The Southern Ocean has become less salty during the past few decades. An analysis of sea-ice transport in the ocean suggests that this phenomenon can be explained by coupled changes in sea-ice drift and thickness. See Letter p.89
The vast band of water that encircles the Antarctic continent, known as the Southern Ocean, is the world's dominant ocean sink for heat and carbon dioxide1. It also has a crucial role in the global overturning circulation — the sinking, at high latitudes, of cold, dense surface waters to the deep ocean, and the compensatory rising of deep waters originating from lower latitudes. The Southern Ocean's salinity has fallen during the past half-century, in the surface and intermediate waters of the open ocean2,3 and coastal regions4, and in deeper waters5. This freshening of surface waters has increased stratification (the vertical gradient of water density), potentially inhibiting upwelling of deeper water and affecting CO2 uptake6. On page 89, Haumann et al.7 show that the freshening can be explained by changes in Antarctic sea-ice production and transport (Fig. 1).
Previous explanations for this freshening have included a net increase in the difference between the amount of precipitation and the amount of evaporation over the ocean8, and increased input of glacial meltwater4,5. However, the former is inadequate for explaining the freshening in surface and intermediate waters of the open ocean, and the latter overestimates the freshening of deep waters. The constant movement of sea ice redistributes a substantial amount of fresh water9, so Haumann and colleagues chose to investigate the potential contribution of sea-ice transport to this observed freshening.
When sea ice forms, most of the salt is lost to the upper ocean, so the ocean loses fresh water and its salinity increases. When the ice melts, the fresh water is returned to the ocean. The net impact on the upper ocean would be minimal were it not for prevailing winds that tend to push the ice from coastal waters, where most of it is formed, to the north, where it melts. This drives a net transport of fresh water that contributes to the overturning circulation of the Southern Ocean. The saltier water that results from ice formation in coastal regions contributes to the generation of Antarctic Bottom Water, and the fresh meltwater input to the north mixes with upwelling deep water to modify the upper waters of the open Southern Ocean9.
Determining any trends in this sea-ice-driven freshwater transport is challenging, in part because of a lack of reliable data. The volume of sea ice transported can be calculated as the product of ice concentration (the fractional area of the ocean covered by ice), ice thickness and ice drift rate. However, satellite-derived ice-drift rates have significant biases relative to those measured by drifting buoys, and potential biases due to changes in data sources and satellite sensors over time. And there is no long-term data set for ice thickness.
Haumann et al. addressed these challenges by carefully reconstructing time series for each of these variables for the period from 1982 to 2008. First, they established a consistent satellite ice-drift time series by removing inconsistent data associated with the transitions between satellites, and stitched together different time periods by correcting for estimated biases. They then scaled the satellite ice-drift series to make it consistent with observed buoy drift.
To reconstruct a time series of ice thickness, the authors turned to a model-based estimate of ice-thickness trends constrained by observations of ice concentration. They then adjusted for potential biases in the modelled thicknesses using both sparse in situ data10 and ice-thickness estimates from satellite data11. The time series for ice drift and thickness allowed Haumann et al. to make more-robust estimates of freshwater transport than were previously possible. This, in turn, allowed them to estimate the impact of transport trends on the salinity of the Southern Ocean using a simple model of water-mass exchange between the surface and the deeper waters.
The researchers show that the net transport of sea-ice-driven fresh water is substantial: larger than the inputs from glacial melt and comparable to the net input of precipitation and evaporation9. The estimated temporal trends are also sizeable: there is a 20% increase in transport over the 26-year study period. Notably, however, there is considerable regional variability in freshwater transport trends, including a large increase in the Pacific sector of the Southern Ocean (which encompasses the Ross Sea, where positive trends in northward ice drift and extent are largest12). Transport has decreased slightly elsewhere. Overall, Haumann et al. estimate that sea-ice-driven transport has contributed enough fresh water to the open-ocean surface and intermediate waters to explain the observed freshening.
A compelling result is that the calculated trends in sea-ice-driven freshwater transport are consistent with other observed patterns of change. First, the increases in freshwater transport occur in the Pacific sector, where increased freshening in surface waters has been strongest2. Second, the increase in salt input due to sea-ice production in the coastal Pacific sector might explain why the observed freshening of Antarctic Bottom Water is less than that predicted from increased glacial melt5.
It is striking that major changes to ocean properties can occur as a result of relatively small average changes in sea-ice cover. Sea-ice extent has increased only slightly overall during the period covered by the time series, albeit with strongly contrasting regional patterns of change13. These regional changes were partly wind-driven12, but, as Haumann et al. show, there may be little to no trend in the mean drift speed of sea ice. This demonstrates that it is the coupled trends in regional ice thickness and ice drift that are key to driving freshwater redistribution.
An important caveat to the findings is that the uncertainty in the derived trends is considerable, and potentially underestimated. The corrections for bias in ice drift are large, and are difficult to quantify for the earlier years, for which there are almost no independent data available to provide validation. Nevertheless, the authors' estimates of freshwater transport remain similar when they are based on ice drift estimated from surface winds, which are a reasonable proxy for drift. The need for better ice-thickness estimates is also clear; ice thickness is the largest source of uncertainty in the results, and ice-thickness trends are the least well constrained by observations. However, a recent complementary study9 that used a broader array of observations collected between 2005 and 2010 to constrain a coupled ice–ocean model broadly supports the regional patterns of sea-ice-driven freshwater transport estimated in the current study, allaying concerns about the uncertainties.
Haumann and colleagues' findings emphasize that Antarctic sea ice is not merely a passive indicator of climate change and variability, but also a driver of changes in the climate system. Through its potential influence on ocean stratification and CO2 uptake, sea ice might have a bigger role than previously thought.
The implications of these results for the Southern Ocean in a warming world are uncertain, because climate models do not properly capture the observed changes in Antarctic sea ice14. However, anticipated future declines in ice extent and volume would suggest that sea-ice freshwater transport should decrease. If so, then future losses of sea ice can be expected to play a prominent part in changes in the Southern Ocean's overturning circulation.