Multidecadal warming of Antarctic waters

See allHide authors and affiliations

Science  05 Dec 2014:
Vol. 346, Issue 6214, pp. 1227-1231
DOI: 10.1126/science.1256117


Decadal trends in the properties of seawater adjacent to Antarctica are poorly known, and the mechanisms responsible for such changes are uncertain. Antarctic ice sheet mass loss is largely driven by ice shelf basal melt, which is influenced by ocean-ice interactions and has been correlated with Antarctic Continental Shelf Bottom Water (ASBW) temperature. We document the spatial distribution of long-term large-scale trends in temperature, salinity, and core depth over the Antarctic continental shelf and slope. Warming at the seabed in the Bellingshausen and Amundsen seas is linked to increased heat content and to a shoaling of the mid-depth temperature maximum over the continental slope, allowing warmer, saltier water greater access to the shelf in recent years. Regions of ASBW warming are those exhibiting increased ice shelf melt.

Bringing up the problem of ice shelf melting

Warm water intruding from below is heating up the ocean that covers the continental shelf of Antarctica. Schmidtko et al. report that Circumpolar Deep Water has been warming and moving further up onto the shelf around Antarctica for the past 40 years, causing higher rates of ice sheet melting (see the Perspective by Gille). These observations need to be taken into account when considering the potential for irreversible retreat of parts of the West Antarctic Ice Sheet.

Science, this issue p. 1227; see also p. 1180

The Antarctic ice sheet is the largest reservoir of terrestrial ice and is a significant contributor to sea level rise in a warming climate (1). Massive ice shelf disintegration and rapid acceleration of glacial flow have occurred in recent decades (2) or are potentially looming (3). These events are generally linked to enhanced basal melt (4, 5), which reduces buttressing and accelerates glacier flow. An increase in basal melt may be linked to stronger sub–ice shelf circulation of Circumpolar Deep Water (CDW). It is not known whether changes in the delivery of warm water to the underside of the ice shelf are caused by increased heat content, increased volume flux responding to changes in wind and buoyancy forcing, or some combination of the two (6, 7). Here, we focus on documenting long-term temperature and salinity changes in ocean properties over the continental shelf and slopes.

We refer to the water occupying the sea floor on the Antarctic continental shelf as Antarctic Continental Shelf Bottom Water (ASBW) and the temperature minimum layer, representing the remnant of the winter mixed layer, as Winter Water (WW). Atmospheric processes, adjacent water masses, ice shelf and continental freshwater fluxes, and bathymetry-dependent cross-shelf water exchange at the shelf break determine the hydrographic properties of ASBW. Hence, ASBW is a mixture of CDW, WW, and waters originating from continental runoff, ice shelf melting, or sea ice formation. Temporal changes in ASBW may represent changes either in formation processes or in properties of source waters. ASBW has freshened in the Ross Sea (8) and northwest Weddell Sea (9), with the former generally linked to enhanced upstream ice shelf melt. Recent research has identified large-scale warming of both CDW north of 60°S (10) and Antarctic Bottom Water (AABW) (11). However, statistically significant long-term warming of ASBW has been observed only in localized regions (12), and trends in CDW properties over the Antarctic continental slope have not previously been described in detail.

