Climate Change 2001:
Working Group I: The Scientific Basis
Other reports in this collection

2.2.2.5 Sub-surface ocean temperatures and salinities

While the upper ocean temperature and salinity are coupled to the atmosphere on diurnal and seasonal time-scales, the deep ocean responds on much longer time-scales. During the last decade, data set development, rescue, declassification and new global surveys have made temperature and salinity profile data more readily available (Levitus et al., 1994, 2000a).

Global

Figure 2.11: Time-series for 1948 to 1998 of ocean heat content anomalies in the upper 300 m for the two hemispheres and the global ocean. Note that 1.5 x 1022 J equals 1 watt-year-m-2 averaged over the entire surface of the earth. Vertical lines through each yearly estimate are ± one standard error (Levitus et al., 2000b).

Levitus et al. (1997, 2000b) made annual estimates of the heat content of the upper 300 m of the world ocean from 1948 through to 1998 (Figure 2.11). The Atlantic and Indian Oceans each show a similar change from relatively cold to relatively warm conditions around 1976. The Pacific Ocean exhibits more of a bidecadal signal in heat storage. In 1998, the upper 300 m of the world ocean contained (1.0 ± 0.5) x1023 Joules more heat than it did in the mid-1950s, which represents a warming of 0.3 ± 0.15°C. A least squares linear regression to the annual temperature anomalies from 1958 to 1998 gives a warming of 0.037°C/decade. White et al. (1997, 1998b) computed changes in diabatic heat storage within the seasonal mixed layer from 1955 to 1996 between 20°S and 60°N and observed a warming of 0.15 ± 0.02°C or 0.036°C/decade.

Extension of the analysis to the upper 3,000 m shows that similar changes in heat content have occurred over intermediate and deep waters in all the basins, especially in the North and South Atlantic and the South Indian Oceans. The change in global ocean heat content from the 1950s to the 1990s is equivalent to a net downwards surface heat flux of 0.3 Wm-2 over the whole period.

Pacific
The winter and spring mixed-layer depths over the sub-tropical gyre of the North Pacific deepened 30 to 80% over the period 1960 to 1988 (Polovina et al., 1995). Over the sub-polar gyre, mixed-layer depths reduced by 20 to 30% over the same period. The surface layer of the sub-polar gyre in the north-east Pacific has both warmed and freshened, resulting in a lower surface density (Freeland et al., 1997). Wong et al. (1999) compared trans-Pacific data from the early 1990s to historical data collected about twenty years earlier. The changes in temperature and salinity are consistent with surface warming and freshening at mid- and higher latitudes and the subsequent subduction (downward advection) of these changes into the thermocline. From 1968/69 and 1990/91, the South Pacific waters beneath the base of the thermocline have cooled and freshened (Johnson and Orsi, 1997); the greatest cooling and freshening of -1.0°C and 0.25, respectively, occurred near 48°S and were still observed at 20°S. All the deep water masses show a cooling and freshening at these high southern latitudes.

Arctic
Recent surveys of the Arctic Ocean (Quadfasel et al., 1993; Carmack et al., 1995; Jones et al., 1996) have revealed a sub-surface Atlantic-derived warm water layer that is up to 1°C warmer and whose temperature maximum is up to 100 dbar shallower than observed from ice camps from the 1950s to the 1980s, as well as from ice-breaker data in the late 1980s and early 1990s. Warming is greatest in the Eurasian Basin. Annual surveys of the southern Canada Basin since 1979 (Melling, 1998), have shown a warming and deepening lower Atlantic layer, the lower halocline layer cooling by 0.12°C and the upper halocline layer warming by 0.15°C. Steele and Boyd (1998) compared winter temperature and salinity profiles obtained over the central and eastern Arctic Basins from submarine transects in 1995 and 1993 with Soviet data collected over the period 1950 to 1989 (Environmental Working Group, 1997). They showed that the cold halocline waters cover significantly less area in the newer data. This is consistent with a decreased supply of cold, fresh halocline waters from the Pacific Shelf areas.

