Drones' New Mission: Save the Forests

CORVALLIS, Oregon—“Everyone step back 20 meters!” Michael Wing shouts before a Phantom 3 quadcopter loaded with sensors takes off in a clearing. Even though the drone’s pilot, Jonathan Burnett, has many hours of flying under his belt, caution is at the forefront. “All it takes is a fraction of a second to damage your arm, your hand, or your face," Wing says, pointing at a set of six-inch blades.

I take the message to heart and back up a lot farther than I need to. By the time I stop walking Burnett has started up the rotors of the three-pound, two-foot-long machine. The buzz isn’t as loud as I expected, but I’m still glad that I’m not standing next to it.

Then, in a flash, the drone is in the air. It shoots upward, levels off, and then zips across the field. At the controls, Burnett swings it around and points it where it needs to go. Every change in direction is in the blink of an eye.

FULL COVERAGE: Fight for the Forests

Wing and Burnett aren’t flying for fun or testing a way to deliver your packages by air, Jetsons style. Rather they’re scientists, and the flying robot buzzing overhead may just help save the world’s forests and the wildlife that depend on them. That includes nearly a million acres of Douglas fir trees in the Pacific Northwest that are dying of a disease that is spreading rapidly throughout the region.

Douglas firs are among the most important tree species for carbon sequestration, foot soldiers in the fight against climate change. Sick trees stop growing and absorbing carbon, reducing their effectiveness. The disease, called Swiss needle cast, also threatens the recovery of two endangered birds, the northern spotted owl and the marbled murrelet. Both species prefer to nest in old-growth Douglas firs. Many of the trees set aside for these rare birds are suffering from the disease.

The future of these forests—and those around the world—may depend on new technologies like those being developed in Wing’s Lab at Oregon State University here.

Unmanned aerial vehicle technology has come a long way in four years, and experts expect that drones will get even better in the decade to come. Yet, even as the use of drones expands, several technical challenges and roadblocks threaten researchers’ ability to make the most of them. Will they be solved in time to make a difference for the planet’s forests?

Tree Plague

As we drive over rough mountain roads, the ancient truck bouncing on every rut and rock, the forest around us seems healthy. Thriving, even.

We’re surrounded on all sides by enormous Douglas firs, towering 100 feet and more above. On a typically rainy Oregon afternoon in mid-March, they practically glow green.

Then we turn a corner. Suddenly the trees aren’t so green. They’re sad, spindly, and shrunken, with sparse, slightly yellowed needles. Each one looks worse than the last.

“These are the ones hit by Swiss needle cast,” explains Mark Gourley, a forester with Starker Forests, a timber company that manages 84,000 acres of woodland in Oregon, including the one we just traveled half an hour up a mountain to see.

Mark Gourley points out a dying section of Douglas firs plagued by Swiss needle cast disease. (Photo: John Platt)

The disease, which affects only Douglas firs, infects the trees’ needles, blocking their ability to absorb carbon dioxide and release oxygen. It effectively starves the trees of carbon, and infected trees grow 20 to 55 percent less than healthy ones. The lack of nutrition takes a toll, and the trees begin to die.

The fungus that causes the disease is native to the Pacific Northwest, but it’s gotten much worse in recent years, according to the Swiss Needle Cast Cooperative, a coalition of government, academic, and corporate experts. Warmer, wetter springs—fast becoming the norm in coastal Oregon—allow the pathogen to spread. Surveys conducted last year revealed that more than 900,000 acres in Oregon and Washington have been severely affected. That, according to Oregon officials, costs companies at least $78 million in logging income per year.

Enter the drones. Over the past few years, UAVs have become a vital research tool for forest and wildlife research. They’re being used to do everything from counting trees and wildlife to looking deep inside the trees to get a picture of their health. These devices offer researchers the ability to collect data at speeds never before possible and at incredible cost savings.

Robots to the Rescue

Wing ushers me into his basement lab. High-tech parts sit on every surface. Batteries. Cameras and sensors. Miles and miles of wires and cables. Box after box of parts.

Everywhere you look, there are drones—or as Wing prefers to call them, unmanned aviation systems, a less freighted phrase.

“We’ve got a dozen aircraft down here,” he says, pointing around. Here’s a white quadcopter, not much bigger than an ink-jet printer. There’s a bigger gray-and-black flier with eight arms that’s being modified to improve its stability. Above us, on top of a metal cabinet, is a four-foot-long fixed-wing plane made out of foam, orange tape, and hand-wired electronics. This, Wing explains, was the first plane his team used nearly four years ago, and it has been flown all over the state.

With his nice suit, glasses, and close-cropped, slightly graying goatee, Wing projects an air of authority. He’s an associate professor in OSU’s College of Forestry and an expert in remote sensing. Researchers normally rely on satellites and airplanes to make observations about what’s going on at ground level, but over the past few years, Wing and his students have transitioned to the newer, smaller devices. They use drones to count trees, see how well various crops are growing, count wildlife, and observe the effects of climate change.

