We describe a series of experiments using the Geophysical Fluid Dynamics Laboratory (GFDL) modular ocean model in an idealized Atlantic Ocean forced by a fixed freshwater flux and by two different heat flux parameterization schemes. Both of the parameterizations use a bulk formula: one assumes an atmosphere of infinite heat capacity and the other assumes an atmosphere of no heat capacity. These experiments show that the behavior, the stability, and the response to external perturbations of the oceanic thermohaline circulation depend heavily upon the way in which the heat flux is parameterized and upon the associated thermal capacity of the atmosphere to which the ocean is coupled. With an infinite heat capacity atmosphere the ocean's thermohaline circulation is very sensitive to perturbations in freshwater flux. In contrast, with the alternative scheme the ocean's thermohaline circulation is stable. The fundamental difference between the two schemes is that the sea surface temperature (SST) in the first case is not allowed to respond adequately to changes in poleward heat transport, while in the latter it is free to evolve. When the SST is curtailed, the oceanic negative feedback mechanism, which stabilizes the thermohaline circulation, is disabled, and a positive feedback mechanism dominates.

The above result highlights the importance of using an appropriate thermal forcing. It is also shown that the cause for multiple equilibria in the GFDL fully coupled model and in ocean-only models under the scheme assuming infinite heat capacity is different and that multiple equilibria are likely under a no heat capacity atmosphere. Under the scheme assuming an infinite heat capacity atmosphere, interdecadal variability that is thermally driven by a mismatch between local heat storage rate and oceanic heat transport cannot survive. Under the scheme assuming a no heat capacity atmosphere, however, an SST anomaly associated with the variability is able to develop, and regular oscillations persist. Although all the above results suggest that, of the two schemes, the one assuming a no heat capacity atmosphere is a more realistic one for studies of the ocean's climate, fully coupled atmosphere-ocean models are needed to properly investigate the stability and sensitivity of the thermohaline circulation. ¿ American Geophysical Union 1995