We make three large assumptions in constructing the abyssal basin estimates of heat gain and SLR below 4000 m and the deep Southern Ocean estimates of these quantities from 1000 to 4000 m. First, we assume that dθ/dt within a given basin or the Southern Ocean is relatively consistent over the pressure intervals considered. This assumption is often supported by examination of pressure–latitude/longitude sections of dθ/dt estimates, wherein dθ/dt is often roughly vertically uniform on deep pressure horizons, and more so below the sill depth of a basin. For example, P18 crosses four basins: the Amundsen–Bellingshausen, Chile, Peru, and central Pacific (Fig. 4). The Chile and Peru Basins both show uniform (but small) warming below 3000 m and exhibit little variability, while the Amundsen–Bellingshausen Basin shows stronger warming but also higher variability. Second, we assume that the tracklines used to estimate warming rates in each basin or region are representative samples of that basin. The validity of this assumption is dependent on the spatial coverage of the tracklines. In the basins that have multiple meridional and zonal crossings, this assumption seems valid. However, in a few of the basins, such as in the Philippine Sea and the Amundsen–Bellingshausen Basin, repeat sections cross only a small portion of the basin. In these cases, the changes seen in the single region are applied to the whole basin (Fig. 1). Generally, the error analysis ensures that the uncertainties for these basins, with their limited DOF, are appropriately large. The third assumption is that the time scale of the dθ/dt changes observed is longer than the intervals between occupations. This assumption generally seems valid because different sections taken over different time intervals in the same basin yield similar estimates of dθ/dt (e.g., Fig. 5) and a consistent large-scale geographical pattern emerges from the analysis (Fig. 8).
While we focus on deep warming of Antarctic origin here, the temperature and salinity of NADW also varies significantly on interdecadal time scales (Yashayaev and Clarke 2008), with implications for long-term warming (Levitus et al. 2005). Our study includes some of this recent variability, especially in the abyssal North Atlantic. However, the deep Nordic seas have also warmed at a rate around 0.01°C yr−1 during the 1990s (Osterhus and Gammelsrod 1999; Karstensen et al. 2005). The local heat flux that would be needed to account for the warming in theGreenland Sea below 1500 m since 1989 is on the order of 50 W m−2 (Karstensen et al. 2005), with a similar value for the warming in the deep Norwegian Sea starting in 1980 (Osterhus and Gammelsrod 1999). These marginal seas are neither ventilated by AABW nor sampled by repeat hydrographic data available at the CLIVAR and Carbon Hydrographic Data Office (CCHDO), so are excluded from our study. However, if the Arctic Ocean and Nordic seas had been included, the global heat flux estimates for below 2000, 3000, and 4000 m presented here could have increased.
The heating reported here is a statistically significant fraction of previously reported upper-ocean heat uptake. The upper 3000 m of the global ocean has been estimated to warm at a rate equivalent to a heat flux of 0.20 W m−2 applied over the entire surface of the earth between 1955 and 1998 with most of that warming contained in the upper 700 m of the water column (Levitus et al. 2005). From 1993 to 2008 the warming of the upper 700 m of the global ocean has been reported as equivalent to a heat flux of 0.64 (±0.11) W m−2 applied over the earth's surface area (Lyman et al. 2010). Here, we showed the heat uptake by AABW contributes about another 0.10 W m−2 to the global heat budget. Thus, including the global abyssal ocean and deep Southern Ocean in the global heat budget could increase the estimated ocean heat uptake over the last decade or so by roughly 16%. Considering the ocean between 700 m and the upper limits of our control volumes could add more heat (von Schuckmann et al. 2009; Levitus et al. 2005), reducing the percentage of the contribution computed here somewhat.
The global SLR due to upper-ocean thermal expansion is estimated at about 1.6 (±0.5) mm yr−1 (Bindoff et al. 2007) and the contribution from deep Southern Ocean and global abyssal warming estimated here is about 9% of that rate. This percentage is smaller than the relative contribution to the heat budget because the thermal expansion of seawater relative to its heat capacity is reduced at cold temperatures and deep pressures. While the warming in the global abyssal and deep Southern Ocean only contributes to a fraction of the global SLR budget, the local contribution from deep warming in some regions is similar in magnitude to the global upper-ocean contribution. For instance, in the region south of the SAF, the local SLR estimate due to thermal expansion below 1000 m is greater than 1 mm yr−1 (Fig. 8b). The warming of the deep ocean is contributing to global and local heat and SLR budgets and needs to be considered for accurate assessments of the roles of the ocean in climate change.
