Elimination of the relative buoyancy of low-salinity upper waters allowed deep heat to be brought to the surface (so that the polynya can be selfsustaining; Martinson et al., 1981). The event left its mark on the ocean with anomalous cold, low-salinity deep water, which found its way into the deep circulation, as witnessed in the Argentine Basin farther north.
The archive of hydrographic data from throughout the world ocean gives us an instrumental record of climate change that integrates over time and space: the ocean dominates (by a factor of about 103) the heat capacity of the ocean-atmosphere system, and its storage of heat anomalies also greatly exceeds the land surface. Levitus et al. (1999) (also see Levitus et al., 2001 and Barnett et al., 2001) described the integrated ocean heat content from 1950 to 2000 (Figure 2.11). In every ocean, an overall warming during the last 50 years is evident. But in a hiatus 1978–1988, all the ocean basins other than the South Atlantic cooled substantially. The temporary cooling in Pacific and Indian Oceans (both north and south) amounted to about half the net rise (2 × 1023J) over the 50 years. The abrupt cooling is mostly in the top 300 m of the water column. The anomalous air-sea heat flux during the event averaged about 1 watt m–2 which is several times as large as the half-century-long warming (0.31 watt m–2). The cooling event fails to resemble the well-known dip in atmospheric surface temperatures between roughly 1940 and 1970 in timing or duration. Heat is mercurial, and global budgets like this can be surprisingly responsive to localized forcing. Possible contributors to the downward lurches are the eruptions of the volcanoes Agung in 1963, El Chichon in 1982, and Pinatubo in 1991 (Levitus, 2001).
Instrumental records show preferred modes of behavior of the earth’s climate system. Furthermore, those modes might be coupled not only in time (such as the unknown relation between the few-year ENSO and fewdecade PDO variability), but also in space. Although the principal modes of variability at high latitude (the AO/NAO, and related southern hemisphere annular mode) have some independence from ENSO, connections have been proposed. For example, Hoerling et al. (2001) suggested, comparing observations and coupled-model simulations, that the average tropi-