Background
The authors write that "the ability of a model to adequately depict bottom water formation is crucial for accurate prediction of changes in the thermohaline circulation," citing Hay (1993); and they say that "this process is particularly challenging to represent in the current generation of climate models," noting that "the last generation of models in CMIP3 poorly represented Southern Ocean transport and heat fluxes," citing Russell et al. (2006).

What was done
In a nutshell, Heuze et al. say they assessed "Southern Ocean potential temperature, salinity, density and sea ice concentration in fifteen CMIP5 historical simulations (means of the twenty August monthly mean fields from 1986 to 2005)," after which they compared the 20-year mean model fields with historical hydrographic data and Hadley Centre sea ice climatologies.

What was learned
In the words of the four UK researchers, no model reproduces the process of Antarctic bottom water formation accurately, for "instead of forming dense water on the continental shelf and allowing it to spill off," they say that the models "present extensive areas of deep convection, thus leading to an unrealistic unstratified open ocean."

What it means
Due to the fact that Heuze et al. indicate that "bottom water formation processes are poorly represented in ocean models and are a key challenge for improving climate predictions," we should be extremely cautious in accepting what they currently predict about future climate change.

References
Hay, W.W. 1993. The role of polar deep water formation in global climate change. Annual Reviews of Earth and Planetary Sciences 21: 227-254.

Russell, J.L., Stouffer, R.J. and Dixon, K.W. 2006. Intercomparison of the southern ocean circulations in IPCC coupled model control simulations. Journal of Climate 19: 4060-4075.