“We’re never going to make maps better than this,” Matt Nolan says, examining a three-dimensional topographic map of the Toklat River in Denali National Park. He may as well be looking at a compilation of high-resolution satellite images. The map is so precise that there are markers that indicate how the river channel shifts during the summer.
Nolan, a 49-year-old glaciologist and professor at the University of Alaska, Fairbanks, made the map using a new cartographic technique called Structure from Motion (SfM) photogrammetry, or “fodar” as Nolan calls his version, which uses survey-grade GPS. The method combines digital photographs—usually captured from the air—with highly sophisticated algorithms and software to create three-dimensional topographic maps that are accurate to within a few centimeters over a scale of, in some cases, thousands of square miles. Think the new and improved Google Earth but at 100 times higher resolution. The maps can depict everything from Arctic landscapes to coastlines to coral reefs, and when the photographs are taken over time—say, every two weeks or every month—they can show how landscapes change through the seasons with unprecedented precision.
SfM basically works like this: Say you want to know the precise climbing and skiing route on the north face of Longs Peak in Colorado. You’d use a drone or manned aircraft to photograph the area you want to map (drones are much more efficient for small areas; planes are essential to map a wider swath), with sections of different shots overlapping so as to link the entire area through the range of images. Then you’d plug the images into special software, some versions of which are proprietary and some of which are free to the public, and the software would spit out a 3-D map.
The technology is a little geeky and complicated, but according to those who have watched SfM photogrammetry’s rampant increase in use over the past five years, it has the potential to revolutionize maps that are key to solving a range of societal problems, like where to build roads, bridges, and airports; measuring damage from storms; and predicting coastal erosion from past observations.
“The first time I used it, I was blown away,” says U.S. Geological Survey (USGS) research geologist Jonathan Warrick. “It was so elegant and so simple.”
Photogrammetry itself—essentially using 2-D photographs to construct a 3-D image—has been around for more than 60 years. But until the advent of digital photography and enormous improvements in software, someone had to identify objects in the photos and make painstaking measurements and calculations to construct a map. “It was very labor intensive,” Warrick says. “Now, computers can do that work. They can match objects in photos—take a tree in one image and match it with the same tree in another image. And they can do all the calculations and geometric measurements, too.”
The range of applications is vast, particularly when the finished product is what’s called a “time series” map—i.e. one that incorporates photos of the same area taken over a period of time for comparison purposes. Nolan has mapped glaciers and watersheds in Alaska’s Brooks Range to explain how snow depths, glacial ice, and sea ice are changing, all of which are crucial to tracking the effects of climate change. The USGS uses the technique to study coastal erosion from the California coast to Cape Cod, as well as to map coral reefs in Maui. And the National Oceanic and Atmospheric Administration has been testing its potential—via drones and manned aircraft—to measure hurricane damage on the Atlantic Coast. Specialists can also use it to tell whether roads and bridges are decaying or sinking; and where wildlife are congregating via their footprints on the tundra.
Last month, a paper detailing Nolan’s and ski mountaineer Kit DesLaurier’s work in 2014 to determine the highest peak in the American Arctic was published in the journal The Cryosphere. Using fodar, a word Nolan derived from a laser-based technique called Lidar that produces high-quality maps but without photographs and at a much higher cost, the duo proved the long-assumed tallest peak in the region, Mt. Chamberlin, was actually just the third-tallest peak, contrary to what an old USGS map showed. The paper attracted a fair bit of media attention, but Nolan thinks it overshadowed the real story—about how vast the mapping technique’s potential is.
He is so convinced that he stopped applying for research grants through the university this past May, and has instead focused exclusively on his cartography business, Fairbanks Fodar. “I’ve already made five times my annual salary as a professor, and I’m a one-man show and effectively the tip of the iceberg,” he says.
One contract from the Alaska Department of Natural Resources paid Nolan $300,000 to map a 1,600-mile-long, one-mile-wide stretch of lightly populated coastline where native villages are at risk of growing storm surges, which can destroy infrastructure that is built too close to sea level or flush effluent out of so-called sewage lagoons. Over the course of a year, he flew his Cessna 170 for 29 days and 120 hours, snapping photos through a hole in his plane behind the pilot’s seat—up to 10,000 images per day on a standard Nikon DSLR camera.
The trigger is automated and connected to a survey-grade GPS unit, associating each image with a precise location. Depending on what altitude Nolan is flying at when the images are captured (usually between 1,000 and 4,000 feet), the resolution can be so high that each pixel in the image represents just ten centimeters of distance on the ground. He can detect where the ground has sunk a single inch from one flight to the next, a trend that not only tells road workers where their next maintenance project might appear but also where the ground is more stable for future road construction. “You really want to see where these problems are happening before they become catastrophic, so you have time to mitigate and deal with them,” Nolan says.
Although Lidar is still the incumbent manner of making topographic maps due to its highly regarded precision, it can cost the contractee up to 10 times as much as SfM photogrammetry and requires cost-prohibitive equipment that is inaccessible to most scientists. In 2012, when Warrick and a colleague wanted to create a time-series map of sediment movement following the largest dam removal in U.S. history on Washington’s Elwha River, they had two choices given the funding at their disposal: use Lidar and fly over the river valley once, or use SfM photogrammetry and fly over the valley twice a week for two years.
“I kind of look at it like some moon scientists sitting around the water cooler saying, ‘Gee, wouldn’t it be nice to go up to the moon tomorrow and grab some rocks and answer this question we have?’” Nolan says. “We have the technology to go to the moon, it’s just ridiculously expensive so we don’t do it. Well, it’s the same thing with fodar: we had the technology to make super awesome topographic maps, it was just too expensive so we never did it. Now it’s affordable and now we’re starting to use it. We can directly measure change on a centimeter scale over thousands of square kilometers.”
Despite its high cost, most experts believe Lidar—a multibillion-dollar industry—will never go away because it’s still remarkably accurate and its lasers do something SfM photogrammetry’s cameras struggle to achieve: penetrate a forest canopy to provide a precise distance to the ground, if not an image of the terrain. Still, the cheaper and, in some respects, more dynamic SfM technique is creating a niche that figures to expand, thanks to the photographic tool it provides.
Says Warrick: “I certainly see this as a big part of our future.”