Atlantic
The sub-arctic North Atlantic exhibits decadal variability in both temperature and salinity (Belkin et al., 1998). Reverdin et al. (1997) found that the variability of salinity around the entire subarctic gyre for the period 1948 to 1990 was most prominent at periods of 10 years and longer, and extended from the surface to below the base of the winter mixed layer. This salinity signal was only coherent with elsewhere in the north-western Atlantic. A single spatial pattern explains 70% of the variance of the upper ocean salt content of the subarctic gyre, corresponding to a signal propagating from the west to the north-east. Reverdin et al. also found that fluctuations in the outflow of fresh water from the Arctic are associated with periods of greater or fewer than usual northerly winds east of Greenland or off the Canadian Archipelago.

North Atlantic deep waters begin as intermediate waters in the Nordic seas. These waters have freshened over the 1980s and 1990s (Bönisch et al., 1997). In addition, the absence of deep convection over the same period has caused Nordic Sea bottom waters to become warmer, saltier and less dense. The Faroes-Shetland Channel is the principal pathway between the north-east Atlantic and the Norwegian Sea and has been surveyed regularly since 1893 (Turrell et al., 1999). Unfortunately, the quality of the salinity measurements was poor from 1930 through to 1960. Since the mid-1970s, the intermediate and bottom waters entering the North Atlantic through the channel have freshened at rates of 0.02/decade and 0.01/decade, respectively. The decreased salinities have resulted in decreased water densities and a decrease of between 1 and 7%/decade in the transport of deep water into the North Atlantic.

In the Labrador Sea, winter oceanic deep convection was intense during the earlier 1990s, extending to deeper than 2,400 m in 1992 to 1994. This produced a Labrador Sea water mass colder, denser and fresher than has been observed over at least the last five decades (Lazier, 1995; Dickson et al., 1996).

Within the tropical and sub-tropical gyres of the North Atlantic, the deep and intermediate water masses are warming. Ocean station S (south-east of Bermuda, 32°17'N, 64°50'W) has been sampled bi-weekly since 1954. Joyce and Robins (1996) extended the hydrographic record from ocean station S back from 1954 to 1922 using nearby observations. They find an almost constant rate of warming over the 1,500 to 2,500 dbar layer of 0.05°C/decade over the 73-year period 1922 to 1995. This corresponds to a net downward heat flux of 0.7 Wm-2. Sections completed in 1958, 1985 and 1997 along 52°W and 66°W between 20°N to 35°N (Joyce et al., 1999) show a rate of warming of 0.06°C/decade, similar to that seen at Bermuda but averaged over a larger 1,700 m depth interval. Trans-Atlantic sections along 24°N in 1957, 1981 and 1992 show a similar warming between 800 and 2,500 m (Parrilla et al., 1994; Bryden et al., 1996). The maximum warming at 1,100 m is occurring at a rate of 0.1°C/decade. At 8°N between 1957 and 1993, Arhan et al. (1998) showed warming from 1,150 and 2,800 m with the maximum warming of 0.15°C at 1,660 m.

The Antarctic bottom water in the Argentine Basin of the South Atlantic experienced a marked cooling (0.05°C) and freshening (0.008) during the 1980s (Coles et al., 1996). The bottom waters of the Vema Channel at the northern end of the Argentine basin did not change significantly during the 1980s but warmed steadily during a 700-day set of current meter deployments from 1992 to 1994 (Zenk and Hogg, 1996).

The Indian Ocean
Bindoff and Mcdougall (2000) have examined changes between historical data collected mostly in the period 1959 to 1966 with WOCE data collected in 1987 in the southern Indian Ocean at latitudes 30 to 35°S. They found warming throughout the upper 900 m of the water column (maximum average warming over this section of 0.5°C at 220 dbar).



Other reports in this collection