Michael Wing and Jonathan Burnett stand in their basement lab at Oregon State University. (Photo: Andrew Larkin)

One of their research projects involves Swiss needle cast.

Most of the work observing the disease is done by plane, says Burnett, a Ph.D. student in Wing’s lab. Pilots fly back and forth over forests at heights of at least 1,500 feet. Observers in the planes note which forests exhibit signs of the disease. Moderately infected areas have yellow or yellow-brown foliage. Severe infestations show up as brown foliage and sparse crowns.

Drones allow the scientists to get much closer to trees than would be possible with an airplane or a helicopter and without having to walk up to every single tree. They can operate the device from hundreds of feet away, standing in place, while the drone swoops around and above the trees and collects data over the course of a 20-minute flight—which is how long the machine’s battery typically lasts.

All of that is easier said than done, though. Wing and Burnett have each logged thousands of hours of flight time, but they still face challenges. Wind makes the devices hard to control, and any number of technical problems can bring down a drone and its payload of sensors in an instant, taking thousands of dollars with it. “That’s the one thing I’ve learned,” Burnett says. “It’s not if you crash—it’s when.”

Even if a crash doesn’t destroy the drone, it still sets research back. Using drones means you’re in a constant state of repair, Wing says.

The biggest challenge, however, is not just getting drones airborne. It’s figuring out how to use them to collect data. Few people have done this research, so the scientists have to design every experiment almost from scratch. Each project requires them to test different cameras and other hardware, figure out the right height and angles at which to fly, and determine the correct environmental conditions. “We did one experiment looking at salmon beds, and it took us a while to figure out the right angle so the sun wasn’t reflecting off the water,” Wing says. “Jon spent a lot of time working on that problem.”

The drones Wing tested last year carried two near-infrared cameras, providing images of about a hundred Douglas firs at a time. Near-infrared cameras can determine radiation patterns below the surface of an object, like the Predator of movie fame but at a slightly different wavelength. Depending on the color of the image provided by the cameras, the researchers could tell which trees were healthy before the external signs of Swiss needle cast disease became obvious. The greener the image under near-infrared, they found, the healthier the tree. They confirmed their results by going back to verify sick trees in person. “We had about a 90 percent success in identifying both sick and healthy trees,” Burnett says.

Burnett says using drones to identify which trees are infected could help contain Swiss needle cast. Some trees could receive a fungicide treatment. Others might be cut down. Thinning the fir forest could slow the spread of the fungus and allow hemlock trees, which aren’t affected by the disease, to fill the gaps. Such decisions would not be possible without these eyes in the sky.

Detecting Drought

Douglas fir trees aren’t the only ones in trouble.

In Colorado, The Nature Conservancy has put drones to work on its Zapata Ranch to study local cottonwood trees, which are dying at an alarming rate.

Gustavo Lozada, The Nature Conservancy's operations manager, operates a drone in Colorado. (Photo: Courtesy Zapata Ranch)

Gustavo Lozada, an operations manager for The Nature Conservancy, started using drones to study cottonwoods in 2014. He found that a near-infrared camera—like those used in Oregon—could identify which trees were healthy, dead, or in the process of dying.

“We mapped about 700 acres with the near-infrared,” Lozada says. The drone not only saved walking time but allowed researchers to get much better resolution than would have been possible with satellite images. “You can zoom the image up to two centimeters and check out leaves, branches, or grass.”

Just don’t get too close to the birds. During one of the cottonwood flights, an eagle appeared out of nowhere and flew perilously near the drone. Lozada instructed his pilot to let the UAV fall rather than hit the bird. The drone plummeted 100 feet away from the eagle before the pilot was able to get it hovering again just above the ground. “We prefer to crash the drone rather than hurt any wildlife,” Lozada says.

Top: A route for a drone is plotted out before its flight. Middle: The area is mapped using drone photos, then stitched together with the help of Google Earth. Bottom: Software separates different vegetation based on photosynthetic activity. (Images: Courtesy The Nature Conservancy)

The near-infrared sensor gave the researchers insight into the impact of drought on the trees. “If the image is greener, there is more water in that area,” Lozada says. The cottonwoods, he says, are probably suffering from a lack of water in part owing to last summer’s drought. It might also be a sign of a more long-term problem, as several studies have linked other cottonwood die-offs to climate change.

Figuring out what’s happening to the cottonwoods could take years, but for now, Lozada’s drone is helping to establish a baseline of information that will provide contextual clues and show how the trees change over time.

Time is also a factor in work being done in Costa Rica, but there it’s about rebirth.

Rakan Zahawi and his colleagues with the Organization for Tropical Studies have spent the past 15 years studying how forests can be restored on former agricultural land. Most of their monitoring involves long hikes to collect data on the ground, but a few years ago they added drones to their tool kit. For one thing, that allowed them to increase the frequency of their observations. “You can go out and fly this drone every day if you want and look at change in a particular patch or an individual tree,” Zahawi says.

