Labrador Sea descend directly to the interior of the open Atlantic, and are the dominant source of upper NADW (McCartney and Talley, 1984; Clarke and Gascard, 1983). While cooling and sinking occur at a variety of locations within the Nordic seas, the dense waters overflowing the Denmark Strait and crossing the Iceland-Faeroe Ridge into the open Atlantic appear to originate primarily from intermediate-depth convection within the Iceland Sea (Aagaard et al., 1985). These waters, which are denser than those formed in the Labrador Sea, are the dominant source of lower NADW (Worthington, 1976). Together these water masses comprise approximately two-thirds or more of the total southward, trans-equatorial transport of NADW. The remainder may result from entrainment of Antarctic Bottom Water (AABW) and Antarctic Intermediate Water (AAIW) (McCartney and Talley, 1984; Schmitz and McCartney, 1993). The southward-flowing deep waters upwell toward the surface at a variety of locations in the Pacific and Indian oceans. From these locations surface waters are returned to the Atlantic either via the Drake Passage (the "cold-water route") or via the Straits of Indonesia and the Aghullas retroflection around Cape Hope (the "warm-water route"). These return flows complete the Great Ocean Conveyor loop described by Gordon (1986) and Broecker (1991a). Estimates of the net deep-water export from the North Atlantic, based on a variety of different approaches (Broecker, 1991a; Schmitz and McCartney, 1993, and references therein; Wunsch, 1984), range between 14 and 20 sverdrups (1 Sv = 106 m3/sec).
Broecker (1987, 1991b) made a rough calculation of the heat release to the atmosphere resulting from warm-to-cold water conversion by the conveyor, assuming average temperatures of 11°C for northward-flowing thermocline waters and 3°C for southward-flowing NADW. Using an estimated flux of NADW of 20 Sv, approximately 5 × 1021 calories are released to the northern atmosphere per year (i.e., about 0.7 petawatt). According to the circulation scheme of Schmitz and McCartney (1993), however, as much as 5 to 7 Sv of the total southward flow at depth is comprised of relatively old, cold, recirculated deep waters that have not been directly involved in warm-to-cold water conversion at the surface of the northern North Atlantic. Using this transport scheme, which also depicts a larger proportion of the compensatory northward flow at greater, colder depths, the average temperature of surface and near-surface waters subject to warm-to-cold conversion is 8°C, and the flux of NADW produced by convection in the northern North Atlantic is only 13 Sv, yielding an estimated heat release of about 2 × 10 21 calories per year.
The two different estimates can probably be regarded as extremes that bracket the actual heat release to the atmosphere over the northern North Atlantic. (I suggest this because (1) in these calculations the larger estimate of deepwater flux is tied to the larger estimate of temperature difference between exported deep water and returning surface water, and vice versa, and (2) the smaller estimate of NADW flux would predict a nutrient content for the deep Atlantic at the high end of what can be accommodated by current observational estimates.) In any case, the impact of the conveyor, on the basis of heat release alone, is huge. Broecker (1987, 1991a) noted that his estimate of 5 × 1021 calories per year was equivalent to approximately 30 percent of the solar energy absorbed by the troposphere over the northern North Atlantic. The fingerprint of this oceanic heat source is seen in Figure 2. Its removal would result in temperature deficits similar to those seen in Figures 3 and 4. It follows from these examples that changes in the strength of the conveyor may exert a strong influence on the climate of the North Atlantic region.
The proxies used here to assess past variations in the strength of the conveyor circulation are changes in fossil assemblages of planktonic foraminifera (carbonate-shelled protozoans) and diatoms (silica-shelled plants) with known temperature tolerances. Fossil assemblages of planktonic foraminifera, in particular, have been used routinely to chart the history of polar-front movements and related rates of surface-circulation change in the North Atlantic (McIntyre et al., 1976; Ruddiman et al., 1977; Ruddiman and McIntyre, 1981; Bard et al., 1987). The general premise is that northward penetrations of subtropical to subpolar assemblages occurred during periods when poleward flow of Atlantic surface waters was vigorous (as it is today), and that south-