Spies Like Us
Armed with low-cost surveillance technologies, nonprofits aided by “citizen spies” are tracking fracking in Pennsylvania, flaring in North Dakota, and rogue fishing around Easter Island
By W. Wayt Gibbs
It’s a pitch-dark evening in November, and a gaggle of knitters are chatting around a table in the Yarnability knitting shop in Shepherdstown, West Virginia, a historic town 76 miles up the Potomac from Washington, D.C. As the women stitch dishcloths and potholders, an intelligence operation is being hatched in the back rooms of the building. Five tech-savvy environmentalists dial up a video chat to their coder, a member of the Swedish Pirate Party in Gothenburg. The team, which calls itself SkyTruth, is readying its latest surveillance tool for public release in the morning. But the software is not cooperating.
A tall, bearded geophysicist peers over the shoulder of a much shorter, bespectacled software entrepreneur at a computer screen. The shorter one is Paul Woods, SkyTruth’s chief technology officer. His brow furrows as he scans a report showing upload failures for many of the multigigabyte image files—one-meter-resolution aerial photos of 39 counties in Pennsylvania—that he has been trying to place onto Google’s mapping servers. “We’re totally hosed with the Google Maps Engine right now,” he tells the lanky fellow behind him—John Amos, the founder and president of the nonprofit outfit. “It won’t handle any images bigger than 12 gigabytes. So we can’t see about 60 percent of the counties.”
Woods moves to an adjacent workstation to test whether the online interactive map they have been building, which is supposed to serve up the giant photos one small segment at a time, works for a county that Google did ingest. The window fills with a bucolic scene of rolling wooded hills, farmhouses, meadows—and flattened clearings blistered with clunky steel wellheads.
These 39 counties span the heart of the Marcellus shale, ground zero of the fracking boom. Drillers here have pumped liquids and sand into thousands of deep, horizontal wells to unlock the natural gas trapped in the tight rock. The fluid that comes back up includes toxic hydrocarbons, drilling mud backwash, and an ancient tea of groundwater steeped in minerals and metals. Operators store the stuff in freshly dug, plastic-lined ponds until it can be treated and, sometimes, reused. Regulators in Pennsylvania and West Virginia have opened their databases of drilling permits to the public. But these do not reveal where fracking ponds have actually been built and filled.
That information matters to epidemiologists who worry that the chemicals wafting off these wastewater ponds may be harming people who live nearby. To test that hypothesis, they need to know, year by year, where the ponds are. SkyTruth’s operation, if it works, will collect that intelligence.
Woods scans the rural landscape on his screen, searching for regular-shaped ponds near the well pads. As he clicks on each pond, the computer records its coordinates and drops a red X on the spot. He marks four ponds and then hits a submit button to transmit the results to SkyTruth’s server. A new photo appears. One task down, 41,399 to go.
If all sticks to plan, a crowd of hundreds of volunteers—many of them anonymous—will log on to SkyTruth’s website and use the pond-finder tool to complete those remaining tasks. Within a few months, the operation should deliver maps pinpointing every fracking pond that appeared in these parts of Pennsylvania in 2005, 2008, and 2010. When the U.S. Department of Agriculture releases updated aerial imagery from 2013, the recruits will do it all again in order to freshen the frackwater map. Then a final phase will digitize the exact boundaries of every wastewater pond. It’s an ambitious undertaking for a tiny nonprofit. But they’ve done it before.
In the summer of 2013, SkyTruth used this crowd-spying technique to map every well pad put into operation in Pennsylvania from 2005 to 2010. The goal: to produce a more accurate and comprehensive view than one could get from official lists of drilling permits. In a month, more than 200 volunteers performed 90,000 image analysis tasks. Amos says they pinpointed 1,400 active drilling sites. To ensure accuracy, SkyTruth shows every image to ten different analysts; Amos and other experts review split decisions.
