North Atlantic Deep Water cools the southern hemisphere

Abstract

A standard explanation for coupling climate variations in the northern and southern hemispheres involves fluctuations in North Atlantic Deep Water (NADW) production. However, I suggest that the “NADW‐Antarctic” connection may work opposite to that conjectured by many investigators; that is, when NADW production rates are high, southern hemisphere temperatures decrease rather than increase. The revised interpretation is consistent with observations and ocean modeling studies which demonstrate that, although upwelling of relatively warm NADW water around Antarctica promotes sea ice meltback, a second and more important negative feedback is also operating. In order to conserve volume, southward export of NADW across the equator is accompanied by import of an equivalent volume of considerably warmer water from shallower oceanic layers in the South Atlantic. The southern hemisphere loses heat as a result of this exchange. The hemispherically averaged net heat loss may be as high as 4 W/m², an amount comparable to a CO 2 doubling. It is suggested that this more comprehensive view of the role of NADW may explain both decadal‐scale variations in South Atlantic sea surface temperatures in this century and two significant problems in Pleistocene climatology: why southern hemisphere temperatures decreased before CO 2 levels decreased at the end of the last interglacial and why southern hemisphere temperature changes precede changes in northern hemisphere ice volume. It is shown that when NADW production was reinitiated during the last interglacial (120,000 B.P.), high‐latitude southern hemisphere temperatures decreased. The estimated magnitude of altered southern hemisphere heat export is comparable to the ice‐age CO 2 signal and may be able to account for the observed cooling even when CO 2 levels were high. When cast into a frequency domain framework, this interpretation may also help explain why southern hemisphere temperatures lead global ice volume changes.