Deep Water Formation 
Peter D. Killworth 
Department of Applied Mathematics and Theoretical Physics 
Silver Street, Cambridge CB3 9EW, ENGLAND 
In this talk I attempted to cover, from a theoretician's point of view, many of the features 
of deep water formation, specifically as they apply to North Atlantic deep water. 
The talk began by noting that the topic of the meeting was considered highly germane by 
the WOCE (World Ocean Circulation Experiment); their specific objective 4 related specifically to 
water mass formation. [For a discussion of WOCE, see Nowlin (1984).] First, some simple 
arguments on plumes of dense water and filling boxes were given. What determines the time for 
a large-scale environment to be modified by the injection of dense water at its edge is the mass 
flux, not the buoyancy flux. However, it is the denser buoyancy flux, when there are several 
competing plumes (e.g. the Mediterranean outflow versus the Denmark Strait outflow) that deter- 
mines which plume will provide the bottom water for that ocean basin. 
It was noted that the 'obvious' laboratory experiment (rotate a pie-shaped annulus, and 
heat/cool it on the surface) had never been performed. Thus, to some extent our belief that 
deep convection is somehow automatic at high latitudes to close off some ill-defined meridional 
circulation has never been tested. 
A summary of what we believe to be true about deep convection was given, taken largely 
from Killworth (1983). The two fundamental formation mechanisms are shown in Fig. 1. Of the 
two, it is open-ocean convection which forms the water which supplies the Denmark Strait 
overflow -- in all likelihood, as formation in the Greenland Sea remains stubbornly unobserved. 
But it is the slope convection which finally creates North Atlantic deep water, following the 
Denmark Strait overspill. 
surface buoyancy loss 
(WM AA 
(a) 
surface 
buoyancy 
loss 
LE EEC_EC@MRCMCqQCCU;A_CmC_:#]! M€JHHTT]@@#####za##™™7 
(b) 
Fig. 1. Two basic mechanisms for deep water formation. 