Using a comprehensive compilation of observational data sets (table S1) [(13) and supplementary materials], we found that temporal trends in ASBW temperature and salinity have a distinct regional pattern (Fig. 1). The Bellingshausen Sea and Amundsen Sea shelves show significant warming (0.1° to 0.3°C decade–1; Fig. 1, C and E) and salinification (0.01 to 0.04 g kg–1 decade–1; Fig. 1, D and F) since the 1990s, when sufficient observations became available (fig. S2). The Ross Sea is freshening (–0.027 ± 0.012 g kg–1 decade–1), and the western Weddell Sea reveals a slight cooling (–0.05° ± 0.04°C decade–1) and freshening (–0.01 ± 0.007 g kg–1 decade–1); older measurements allow these trends to be calculated from the 1970s. The Cosmonaut Sea is freshening (–0.05 ± 0.03 g kg–1 decade–1). These trends agree with the available but limited regional observations (12). The previously documented regional Ross Sea freshening (8) (–0.03 g kg–1 decade–1) is within the range of the locally mapped values obtained here (–0.01 to –0.05 g kg–1 decade–1; Fig. 1D). Focusing only on data since the mid-1980s (Fig. 1F) produces a more pronounced average Ross Sea freshening (–0.05 ± 0.02 g kg–1 decade–1). The freshening trend in the western Weddell Sea shelf changes sign when only data from the 1990s are used (Fig. 1E and fig. S10). Freshening is observed in the northwestern Weddell Sea (Fig. 1D), although the trend doubles and is similar to reported values when only data from the late 1980s to the mid-2000s are considered (9).

Fig. 1 Temporal means, linear trends, and multiannual variability in Antarctic Continental Shelf Bottom Water (ASBW).

(A to D) Conservative temperature [(A) and (C)] and absolute salinity [(B) and (D)] at the seabed for depths shallower than 1500 m for the period 1975 to 2012 are shown in terms of temporal means [(A) and (B)] and linear trends [(C) and (D)]. Trends not significantly different from zero are hatched. (E and F) Multiannual variability as shown by 5-year median properties since 1975, interquartile ranges, and median trends for selected areas. Abbreviations for surrounding seas: BS, Bellingshausen Sea; AS, Amundsen Sea; RS, Ross Sea; CS, Cosmonaut Sea; WS, Weddell Sea (excluding the Antarctic Peninsula).

Temperature trends are typically uniform across the width of the shelf up to the shelf break. In contrast, salinity trends show a cross-shelf gradient with freshening generally intensified close to the coast (Fig. 1D); this is particularly pronounced over the Ross Sea and Weddell Sea continental shelves. A cross-shelf salinity trend gradient points to physical processes that induce a larger change in salinity than in temperature near the continent. For example, changes in surface meltwater runoff and changes in sea ice melting or freezing would be seen primarily in salinity, whereas changes in ocean-induced ice shelf basal melting would be seen in both temperature and salinity; atmospheric cooling would be seen in temperature but not salinity, unless sea ice is formed. Although changes in meltwater runoff and sea ice formation are consistent with the observed temperature and salinity trends, our analysis cannot distinguish between these mechanisms. Lateral variations of the trend vary according to the time period analyzed. The shoreward intensification of the freshening is significantly stronger when all data are used than when only data since 1990 are used (Fig. 1D and fig. S10). This difference may originate in poor data coverage from recent decades (figs. S1 and S2) or may indicate multiannual variability in any combination of processes described above (Fig. 1, E and F).

We next assessed the origins of ASBW property trends (Fig. 1, C and D) by analyzing changes in the properties and the depth of source water masses: the temperature maximum layer approaching the continental shelf break (CDW core; Fig. 2) and the temperature minimum layer (WW; Fig. 3). CDW shows significant warming and shoaling in most regions around Antarctica (Fig. 2, D and F), with trends on the order of 0.1°C decade–1 and –30 m decade–1, respectively. These values are similar to those of isopycnal heave found farther north (10) and at depth (14). There are few significant changes in CDW salinity close to Antarctica; the most pronounced freshening is observed in the Antarctic Circumpolar Current (ACC) in the Indian Ocean sector, likely owing to variations in front location in regions of large meridional gradients (Fig. 2, B and E).

Fig. 2 Temporal means and linear trends in Circumpolar Deep Water (CDW).

