Despite global warming, the fringe of sea ice around Antarctica is expanding slightly, in contrast to the marked decline of sea ice in the Arctic. Scientists have blamed this curious fact on various forces, from shifting winds to smaller waves, but a new study suggests a more mundane culprit: an error in the way the satellite data have been processed. The miscalculation, the authors say, might be making the sea ice increase appear larger than it is.
Sea ice cover—ice that’s floating free on the ocean surface, rather than on land—has been observed by satellites since the late 1970s. Satellite sensors use different frequencies and polarizations to distinguish ice from water, and scientists have developed several algorithms to then process the data, remove weather effects, account for ponds of melted water on the ice, and the like. Data based on one of the most commonly used algorithms, dubbed Bootstrap, were used in the Intergovernmental Panel on Climate Change’s (IPCC’s) Fourth (2007) and Fifth (2013) assessment reports.
The two reports both commented on the enigma of overall sea ice expansion around Antarctica, despite warming atmospheres and ocean waters. Complicating the story is the observation that the expansion varies regionally and seasonally—it’s not actually expanding everywhere, and in some places it’s clearly retreating. Scientists have attributed this pattern to other effects of climate change—changes in the prevailing wind patterns that generally push the ice northward or in the size of ocean waves, which may shrink and allow the ice to expand or grow and herd it back toward the coast.
Even more puzzling was that the expansion seemed to be picking up speed. The 2013 report noted a significant increase in the expansion rate compared with the 2007 report, which “really jumped out at me,” says Ian Eisenman, a climate dynamicist at the Scripps Institution of Oceanography in San Diego, California. Within the climate science community, he says, that apparent change in the trend of how fast the ice was expanding was instead attributed to, basically, having more years of data between 2007 and 2013.
When Eisenman and his colleagues dug a little deeper, they realized that the two reports relied on two different versions of the Bootstrap data processing algorithm. Satellites come and go, and sensors improve all the time. So, to ensure continuity within a data set whenever a satellite changeover occurs, the data from an outgoing satellite and a new satellite are calibrated. Such a changeover happened in 1992, when a new satellite sensor was launched. Then, around 2007, Bootstrap got an update, in part to help account for ongoing sensor improvements. “There were a lot of changes from version one to version two,” Eisenman says. Several decades of data were recalculated with the new Bootstrap version.
And that’s the rub: The same data from 1979 through 2004, plotted by the two versions of Bootstrap, reveal two distinct trends. Version one (as reported in IPCC’s Fourth Assessment Report) suggests the ice is expanding by about 5.6 x 103 km2 per year. Version two shows the ice expanding more than twice as fast over that time period, at about 14.1 x 103 km2 per year. (IPCC’s Fifth Assessment reports an increase of 16.5 x 103 km2 per year for 1979 through 2012.)
“That implies inaccuracy in one of those versions—although which one isn’t clear,” Eisenman says. If the correct version of the Bootstrap algorithm is version one, then it follows that the sea ice isn’t increasing nearly as much as we thought, his team reports today in The Cryosphere.
Well, maybe. Climate scientist Josefino Comiso of NASA’s Goddard Space Flight Center Cryospheric Sciences Laboratory in Greenbelt, Maryland, is the scientist who worked on the Bootstrap calibration. He says that the revised Bootstrap two is the accurate one, and that the trend depicted in IPCC’s Fifth Assessment Report is correct. “Why not believe the one that has been corrected?” Comiso asks.
Among climate scientists, there seems to be little dispute that Eisenman and his co-authors have identified a very real discrepancy in the satellite algorithms. “It’s an excellent piece of scientific vigilance,” says Paul Holland, a climate scientist at the British Antarctic Survey in Cambridge, U.K. But what’s less clear, Holland says, is how much the discrepancy matters in terms of what’s really going on in Antarctica. Regardless of which trend is used, the data end up telling the same story: Antarctic sea ice is expanding, despite the warming climate.
And that overall increase isn’t even the most interesting thing, Holland adds. Yes, the sea ice is increasing on a continent-wide scale—but that overall slight increase is actually just the sum of stronger increases and decreases at different places ringing the continent—which the new study also reproduces, he points out. “To me, the interesting thing is what’s causing that spatial pattern.”
As for which version of Bootstrap is more accurate, Eisenman and his colleagues are continuing to investigate. Whichever it turns out to be, he says, the study highlights the need for “a more thorough documentation” of satellite data calibration methods, particularly with the growing interest in the sea ice cover at both poles. “There’s a real need for it now that these data sets are getting so much attention."