nature 25 April 2002
Letters to Nature
Nature 416, 832 - 837 (2002)

Rapid freshening of the deep North Atlantic Ocean over the past four decades

BOB DICKSON*, IGOR YASHAYAEV†, JENS MEINCKE‡, BILL TURRELL§, STEPHEN DYE* & JUERGEN HOLFORT‡

* Centre for Environment, Fisheries, and Aquaculture Science, Lowestoft NR33 OHT, UK
† Bedford Institute of Oceanography, Dartmouth, Nova Scotia B2Y 4A2, Canada
‡ Institut fur Meereskunde, 22529 Hamburg, Germany
§ Marine Laboratory, PO Box 101, Aberdeen AB11 9DB, UK

The overflow and descent of cold, dense water from the sills of the Denmark Strait and the Faroe–Shetland channel into the North Atlantic Ocean is the principal means of ventilating the deep oceans, and is therefore a key element of the global thermohaline circulation. Most computer simulations of the ocean system in a climate with increasing atmospheric greenhouse-gas concentrations predict a weakening thermohaline circulation in the North Atlantic as the subpolar seas become fresher and warmer1-3, and it is assumed that this signal will be transferred to the deep ocean by the two overflows. From observations it has not been possible to detect whether the ocean's overturning circulation is changing, but recent evidence suggests that the transport over the sills may be slackening4. Here we show, through the analysis of long hydrographic records, that the system of overflow and entrainment that ventilates the deep Atlantic has steadily changed over the past four decades. We find that these changes have already led to sustained and widespread freshening of the deep ocean.

The Labrador Sea is a critical location for the Earth's climate system. In its upper and intermediate layers, annual-to-decadal variations in the production, character and thickness of its convectively formed mode water (Labrador Sea Water, LSW) directly determine the rate of the main Atlantic gyre circulation5. Through its deeper layers pass all of the deep and bottom waters that collectively form and drive the abyssal limb of the Atlantic meridional overturning circulation. Around its margins pass the two main freshwater flows from the Arctic Ocean to the North Atlantic (by way of the Canadian Arctic archipelago and the East Greenland shelf) that have been implicated in model experiments with a slowdown or shutdown of the meridional overturning circulation1-3.

Over the past 3–4 decades, the entire water column of the Labrador Sea has undergone radical change. From 1966 to 1992, the overall cooling of the water column of the Labrador Sea was equivalent to a loss of 8 W m-2 continuously for 26 years. Its freshening (Fig. 1a) was equivalent to mixing-in an extra 6 m of fresh water at the sea surface6. As a result, the steric height (caused by changes in the density of the water-column) in the central Labrador Sea was typically reduced by 8–10 cm. These are arguably the largest full-depth changes observed in the modern instrumental oceanographic record (see, for example, refs 7, 8, 9).

Figure 1 The origins of deep freshening in the Labrador Sea.
 
High resolution image and legend (150k)

In the upper and intermediate layers of the Labrador Sea to the limit of convection (2,300 m or so), the long cooling and freshening tendency is thought to reflect the sustained if non-steady evolution of the North Atlantic Oscillation (NAO) from its most extreme negative state in the instrumental record during winters of the 1960s to its most extreme and prolonged positive state in the early 1990s10. This evolution brought deepening convection11, 12 and ultimately formed LSW that was fresher, colder, deeper and denser13 than at any other time in the history of deep measurements there.

In the upper water column (about 0–2 km), these long-sustained cooling and freshening trends were halted or partly reversed in the late 1990s (see Fig. 1a, for data post-1995, at depths between 1,000 and 2,000 m). But the deep and abyssal layers of the Labrador Sea (2,300–3,300 m) show a remarkable freshening by more than -0.01 per decade over the past 3–4 decades; the beginnings of this continuing freshening were noted 20 years ago14, 15. Beyond the reach of deep convection, these changes cannot be due to local climate forcing: they probably reflect variations in the two dense overflows that renew and ventilate these deepest layers, or changes in the resident water masses that are entrained as these vigorous flows descend, or both. We consider here why the overflows freshened, and why that freshening was maintained downstream.

We propose that the ultimate source of freshening of both overflows lies in a large-scale and long-term freshening of the upper 1–1.5 km of the Nordic seas, immediately upstream. That multi-decadal change has itself been attributed to a variety of causes, many of which are also associated with the amplifying NAO: an increase in the direct export of sea ice from the Arctic Ocean16-18, an increase in precipitation along the Norwegian Atlantic Current by approximately 15 cm per winter through the extension of storm activity to the Nordic seas18, and a range of factors internal to the Nordic seas19. These include an increased freshwater supply from the East Icelandic Current, a narrowing of the salty Norwegian Atlantic Current towards the coast, and a deepening of the interface between Arctic Intermediate Water and Deep Water in the Norwegian Sea4, 19.

