“This is the first evidence that Greenland’s supraglacial lakes have responded to recent increases in surface meltwater production by draining more frequently, as opposed to growing in size,” says CIRES research associate William Colgan, who co-led the study with CU’s Yu-Li Liang. The results were published online April 15 in Remote Sensing of Environment and will appear in the journal’s August issue.
During the summer, meltwater pools into lakes on the ice sheet’s surface. When the water pressure gets high enough, the ice fractures beneath the lake, forming a vertical drainpipe, and “a huge burst of water quickly pulses through to the bed of the ice sheet,” Colgan says.
The researchers used satellite images, along with innovative feature-recognition software, to monitor nearly 1,000 lakes on a Connecticut-sized portion of the ice sheet over a 10-year period. They discovered that as the climate warms, such catastrophic lake drainages are increasing in frequency. Catastrophic lake drainages were 3.5 times more likely to occur during the warmest years than the coldest.
During a typical catastrophic lake drainage, about 10^7 m^3 of meltwater—equivalent to 4,000 Olympic swimming pools—funnels to the ice sheet’s underside within a day or two. Once the water reaches the ice sheet’s belly, it may turn the ice-bed surface into a Slip ‘n Slide, lubricating the ice sheet’s glide into the ocean. This would accelerate the sea-level rise associated with climate change.
Alternatively, however, the lake drainages may carve out subglacial “sewers” to efficiently route water to the ocean. “This would drain the ice sheet’s water, making less water available for ice-sheet sliding,” Colgan says. That would slow the ice sheet’s migration into the ocean and decelerate sea-level rise.
“Lake drainages are a wild card in terms of whether they enhance or decrease the ice sheet’s slide,” Colgan says. Finding out which scenario is correct is a pressing question for climate models and for communities preparing for sea-level change, he added.
For the study, the researchers developed new feature-recognition software capable of identifying supraglacial lakes in satellite images and determining their size and when they appear and disappear. “Previously, much of this had to be double-checked manually,” Colgan says. “Now we feed the images into the code, and the program can recognize whether a feature is a lake or not, with high confidence and no manual intervention.”
Automating the process was vital since the study looked at more than 9,000 images. The researchers verified the program’s accuracy by manually looking at about 30 percent of the images over 30 percent of the study area. They found that the algorithm correctly detected and tracked 99 percent of supraglacial lakes.
The program could be useful in future studies to determine how lake drainages affect sea-level rise, Colgan said.
CIRES coauthors on the team include Konrad Steffen, Waleed Abdalati, Julienne Stroeve, and Nicolas Bayou.
The study was funded by the Arctic Sciences Program of the US National Science Foundation.
Republished with permission from CIRES.