.