Scientists have successfully deployed an unmanned drone over the frozen ocean to spy on the vagaries of seals in the impenetrable winter wilderness of floes and bergs off Alaska's west coast, a Colorado scientist reported last week at the fall meeting of the American Geophysical Union in San Francisco.
The study combined eye-in-the-sky battlefield technology with state-of-the-art computing wizardry to discriminate telltale shadows from the fat bodies of marine mammals lounging on ice.
"Biologists are thrilled about the image recognition software because it could change the way we monitor seal populations," says lead researcher Elizabeth Weatherhead. "We can send an unmanned craft out from a ship, collect 4,000 images, and have them analyzed before dinner."
Originally designed for intelligence and military recon by Boeing, the ScanEagle has a 10-foot wingspan, with the capability to loiter at low elevation or soar as high as 16,000 feet.
This particular aircraft, operated by the University of Alaska, flew sorties in May and June of 2009 over the Bering Sea from the NOAA vessel McArthur II -- recording 27,000 images that ultimately contained seals, according to Weatherhead, a senior scientist at the Cooperative Institute for Research in Environmental Sciences, a joint venture of CU-Boulder and the National Oceanic and Atmospheric Administration. Alaska's ScanEagle weighs 27 pounds loaded with gear and fuel, and can fly for 20 hours at speeds up to 75 knots.
While the ScanEagle snapped pics and video of jostling bergs and offered insight into the behavior of the ice pack, it was the use of new image recognition software by Boulder Labs Inc. that helped pinpoint the hard-to-find, fish-eating mammals.
"The results show that the seals have distinct preferences for specific types of ice, demonstrating that ice extent is not the only factor affecting seal populations," Weatherhead says.
The researchers have created a sort of "Where's the Seal" game, demonstrating how difficult it is to find seals hauled out on the surface. If you want to test the "image recognition software" loaded in your own personal brain, give it a try.
For much more detail on the research, plus a link to a short video, go here. An older story about the project, with links to photos, is here.