Recent variability of the North Atlantic thermohaline circulation inferredfrom surface heat and freshwater fluxes

An annual-mean surface-forced component of the North Atlantic thermohaline circulation (THC) in density space, psi(surf)(Theta, sigma), is diagnosed f rom observed surface heat and freshwater fluxes. The climatological mean of psi(surf) over 1980-97 indicates steady overturning rates for those water masses that are principally formed through surface buoyancy forcing: simila r to 8 Sv of Greenland Sea Deep Water (GSDW), similar to 4 Sv of Labrador S ea Water (LSW), and similar to 4 Sv of 18 degrees Water (STMW). Obtained fo r each year over 1980-97, psi(surf) reveals interannual-to-decadal variabil ity in the renewal and overturning of these water masses. The total surface -forced overturning rate varies in the range 10-20 Sv, reaching peak values in 1989-90 and generally decreasing over the period 1990-97. More dramatic changes are apparent in the overturning of different water masses. The ove rturning of STMW exhibits large-amplitude interannual variability. A backgr ound positive trend over 1980-97 is dominated by variance of 13.4 Sv. By co ntrast, the overturning rates of LSW and GSDW vary more slowly, on decadal timescales. LSW overturning rates increase from near zero in the early 1980 s to a maximum of similar to 10 Sv in 1990 and return to near zero by 1997. Seemingly in antiphase, GSDW overturning rates decline from a peak rate of similar to 11 Sv in 1981 to a minimum of similar to 3.5 Sv in 1991, and th en increase somewhat up to 1997. The variability in these overturning rates is related to changes in the Nor th Atlantic oscillation (NAO) and the Arctic oscillation (AO) over 1980-97. Correlation coefficients between total overturning rate and three versions of the NAO index are significantly positive at a 99% confidence level. Als o highly significant is a correlation of LSW overturning and the AO index, and an anticorrelation between overturning of LSW and GSDW. The overturning of STMW is more weakly (and not significantly) anticorrelated with NAO ind ices, although there is highly significant anticorrelation between interann ual changes in STMW overturning and one version of the NAO index. Stronger LSW (GSDW) overturning in years when NAO indices are relatively high (low) is consistent with recent hydrographic evidence. The extent to which these relationships support a hypothesis that the NAO drives interannual-to-decad al variability in the THC, through wintertime atmospheric buoyancy forcing of the ocean, is discussed.