(A to F) Conservative temperature [(A) and (D)], absolute salinity [(B) and (E)], and core depth [(C) and (F)] of CDW between 1975 and 2012 are shown in terms of temporal means [(A) to (C)] and linear trends [(D) to (F)]. Trends not significantly different from zero are hatched. Fronts (28) related to the ACC are shown as black lines. Only regions where the ocean depth exceeds 1500 m and the underlying CDW is cooler than 2.8°C are shown. (G) Trends and data for selected locations.

Fig. 3 Temporal means and linear trends in Winter Water (WW).

(A to F) Conservative temperature [(A) and (D)], absolute salinity [(B) and (E)], and core depth [(C) and (F)] of WW between 1975 and 2012 are shown in terms of temporal means [(A) to (C)] and linear trends [(D) to (F)]. Trends not significantly different from zero are hatched. Fronts (28) related to the ACC are shown as black lines. Only regions where the ocean depth exceeds 1500 m and the underlying CDW is cooler than 2.8°C are shown.

In addition to regional changes in water mass properties, changes in the depth of the CDW core also show spatial variability. Deepening of the CDW core only occurs along the northern limb of the Weddell Gyre and in the Cosmonaut Sea (Fig. 2F). Areas where the CDW core deepens tend to be correlated with cooling. The strongest shoaling of CDW is found in the Bellingshausen and Amundsen seas, where temporal shoaling rates locally exceed –50 ± 18 m decade–1. Around the Antarctic margins, the core of CDW typically deepens approaching the continental shelf, as in the Weddell and Ross gyres (Fig. 2C and Fig. 4B). However, in the Bellingshausen and Amundsen seas, the core of CDW slopes upward approaching the shelf (Fig. 4A).

Fig. 4 Schematic differences of shelf water masses.

(A) Schematic of the Amundsen and Bellingshausen seas with warm water at the bottom of the water column on the shelf. (B) Schematic of the shelf in the Ross and Weddell seas, with a strong Antarctic Slope Current.

The other water mass present over the shelf and slope is WW (Fig. 3). WW is freshening along the southwestern limb of the Ross Gyre and in the western Indian Ocean sector, while WW is becoming saltier in the northwestern Weddell Sea (Fig. 3E). Significant shoaling of WW is present in the eastern Indian Ocean sector and Amundsen Sea (Fig. 3F). Origins of the salinity changes in WW are likely to differ among regions, including changes to sea ice concentration caused by ice motion trends (15) and changes to sea ice concentration caused by increased accumulation of ice shelf meltwater (16).

The trends in Figs. 1 and 2 show a link between changes in the properties of ASBW and those of CDW over the continental slope. The CDW thermal structure may be broadly categorized into two regimes, shown schematically in Fig. 4: CDW (i) sloping upward or (ii) sloping downward toward the shelf break. The Weddell and Ross gyres are subject to a large-scale cyclonic wind stress, leading to strong easterly winds over the shelf break that depresses isotherms (Fig. 4B). Cyclonic wind patterns are also found over the Amundsen and Bellingshausen seas, but the center of the low-pressure systems is found over the continental shelf. Thus, in these regions, weaker easterly or even westerly winds are found at the shelf break, causing upward-tilting CDW and enhanced onshore intrusions (17, 18) (Fig. 4A). Furthermore, the Bellingshausen and Amundsen seas do not export a significant volume of AABW, and a bottom-intensified along-slope Antarctic Slope Current does not form (Fig. 4A).

Although warming and shoaling of CDW is present around Antarctica, the regions that have experienced the most intense warming and shoaling are associated with areas where CDW has greater access to the shelf break (Fig. 4A). Additionally, in all regions where CDW both slopes upward toward the shelf break and is shoaling over time, ASBW temperatures are rising. This link suggests that changes occurring to CDW offshore in the Bellingshausen and Amundsen seas (Fig. 2, D and F) are more likely to be transmitted onto the shelf and to be reflected in regional ASBW properties. By contrast, in regions where CDW slopes downward toward the shelf break, offshore trends in CDW properties show no correlation to bottom water masses on the continental shelves (e.g., near the Ross, Amery, Filchner, and Ronne ice shelves) (Fig. 4B). Thus, the segments of continental shelf in the Weddell and Ross seas are in effect shielded from offshore changes in CDW. The one possible exception, though, is the intense shoaling of CDW in the southern Weddell Sea, which raises questions about the future stability of ASBW temperatures in front of the Filchner and Ronne ice shelves (19) (Fig. 2, F and G).

