For an adequate interpretation of the temperature and salinity trends, we have to perform a water mass analysis. This analysis is also used to examine the processes behind the observed temporal changes.
Fig. 4 shows the temporal evolution of the decay-corrected tritium/theta correlation, as well as the theta/salinity, and the CFC11/theta correlations. Also shown are the properties of possible sources of the deep water (2000m depth to the bottom) in the 1980s. The deep water in the central Greenland Sea is formed by a mixture of water convected from the upper water column, presented as Greenland Sea Surface Water (GSSW), and water advected from the deep Norwegian Sea (Norwegian Sea Deep Water, NSDW) and the Arctic Ocean (Eurasian Basin Deep Water, EBDW; Heinze et al. [1990], Aagaard et al. [1991], Rhein [1991], Schott et al. [1993], Bönisch and Schlosser [1995]). As mentioned earlier, deep convection occurs only during winter (Visbeck et al. [1995]). Because the time series does not include the seasonal cycle due to lack of data, we use the theta/S characteristics observed by Schott et al. [1993] during deep convection in the Greenland Sea to characterize winter conditions of the surface water. Figs. 4a and b indicate that the water found in the lower water column of the central Greenland gyre had higher transient tracer concentrations than all the other deep waters in this region.
The TU81/theta plot is shown in Fig. 4a. In 1950, the tritium concentration of all waters was practically 0 (about 0.2 TU in surface waters (Dreisigacker and Roether [1978]) and 0.03 TU in the deep water (Bönisch and Schlosser [1995])). In the surface waters, the tritium concentrations increased rapidly during the 1950s and 1960s due to the atmospheric nuclear weapon tests, and declined since. Since the deep water of the central Greenland gyre contained a significant component of recently ventilated near-surface water, its tritium concentration followed the increase of the surface waters, although at a lower rate. This is evident from Fig. 4a for the time between 1950 and 1981. For this period, the TU81/theta ratio of the deep water in the Greenland Sea does not fall on a mixing line between surface water and the deep waters in the 1980s, because the initial (1950, marked in Fig. 4a) tritium concentration of the deep water was practically zero and has to be taken into account as one endmember of the mixing diagram for the following reasons. Since water with high tracer concentrations mixed into the deep water over time, even the decay-corrected tritium concentration of the deep water which was tracer free in 1950 changed continuously. The tracer concentrations increased due to the addition of near surface water with high tracer cooncentrations. The initially tracer-free deep water from 1950 has therefore to be taken into account as one endmember of the mixing diagram. The tritium concentration in the deep water never reached equilibrium with the supplying waters.
Although the tritium concentration of the surface water was still much higher (about 5 TU, see Fig. 2d) than that of the deep water at the beginning of the 1980s, the decay-corrected tritium increase in the deep water nearly stopped at the same time, as its temperature started to increase. Fig. 4a suggests that this is caused by increased contributions of NSDW and EBDW. During the 1980s and early 1990s, the CFC11/theta ratio (Fig. 4b) showed the same behaviour as the TU81/theta ratio. The dashed lines in Figs. 4a and b indicate the proposed evolution of the tracer/theta ratios, if no water with higher transient tracer concentrations contributed to the formation of the deep water during the 1980s and 1990s. Since the properties of the deep water did not follow this line, we conclude that water with higher transient tracer concentrations continued to contribute to the deep water in the central Greenland Sea between 1980 and 1994. Since the deep water of the central Greenland Sea has higher transient tracer concentrations than the deep waters of the adjacent basins contributing to its formation (Norwegian Sea and Eurasian Basin), this water can only originate from the overlying water column of the Greenland gyre. The contribution of such water was predicted by the tracer balances published by Schlosser et al. [1991] and Bönisch and Schlosser [1995]. However, this water does not necessarily have to originate directly from the surface as implied by their model, it might well come from intermediate depths (see discussion below). This simple box model by Bönisch and Schlosser [1995] we used before for interpretation of the evolution of the tracer data of the deep water can not distinguish, if near surface water or intermediate water renewed the deep water.
The theta/S characteristics of the deep water changed little between 1972 and 1981 (Fig. 4c). The theta/S characteristics were in a quasi-steady state with the source waters. The dashed lines indicate the possible mixing domaine between GSSW, EBDW, and NSDW. In the 1980s, the theta/S characteristics moved towards those of EBDW and NSDW, i.e., the deep water became warmer and more saline. The relative contribution of GSSW decreased, causing the relative contributions of EBDW and NSDW to the deep water of the Greenland Sea to increase. Fig. 4c also shows the evolution of the theta/S characteristics of the most dense (sigma2 > 37.45) part of the overlying intermediate water of the central Greenland Sea. The temporal evolution of the theta/S characteristics of the deep water in the 1980s and 1990s could also be caused by mixing with water from the intermediate layer, EBDW, and NSDW. By the end of 1994, the theta/S characteristics of the deep water were still changing, no new quasi-steady state had been reached.
All data indicate a change in the process of deep water formation in the Greenland Sea at around 1980. The transient tracers clearly demonstrate that water with relatively high transient tracer concentrations still contributed in the deep water formation in the 1980s and 1990s, but at a much lower rate than in the period before 1980. This water must originate from the upper or intermediate water column of the Greenland Sea.