When atmospheric greenhouse gas levels are specified to increase in GFDL's multi-century coupled climate model simulations, the North Atlantic's meridional overturning circulation (MOC) noticeably weakens. While this general result is evident in three generations of GFDL coupled models used to conduct IPCC-style experiments during the past two decades (the R15, R30, and CM2 models), differences do exist between the models' MOC responses. That response differences exist is not necessarily surprising, since the three GFDL models differ considerably in their grid resolutions and in the physics of the component models. Such MOC response differences are consistent with the observation that significant Atlantic MOC weakening is common, but not ubiquitous, in climate change projection experiments performed with various IPCC-class global AOGCMs.

Comparisons of the Atlantic MOC simulated in the three GFDL models' control experiments will be presented, including the mean MOC strength, the preferred locations of deep water formation, and the MOC's low- frequency variability. Similarly, the three models' MOC response to increased greenhouse gas forcing will be compared. Sets of "partially- coupled" experiments run using the older R15 and R30 models help quantify to what extent MOC weakening in a climate change experiment can be attributed to changes in each of three types of air-sea fluxes that drive the ocean circulation - the surface heat, freshwater, and momentum fluxes. Preliminary results suggest that the role of freshwater flux changes is greatest in the R15 model, while in the R30 model most of the MOC weakening is associated with changes in surface heat fluxes. Greenhouse gas induced changes in wind stress forcing appear to have minimal impact on the MOC response in the climate change experiments. Finally, the GFDL models' MOC responses to increasing levels of greenhouse gases will be compared with the results of models from other research centers.