Recent dynamical studies have highlighted the importance of surface forcing on links between offshore and onshore properties. In Ross and Weddell gyre idealized model configurations, surface forcing by the polar easterlies controls onshore transport of WW and CDW as well as the characteristic slope of mid-depth density surfaces toward the shelf break (17). In the Amundsen Sea, offshore atmospheric anomalies modify the thermal gradient across the continental shelf and under the ice shelf (18). Buoyancy forcing may also be critical to long-term trends. A simple balance model (6) suggests that regional properties are largely set by atmospheric patterns of heating and cooling, but changes in ASBW may depend on the penetration depth of convection. Finally, over the continental slope, mesoscale variability plays a critical role in setting temperature distributions and is sensitive to changes in wind forcing (20). It is critical to note that these dynamical processes occur at scales that are necessarily smoothed by our analysis. However, the large-scale trends point to key regions that require more intense analysis through high-resolution modeling with improved bathymetric data to determine the mechanistic cause of these changes.

Because ice shelf melting and ice sheet collapse depend on the delivery of warm water by ocean currents, ice sheet predictability requires accurate estimates of oceanic heat transport. Shoaling CDW has the potential to significantly increase this heat transport and must be represented in climate models for future ice shelf predictions. Current climate models still have limitations in Southern Ocean mixed layer processes (21) and are thus unlikely to represent WW and the underlying CDW shoaling correctly. In addition to onshore transport, capturing changes in exported AABW properties is also critical for modeling the global circulation. The changes in ASBW salinity taking place in AABW formation regions such as the Weddell and Ross seas (Fig. 1D) resemble the observed freshwater changes in AABW around Antarctica (14, 22). AABW has freshened in the Indian (22) and Pacific sectors and to a lesser extent in the Atlantic sector (14). Temperature changes seen in AABW (11, 23) are not apparent in ASBW (Fig. 1C).

The observed trends in water mass properties have a number of implications. Changes in ASBW may influence Southern Ocean ecosystems through the distribution of salps and krill (24), although the impact of warming oceans on Antarctic krill is still debated (25) because of their spatially diverse life cycles. We expect that megafaunal communities (26) on the shelf—in particular, in the Bellingshausen Sea and Amundsen Sea sectors—have experienced warmer and more frequent warm-water intrusions of CDW. Further shoaling of CDW will increase both the number of ice shelves and the area of each ice shelf that are influenced by offshore warming. This has the potential to increase the spatial extent of enhanced basal melt and to lead to irreversible retreat of a portion of the West Antarctic Ice Sheet, which will have an impact on global sea level (3, 27).

Supplementary Materials

Figs. S1 to S11

Table S1

References (29, 30)

References and Notes

  1. Acknowledgments: All data used in this manuscript are publicly available; data sources are detailed in the supplementary materials. We thank all officers, crews, and scientists involved in collecting and calibrating the data in the often harsh Southern Ocean environment, as well as everyone involved in making the data publicly available. We thank three anonymous reviewers for helpful discussions and feedback. Supported by the German Federal Ministry for Education and Research (BMBF) project MiKlip (S.S.); NERC Antarctic Funding Initiative research grant GENTOO NE/H01439X/1 (S.S. and K.J.H.); a Grant-in-Aid for Scientific Research of the Ministry of Education, Culture, Sports, Science and Technology, Japan, and a Daiwa Foundation Small Grant (S.A.); and NSF award OPP-1246460 (A.F.T.).
View Abstract

Navigate This Article