Although we cannot yet partition the recent freshening of the Nordic seas into its contributory components, it is clear from some of our longest observational records that the change has occurred over a sufficiently deep layer (1–1.5 km) to affect the hydrographic character of both dense overflows crossing the Greenland–Scotland Ridge at sill depths of 700–900 m. Century-long hydrographic sections monitoring the outflow of upper Norwegian Sea Deep Water and Arctic Intermediate Water through the Faroe–Shetland channel (FSC sill and FSC NSAIW in Fig. 2) confirm that salinities have decreased there almost linearly by 0.01 per decade since the mid 1970s20, and our more gappy record from the deepest part of the Denmark Strait sill (DS in Fig. 2, representing salinities at 500–550 m depth for temperatures below 0 °C) indicates a very similar net freshening over the same period.

Figure 2 Evidence of sustained and rapid freshening throughout the system of overflow and entrainment that ventilates the deep Atlantic.
 
High resolution image and legend (184k)

In addition to the change at both sills, Fig. 2 illustrates the salinity change at various intervals downstream where these overflows deepen, entrain, mix and spread, ultimately forming the deep and abyssal layers of the Labrador Sea (locations in Fig. 1b). All time series but one are to a common scale (the exception is the upper-ocean time series SEI, plotted at half-scale), and the bracketed values against each curve refer to the mean freshening rate over the common period 1965–2000. (Reflecting earlier usage (salinity defined in p.p.t.), a freshening rate of 0.014 per decade will be abbreviated here to '-14 p.p.m.').

On leaving the Faroe Bank channel, the vigorous plume of Iceland–Scotland Overflow Water (ISOW; freshening at between -7 and -14 p.p.m. per decade; Fig. 2) will broaden, slow and deepen through entrainment of 'resident' water masses at the head of the south Icelandic basin before continuing south at 1,500–2,500 m depth along the eastern flank of the Reykjanes Ridge. Constructing a salinity time series for the deep salinity maximum that marks the ISOW core in this location, we find evidence of a very similar freshening rate (RR; -9 p.p.m. per decade). The same rate, period and steadiness of freshening is maintained as the ISOW-derived layer passes at depth around the Irminger Sea (there known as North East Atlantic Deep Water, thus EIS NEADW, -13 p.p.m. per decade in Fig. 2) and into the Labrador Sea (LS NEADW, -12 p.p.m. per decade).

When we similarly construct salinity series for the coldest, deepest water overflowing the Denmark Strait sill (DS, freshening at -13 p.p.m. per decade, Fig. 2), for the fully entrained DSOW-derived water mass descending the Greenland slope in the western Irminger Sea (WIS DSOW, -15 p.p.m. per decade) and for the DSOW-derived water mass occupying the deepest layers of the Labrador Sea (LS DSOW, -12 p.p.m. per decade), we find an essentially similar trend and period of freshening.

Differences in detail between the freshening trends of the two overflows are understandable if we consider that the processes contributing to the freshening of the Nordic seas are uneven in distribution. For example, Fig. 2 shows an accelerated freshening at the Denmark Strait sill (DS), in the near-bottom overflow core off southeast Greenland (WIS DSOW) and in the abyssal Labrador Sea (LS DSOW) in the mid-1990s, followed by recovery to the common trend-line, which was not experienced in the NEADW layers close by (EIS NEADW and LS NEADW). We suggest that this reflects the peak 1–2 years earlier in the efflux of ice from the Fram Strait which would more directly affect the western overflow. To illustrate this, we superimpose in Fig. 2 an observed17 ice-flux series (scale inverted).

Because both overflows are considerably altered by entrainment and mixing as they deepen and spread to the Labrador Sea21, we might expect their freshening signal to dilute downstream, particularly along the 5,000-km path length of the eastern overflow. So the maintenance of a more-or-less uniform freshening rate along that path (Fig. 2) is surprising. We identify two locations in particular where that freshening would be reinforced and so maintained.