Several possible mechanisms could affect the warming reported here, perhaps in combination. First, changes in buoyancy forcing in AABW formation regions could reduce formation rates or change water properties, resulting in local and remote warming (Masuda et al. 2010), possibly over long time scales. This change may be due partly to changes in air–sea heat flux in addition to melting continental ice (Jacobs and Giulivi 2010). Second, long-term intensification and southward movement of westerly winds that drive the ACC appear to result in southward shifts of ACC fronts, also creating warming in the Southern Ocean (Gille 2008; Sokolov and Rintoul 2009), perhaps to great depth as seen here. These stronger winds may also spin up the Weddell gyre, increasing the temperature of abyssal waters escaping that gyre to flow northward (Jullion et al. 2010), and perhaps similarly the Ross gyre. Finally, wind strength and position also affect ice coverage and warm water transport into formation regions, changing formation rates of AABW (Santoso and England 2008).
Abyssal warming in the Southern Ocean is likely to be at least partly caused by advection of warmer water directly from the sources. These basins are located directly downstream from the source regions for AABW and are filled on time scales captured by the data used in this study, as shown by transient tracer burdens (Orsi et al. 1999). Outside of the Southern Ocean, advection times from AABW formation regions to some abyssal locations, such as the North Pacific, can approach 1000 years (Masuda et al. 2010). However, a deep warming signal can propagate from the Southern Ocean to the North Pacific via planetary (Kelvin and Rossby) waves in less than 50 years (Nakano and Suginohara 2002), moving northward on western boundaries, eastward on the equator, poleward at eastern boundaries, and then westward into the interior (Kawase 1987). This remote warming signal could be driven by an increase and southward shift in Southern Ocean westerly winds (Klinger and Cruz 2009) or reductions in buoyancy fluxes near the AABW formation regions that decrease AABW formation rates (Masuda et al. 2010). Alternatively, the more distant changes could also be advective, resulting from changes in AABW formation centuries ago.
The local abyssal heating rates outside of the Southern Ocean (Fig. 8a) are comparable to geothermal heating, typically 0.05 W m−2 away from ridge crests, which can have a significant impact on abyssal ocean circulation and water properties (Joyce et al. 1986; Adcroft et al. 2001). However, if the ocean circulation and geothermal heat fluxes are in steady state, this heating should not cause trends in abyssal temperatures. But, if the abyssal circulation were to slow, geothermal influences might contribute to a change in abyssal temperatures and even circulation.
To gain more precise estimates of the deep ocean's contribution to sea level and global energy budgets, and to understand better how the deep and abyssal warming signals spread from the Southern Ocean around the globe, higher spatial and temporal resolution sampling of the deep ocean is required. The basin space-scale and decadal time-scale resolution of the data used here could be aliased by smaller spatial scales and shorter temporal scales. Furthermore, the propagation of the signal can only be conjectured, not confirmed, with the present observing system.
In summary, we show that the abyssal ocean has warmed significantly from the 1990s to the 2000s (Table 1). This warming does not occur uniformly around the globe but is amplified to the south and fades to the north (Fig. 8). Both Indian and Atlantic oceans only warm on one side, with statistically insignificant cooling on their other side. The recent decadal warming of the abyssal global ocean below 4000 m is equivalent to a global surface energy imbalance of 0.027 (±0.009) W m−2 with Southern Ocean deep warming contributing an additional 0.068 (±0.062)W m−2 from 1000 to 4000 m. The warming contributes about 0.1 mm yr−1 to the global SLR. However, in the Southern Ocean, the warming below 1000 m contributes about 1 mm yr−1 locally. Thus, deep-ocean warming contributions need to be considered in SLR and global energy budgets.
Acknowledgments. Our heartfelt thanks go to all those who helped to collect, calibrate, and process the WOCE and GO-SHIP data analyzed here. Discussions with John Lyman were useful. Comments from Susan Hautala, Takeshi Kawano, Michael Meredith, LuAnne Thompson, Joshua Willis, Carl Wunsch, and two anonymous reviewers improved the manuscript. The findings and conclusions in this article are those of the authors and do not necessarily reflect the views of the National Oceanic and Atmospheric Administration (NOAA). The NOAA Climate Program Office and the NOAA Office of Oceanic and Atmospheric Research supported this research.
Return to Abstract
Return to Previous section
Go to References