which could delay by several decades the predicted slowing of the THC (e.g., Delworth and Mann, 2000).
Numerical simulations suggest that this enhanced production of intermediate-level water (which forms the upper North Atlantic Deep Water) is enough to energize, strongly and quickly, the entire THC of the Atlantic, and its closely correlated meridional heat transport (Häkkinen, 1999; Cheng, 2000). Model simulations allow us to make connections between observable and unobservable quantities, for example the sea-surface elevation field (seen by satellite altimeter), meridional heat transport and meridional volume flux (Häkkinen, submitted). These model simulations emphasize the fast-track response of the intermediate-depth THC during abrupt swings of climate, which may be quite different from changes in the deeper branch of the THC driven by overflows from farther north.
These changes during the instrumental era are in themselves large perturbations of the climate system. Still larger changes have been recorded in paleoclimate data, suggesting that complete shut-down of Labrador Sea deep convection, and its associated contribution to the THC, is possible. Hillaire-Marcel et al. (2001) inferred long periods without Labrador Sea convection during the previous interglacial period. Under a more heavily perturbed, warmer climate, we could see this state recur; learning to recognize its possible onset is a major goal of current research.
Important events embedded in the longer-term decadal variability of the Atlantic are the “great salinity anomalies,” which seem to involve out-pourings of low-salinity surface water and ice from the Arctic. One of the strongest developed during the late 1960s. The salinity of the upper subpolar Atlantic decreased during a series of mild winters at the same time that intense northerly winds occurred over the Fram Strait. It has been inferred that a great increase in Arctic ice and low-salinity water was driven into the Atlantic, although the influence is controversial. The combined effects of mild winters and buoyant surface layer all but halted deep convection in the Labrador and Irminger Seas. Low-salinity water was tracked as it progressed cyclonically around the high-latitude Atlantic (Dickson et al., 1988) and was seen moving into the Norwegian Sea in 1979; its influence was seen as far south as the Azores. “Lesser” salinity anomalies also appear in the instrumental record (Belkin et al., 1998). Finally, the long overall decrease in the salinity of the upper subpolar Atlantic since 1972 can similarly be at-