The first concerns the entrainment and mixing that take place as the eastern overflow leaves the Faroe Bank channel at >1 m s-1 to descend and decelerate through the resident water masses at the head of the south Icelandic basin22. Selecting the 500–1,000 m layer immediately southeast of Iceland as representative of the entrained water masses, Fig. 2 indicates that its mean freshening rate was more than twice as large as that of the overflow itself (-32 p.p.m. per decade; SEI in Fig. 2). As Fig. 3 explains, however, we need three end-members—all of them freshening with time—to account for the changing ISOW characteristics at the base of the Reykjanes Ridge. Selecting the water that overflows the Faroe–Shetland channel at sill depth (800 m) as the 'source' end-member, and the warm saline water mass at 700–800 m south of Iceland as the likely initial 'entrainment' (SEI700–800; Fig. 3), we find that we cannot form the product water mass that we encounter along the lower east side of the Reykjanes Ridge (RRISOW) or explain its freshening with time unless we add the cooling and freshening influence of a sizeable component of LSW. However, the freshening rate of LSW at source was also at least as large as that of the overflow (-13 p.p.m. per decade, 1965–2000, for LSW defined annually as the largest volumetric temperature–salinity class). Broadly speaking, Fig. 3 implies that the overflowing source water more than doubles in volume (about 2.5 times ) through entrainment and mixing along the margins of the south Icelandic basin, and that the other two end-members (SEI700–800 and LSW) contribute about equally to the fully entrained product.

Figure 3 Maintenance of the freshening rate of Iceland-Scotland Overflow Water (ISOW) on its long spreading-path to the Labrador Sea.
 
High resolution image and legend (54k)

The divergence of curves EIS and WIS with time in Fig. 2 suggests that the NEADW layer undergoes further re-freshening in waters of the western Irminger Sea. Both curves show the salinity change on the isopycnal that lies at the core of the NEADW layer in the Irminger Sea (sigma2 = 36.99–37.01), but close to Greenland where this isopycnal shoals to 1,500–2,250 m, the freshening rate (WIS; -22 p.p.m. per decade) is much greater than in the centre of the basin (EIS, -13 p.p.m. per decade, depth range 2,120–2,550 m). We infer from this that NEADW entering from the eastern basin is re-freshened as it approaches, shoals and then passes south against the Greenland slope, either by mixing with the underlying DSOW layer, or the overlying LSW layer, or both. Thus salinities of WIS and EIS would be similar around 1970 when LSW formation was too shallow to influence that isopycnal, but diverge into the 1990s as deepening convection in the Labrador Sea carried colder and fresher LSW to a sufficient depth to influence NEADW properties in the western Irminger Sea.

We conclude then that the freshening rate of the eastern overflow was maintained downstream by mixing with waters that were themselves freshening at an equal or greater rate. A similar conclusion applies to the overflow from the Denmark Strait as it descends the Greenland slope to the abyssal Labrador Sea. As its volume rapidly doubles by entrainment south of the sill21, its relatively uniform freshening rate is not simply the result of a short spreading-path but must reflect mixing with overlying waters of a similar or greater rate of freshening—LSW at -13 p.p.m. or WIS NEADW at -22 p.p.m. per decade.

We have described a widespread, sustained, rapid and surprisingly uniform freshening of the deep and abyssal North Atlantic, south of the Greenland–Scotland Ridge, over the past four decades. Because the freshening affects both overflows and their spreading pathways downstream, it seems to confirm our supposition that subarctic change can be rapidly transferred to the deep Atlantic, and has already directly affected the abyssal limb of the Atlantic meridional overturning circulation. Other observations confirm that the deep and abyssal freshening we describe has already passed equatorward along the North American seaboard in the Deep Western Boundary Current (W.M. Smethie, personal communication).

The question remains as to whether and to what extent these changes or these transfers reflect the onset of global change. With "new and stronger evidence"3 of anthropogenic warming, coupled climate models seem to be reaching some kind of consensus that a slowdown of North Atlantic Deep Water production and of the meridional overturning circulation will be one outcome. However the issue of whether such effects are yet evident in our ocean time series remains open. Hansen et al.4 have been able to couple a moderately long, modern set of direct flow measurements to a half-century of frequent hydrography at OWS M to provide evidence of a 20% decrease in the coldest and densest part (T < 0.3 °C, sigmat >28.0, sigmat is in situ density) of the overflow from the Faroe Bank channel since 1950, and such measurements as we have from the Denmark Strait give no sign of any compensating increase in the cold dense outflow through that sill23. In the overflow hydrography that we report here, we provide the companion finding that a means exists of transferring the 'signal' of high-latitude climate change to the deep and abyssal headwaters of the global thermohaline circulation. However, the question of ultimate cause really hinges on whether the long amplification of the NAO, which so pervades our recent records of ocean circulation and hydrography, is itself attributable to global change; that issue is at present unresolved12, 24.