Drones also help provide a bigger picture. Zahawi and his team fly drones in a circle around plots of forest, taking thousands of photos along the way that are then stitched together into a 3-D image.

That’s cool in itself, he says, but it also allows them to compare the data from multiple flights to tell how tall a forest is growing over time. Canopy height is a good sign that a forest is becoming hospitable for birds and bats. There are also climate considerations: “You can use it to calculate how much carbon’s being fixed aboveground,” Zahawi says.

The software that analyzes data from drones is an essential part of the process. “We have been using a lot of photogrammetry software,” says Lian Pin Koh, a professor at the University of Adelaide in Australia. “It stitches together the images to produce a 3-D landscape.” Once the model is up on the screen, you can zoom in and out and manipulate the image to get a better understanding of the landscape.

Drone work has come a long way in the four years since Koh and his partner, Serge Wich of Liverpool John Moores University in the United Kingdom, launched their nonprofit, Conservation Drones. They started with a few test flights on the Indonesian island of Sumatra. Now they assist researchers all over the world. Their technology has been used to track radio-collared animals, observe an endangered island tree species, monitor bird colonies on remote islands, and—perhaps most notably—uncover evidence of illegal logging.

Conservation Drones revealed rogue timber operations in Sumatra in 2012 and has since trained officers at the Sumatran Orangutan Conservation Program to keep the fight going. The program’s drone operators look not only for felled trees but also for things like blue tarps, a telltale sign that loggers have set up camp in a forest. “They have been flying the drones over protected areas, and they have been able to pick up evidence of illegal logging just by regularly flying over the same area,” Koh says. “That data has been very useful in helping the local authorities to track down who the perpetrators are and take action against them.”

Koh says he’s impressed by how quickly the technology has developed over the past few years and how researchers around the world have embraced it. “I am hopeful that this technology will soon be an everyday thing,” he says.

Data Overload

Drones have a few downsides. For one thing, they produce almost too much data. “If you take a picture every two seconds during an hour flight, that’s 1,800 pictures,” Wich says. “Let’s say you need a minute per picture to pick out what you’re looking for, like an orangutan nest. That’s 30 hours of just looking at pictures—almost a week.”

Also, just because you hold a doctorate doesn’t mean you’re qualified to do research by drone. Wing says a computer-savvy scientist can figure out how to get a drone into the air for short flights in as little as an hour, but mastering the controls—not to mention getting licensed by the Federal Aviation Administration—takes much longer. You need to learn how to fly a drone up, down, forward, back, and as far as you can see. That takes a lot of practice. A typical drone flight-training course requires 20 to 40 hours of flight time, and potential pilots also need to pass medical and knowledge exams. “Not everyone’s capable of fulfilling that,” Wing says.

Another drawback is weather. My first attempt to see Wing’s drones in action is canceled after a 20-mile-per-hour windstorm hit the area. The second trip is derailed by sun. Bright light, Wing tells me, would create too many reflections for the cameras and sensors.

We watch the weather for another opportunity. The third try, too, seems iffy, with just enough rain drizzling to make Wing nervous. Water not only would damage the drone’s motors and autopilot but also could ruin the camera or sensors as thoroughly as a crash.

The day I visit his lab, the weather holds.

The drone’s flight over a field near the OSU campus is flawless, but not everything is working perfectly. The machine is lacking one critical piece of hardware: a thermal sensor. The team plans to use it as part of a new research project to see if it can count the elk hiding under trees in a nearby wildlife reserve. The sensor, the researchers hope, can not only help count the animals but also reveal how large and therefore how old they are, to provide clues about the reproductive health of the herd. “The sensor is in Texas,” Wing says. After the team bought the device, it found out it needed an extra part to enable it to store data. A small company said it could modify it. Months later, the sensor still hasn’t arrived back in Oregon.

At Oregon State University, Jonathan Burnett test-flies a drone to be equipped with a thermal sensor as Michael Wing and reporter John Platt observe. (Photo: Andrew Larkin)

Equipment problems like that are frustrating, Wing says. Most drones, batteries, and related parts are made in China, and delivery can be inconsistent. “Some things we’ve ordered haven’t shown up for months,” he says. That slows research, as does drones’ 20-minute battery life, which is problematic for experiments demanding an hour of flying time.

It doesn’t slow Wing down though. His team has a full agenda in place for Oregon’s less rainy spring months, including continuing its work with Swiss needle cast. Once the thermal sensor arrives, the team will go into the forests to test its elk-counting skills. Another project will look at logged forests to see how they’re regenerating, while a third will see the researchers travel to Arizona to study how seedlings are faring in warmer temperatures.

Looking farther ahead, Wing predicts drone technology will continue to improve and grow, as will the list of ways they can be used to study the world’s forests and the animals that live in them. “This is just a start,” he says. “It’s going to look a lot different in a few years.”