Drones look for sharks, map NC beaches
Tourists walked along the shoreline of Bird Shoal, an island in the Rachel Carson Reserve across the water from Beaufort, soaking in the afternoon sun last week, never seeming to notice the non-indigenous bird that flew back and forth overhead in a perfect grid pattern.
As the visitors collected scallop shells and barnacle clusters, a group of scientists collected an ocean of data using a fleet of softly humming drones.
“It’s coming back around,” Dave Johnston said as one of the tiny aircraft buzzed by, about 275 feet above the sand. “Looks great.”
Johnston is executive director of the Marine Conservation Ecology Unoccupied Systems Facility – more commonly called the Drone Center – that opened last year at Duke University’s Marine Laboratory in Beaufort.
The Marine Lab has begun using drones to measure and map coastal erosion, scope out ocean debris, count gray seals and sea turtles, sample whale blow and even spot sharks. The lab is funded by grants and working with public and private partners, including researchers from UNC, NCSU, the N.C. Department of Wildlife Resources and the National Park Service.
The data eventually could have widespread implications, helping communities monitor erosion on barrier islands that could increase flooding risks there and on the mainland; help determine whether protected marine species are recovering or declining; guide the cleanup of the debris that’s washed ashore by storm surges and measure whether habitats recover when trash is removed; gather information about protected species such as right whales, whose expectorant can even reveal a pregnancy; and alert swimmers when dangerous sharks are prowling close to shore.
Johnston looks at drones with the creative glint that sparkles in the eye of a kid with a new toy whose full power is not yet tested.
“There are almost endless possibilities,” said Johnston, whose job description at Duke includes finding ways to use new technology in education and research.
To put drones to work, Johnston and his team, which includes program manager Everette Newton, a retired Air Force colonel, and Julian Dale, lead engineer, first had to get clearance to fly from the Federal Aviation Administration. FAA rules presently ban “commercial uses” of drones, including for academic research.
While waiting months for their FAA exception last fall, the Duke team took some of its drones to Costa Rica, where team members worked with researchers from UNC who were mapping aggregations of endangered olive ridley sea turtles. They also traveled to Canada to help the government there count endangered gray seals.
“This can be really dangerous work,” when done by traditional methods, said Kerry Irish, who is Johnston’s wife and spokeswoman for UNC’s Institute of Marine Sciences. Planes and helicopters flying close enough to get good photos sometimes crash, killing or injuring photographers, and those aircraft disturb some of the kinds of animals they’re surveying.
Hiring a plane for aerial photography is also costly, Johnston said, “and you can’t always rustle up a plane when you need one.”
Depending on the sophistication of the machine, a drone can be manually directed or programmed to follow a particular flight pattern. Drones can be outfitted with still or video cameras, whose shutters can be triggered by the operator on the ground or set to automatically snap photos at regular intervals.
Depending on the setup, the photos can be seen by the operator as they’re being made, or not until the rig is back on the ground.
Researchers use computer programs to knit the photos together so they meld into a seamless tableau, providing a broad view comparable to a satellite photo but with higher resolution and with greater immediacy. And while a fresh satellite photo of a particular area might be available once a year, Johnston said a drone team could photograph the same spot as frequently as needed.
Johnston displays the drones’ photographs of the Canadian seals on a large screen in his waterfront lab. At first glance, the pictures have a lunar landscape quality: a pale gray wash with darker gray speckles in two sizes. Then Johnston zooms in, and zooms again. The large speckles take the soft, curving shape of adult seals and the small speckles of seal pups.
He zooms in once more, and the view is detailed enough to see that some of the pups are nursing.
For the seal project, researchers had to manually count the animals, coming up with about 5,000 at the site. Johnston said the margin of error is lower than with airplane photography because the images are taken so quickly that if a seal waddles away, the next frame will capture it in the place where it moved and a slight blur will indicate it’s the same critter.
Eventually, drones will be paired with software that can recognize particular features: seals, different types of turtles or shore bird, and will be able to send back a count to observers on the ground.
Johnston’s team is in the early stages of developing such a system for monitoring sharks in Atlantic waters. They have built a family of shark decoys to see if they can train a drone to spot and recognize them.
“We’re at least three to five years away,” he said, but it’s conceivable that drones could be sent out to fly over crowded beaches, spot something swimming in the water that matches the profile of a dangerous shark, and send word back to the beach patrol.
The lab has about 10 drones, including fixed-wings, which are light and can stay in the air longer before having to land for a battery change, and helicopters, which tend to use more battery power but can hover over a spot for extended periods and can fly lower without falling out of the sky. They have given nicknames to some of their favorites: Maverick and Ice Man, two of the fixed-wing birds, named for characters from the movie “Top Gun,” and Han Solo, a helicopter named for Harrison Ford’s character in “Star Wars.”
The researchers are testing different combinations of drones, cameras and software, trying to find cost-efficient packages that can capture the right type of images for a project at the right price. Duke has drones of the kind that can be purchased offtheshelf for less than $1,000, and programmable models that run closer to $50,000.
None of them, Johnston says, were designed for a marine environment.
“The first thing they tell you about a drone is, ‘Don’t fly it over water,’” Johnston said, because of the risk of a failure that results in the machine taking a fatal dip. “That’s 80 percent of what we do.”
One hazard the team has learned to manage is landing a drone on a moving boat. The craft navigate by GPS, linking an on-board system with one in the operator’s laptop or hand-held controls. But the motion of a boat on the water confounds GPS, so sometimes, Dale, the engineer, brings the aircraft in as close as possible and Newton, the retired airman, reaches up with a steady hand and snatches it down.
The pair practiced that maneuver several times last week in the low dunes on Bird Shoal, not to stay out of the water but to avoid hard landings that could drive sand into delicate moving parts.
That day, two teams of scientists were at work half a football field apart, a group from UNC taking measurements with traditional surveying tools, and the team from Duke with its flying electronics.
The hope is that one day the drones will replace the more cumbersome but highly reliable laser-line scanner UNC research colleagues hauled to the island with a boat and a wheel barrow. But for now, they’re working in tandem, with the laser-line crew setting targets with super-accurate GPS and the drones flying over and mapping where it finds them. Back in the lab, researchers will compare and see how well the drones’ performed.
Precision and reliability will be key on projects such as measuring coastal erosion, where it will matter whether a shoreline lost three feet of sand or if an inlet grew by 50 yards.
Recently, as tropical storm Bonnie approached, Johnston and his team took some drones and got some measurements of the tip of one of the islands in the reserve where erosion has accelerated in recent years. If the storm had turned into a major event, they would have been able to go back and photograph the same area immediately after the storm. The more scientists know about the damage caused by specific storms – where they know the wind velocity and tidal heights and other details – the better they can predict potential damage of future storms and hurricanes.
In the past, scientists have had to wait for months or years between comparable images, making it difficult to link changes with particular storms, construction, dredging or other activities.
With flying robots, Johnston said, “As long as it’s not blowing too hard, or raining too hard, we can go.”