Received 15 October 2001;accepted 21 March 2002

------------------

References
1. Rahmstorf, S. & Ganopolski, A. Long-term global warming scenarios computed with an efficient coupled climate model. Clim. Change 43, 353-367 (1999) | Article | ISI |
2. Delworth, T. L. & Dixon, K. W. Implications of the recent trend in the Arctic/North Atlantic Oscillation for the North Atlantic thermohaline circulation. J. Clim. 13, 3721-3727 (2000) | ISI |
3. IPCC Climate Change 2001: The Scientific Basis (eds Houghton, J. T. et al.) (Cambridge Univ. Press, Cambridge, 2001)
4. Hansen, B., Turrell, W. R. & Østerhus, S. Decreasing overflow from the Nordic seas into the Atlantic Ocean through the Faroe-Shetland Channel since 1950. Nature 411, 927-930 (2001) | Article | PubMed | ISI |
5. Curry, R. & McCartney, M. S. Ocean gyre circulation changes associated with the North Atlantic Oscillation. J. Phys. Oceanogr. 31, 3374-3400 (2001) | ISI |
6. Lazier, J. R. N. in Natural Climate Variability on Decade-to-Century Time Scales (eds Martinson, D. G. et al.) 295-304 (National Academy Press, Washington DC, 1995)
7. Verduin, J. & Quadfasel, D. in European Sub-Polar Ocean Programme II, Final Scientific Report (ed. Jansen, E.) A1, 1-11 (Univ. Bergen, Bergen, 1999)
8. Lascaratos, A., Roether, W., Nittis, K. & Klein, B. Recent changes in deep water formation and spreading in the eastern Mediterranean Sea. Prog. Oceanogr. 44, 5-36 (1999) | ISI |
9. Gordon, A. Weddell Deep Water variability. J. Mar. Res. 40, 199-217 (1982)
10. Hurrell, J. W. Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science 269, 676-679 (1995) | ISI |
11. Dickson, R. R., Lazier, J., Meincke, J., Rhines, P. & Swift, J. Long-term co-ordinated changes in the convective activity of the North Atlantic. Prog. Oceanogr. 38, 241-295 (1996) | Article |
12. Hurrell, J. W. & Dickson, R. R. in Ecological Effects of Climate Variations in the North Atlantic (eds Stenseth, N. C., Ottersen, G., Hurrell, J. W. & Belgrano, A.) (Oxford Univ. Press, in the press)
13. Sy, A. et al. Surprisingly rapid spreading of newly formed intermediate waters across the North Atlantic Ocean. Nature 386, 675-679 (1997) | ISI |
14. Brewer, P. G. et al. A climatic freshening of the deep Atlantic north of 50°N over the past 20 years. Science 222, 1237-1239 (1983) | ISI |
15. Swift, J. H. Climate Processes and Climate Sensitivity (eds Hansen, J. E. & Takehashi, T.) 39-47 (AGU Geophysical Monograph 29, American Geophysical Union, Washington DC, 1984)
16. Vinje, T., Nordlund, N. & Kvambekk, A. Monitoring ice thickness in Fram Strait. J. Geophys. Res. 103, 10437-10449 (1998) | ISI |
17. Vinje, T. Fram Strait ice fluxes and atmospheric circulation, 1950-2000. J. Clim. 14, 3508-3517 (2001) | ISI |
18. Dickson, R. R. et al. The Arctic Ocean response to the North Atlantic Oscillation. J. Clim. 13, 2671-2696 (2000) | ISI |
19. Blindheim, J. et al. Upper layer cooling and freshening in the Norwegian Sea in relation to atmospheric forcing. Deep-Sea Res. I 47, 655-680 (2000) | ISI |
20. Turrell, W. R., Slesser, G., Adams, R. D., Payne, R. & Gillibrand, P. A. Decadal variability in the composition of Faroe-Shetland Channel bottom water. Deep-Sea Res. I 46, 1-25 (1999) | ISI |
21. Dickson, R. R. & Brown, J. The production of North Atlantic Deep Water: Sources, rates and pathways. J. Geophys. Res. C 99, 12319-12341 (1994) | ISI |
22. Hansen, B. & Østerhus, S. North Atlantic-Nordic seas exchanges. Prog. Oceanogr. 45, 109-208 (2000) | ISI |
23. Girton, J. B., Sanford, T. B. & Käse, R. H. Synoptic sections of the Denmark Strait Overflow. Geophys. Res. Lett. 28, 1619-1622 (2001) | ISI |
24. Thompson, D. W. J., Wallace, J. M. & Hegerl, G. C. Annular modes in the extratropical circulation Part II: Trends. J. Clim. 13, 1018-1036 (2000) | ISI |