Comparison of insolation and various temperature proxy records for the last deglaciation (0-20 kyr BP). Beginning at the left, they are: the slow change in summer insolation for the Northern Hemisphere during that interval (Berger, 1978); oxygen isotope results from the Greenland ice core Dye 3 (Dansgaard et al., 1984); oxygen isotope results from Lake Gerzensee marls in Switzerland (Siegenthaler et al., 1984); and the mean annual temperature estimate for the British Isles based on fossil coleoptera (Atkinson et al., 1987). Note that summer insolation in the Northern Hemisphere, expressed as a percentage deviation from today's values (first panel), changed gradually over the interval 20 kyr to present. However, various climate proxy indicators from around the North Atlantic region indicate that air temperatures changed abruptly (remaining panels). Detailed studies of Dye 3 by Dansgaard et al. (1989) demonstrate that the large isotope shift at about 10,000 yr BP occurred within 50 years and corresponded to a temperature change of 7°C.
Deviation of local temperatures from the zonal average, for January, in °C. (From Barry and Chorley, 1982; reprinted with permission of Routledge.) The largest positive anomalies overlie the area of conveyor overturning in the northern Atlantic and Norwegian Sea.
Atlantic (see, e.g., Bryan, 1986; Manabe and Stouffer, 1988). In many cases the model circulation is observed to collapse within a few decades of surface freshening (Maier-Reimer and Mikalojewicz, 1989; Stocker and Wright, 1991).
As an example of the possible climate change arising from collapse of the conveyor circulation, Figure 3 shows the depression of (annual) surface ocean and air temperatures associated with the conveyor-off mode in the GFDL coupled atmosphere/ocean general-circulation model (Manabe and Stouffer, 1988). The main point to be made here is that even though the sensitivity of the overturning circulation to fresh-water and temperature forcing at the surface in this (or any other) model cannot yet be validated, cooling of the surface ocean in response to weakened or eliminated conveyor circulation is capable of producing pronounced cooling of the air over and downwind of the North Atlantic. This is further borne out by the results of an earlier experiment in which Rind et al. (1986) used the NASA-GISS atmospheric general-circulation model with sea surface temperatures specified (rather than freely pre-