Geospatial intelligence has long been the purview of highly technical specialists, usually working for the rich and powerful. No longer. SkyTruth and other groups like it are downloading highly detailed imagery free, or nearly so, from NASA, NOAA, the USDA, and the European Space Agency. Some have done deals with satellite companies such as DigitalGlobe, Planet Labs, and exactEarth to get photos rivaling those used by militaries—but at a cost they can afford with foundation grants. Google has been donating to nonprofits bandwidth, cloud storage, and access to its Map Engine and Earth Engine services through its “Geo for Good” program. Open-source platforms such as PyBossa make it feasible for a small group to harness the analytical pattern-recognition power of the crowd.
The democratization of these technologies could alter the balance of power for the resource extraction industries. Mining, drilling, fishing, and logging companies are all coming under surveillance by activists as the costs of remote sensing plummet.
In 2013, for example, SkyTruth and Space for All, another small NGO, flew a high-altitude hydrogen balloon over oil fields in North Dakota to collect reconnaissance video of flaring from Bakken oil wells. Those data were used to calibrate a new visualization that maps flaring all over the globe; the map is hosted by Google and updated nightly using data provided by NOAA.
THIS FOREST IS BUGGED
Rainforest Connection is a tiny nonprofit launched two years ago in San Francisco by Topher White. White, 32, is “upcycling” used Android smartphones into solar-powered spy gadgets, which he mounts in the canopy of protected forests. The devices record periodic snippets of audio, uploading the clips over the cellular network to his servers, which scan the waveforms for telltale sounds of illegal forest clearing: the buzz of chain saws or the rumble of trucks.
In 2013, White hid four of his bugs in treetops around 134 hectares of the Kalaweit gibbon sanctuary in Sumatra. Within a day of going live, White says, the system homed in on a group of loggers and sent out an alert. Sanctuary managers then drove off the interlopers. Within two weeks, he says, illegal logging in the sanctuary ceased completely. “The loggers quickly decided to move on,” White says. His next project, to be done in conjunction with the Zoological Society of London, will use 30 phones to monitor 10,000 hectares around the periphery of a large sustainable forestry reserve and elephant habitat in Cameroon for sounds of poaching or illegal logging.
Bit by bit, other activist groups have picked up on the power of remote sensing data and crowdspying. Earlier this year, the World Resources Institute launched a global, interactive map of forest clearing that is updated regularly. Other environmental startups have begun launching drones to shadow fishing boats as well as using cellular listening devices tucked into the rainforest canopy to detect and halt illegal logging or fire-clearing.
It’s not clear yet whether these tactics will sharpen the teeth of the watchdogs. But they are putting eyes on activities that have been largely out of sight—and often out of mind. Already, there are encouraging signs that when resource extractors know they are being watched, they behave better.
Amos followed an unusual path to environmentalism. A graduate degree in geophysics led to jobs interpreting satellite imagery for Big Oil clients. In one early project, he recalls, his task was to trace any sheen on the waters of the Gulf of Mexico, visible from space, to places where crude oil oozes naturally out of the seabed; these might point in turn to rich reservoirs miles underground. “It worked,” he says. Natural seeps provided clues to oil plays off the coasts of Brazil, Australia, Indonesia, and West Africa.
But Amos gradually grew disillusioned with the work as he scrutinized Landsat photos of Wyoming, where he had lived for several years, and saw sprawling tracts of prairie stripped bare to mine the coal beneath them. “I was seeing pictures I thought should be on the front page of the New York Times, but these images were only accessible to experts like me backed by deep-pocketed companies like Exxon,” he says.
Amos quit in 1999 and started SkyTruth to bring remote sensing know-how to the environmentally minded masses. “I saw so many compelling environmental stories in the imagery I was working with: deforestation, impacts from oil and gas drilling, from mining, urban sprawl,” he says. “I thought that anybody who cared about what was happening to the environment anywhere on the planet ought to be able to see that with their own eyes. And I knew we had the technology to make that available.” But the costs and learning curves were too formidable for most activists, and SkyTruth remained a one-man operation until 2008.
But in 2010, shortly after the horrific blowout at BP’s Macondo well, Amos caught the attention of the international media when he and Ian MacDonald, an oceanographer then at Texas A&M University, questioned the thousand-barrels-a-day spill rate that BP and the U.S. Coast Guard had estimated. When Amos and MacDonald measured the extent of the enormous slick on satellite photos, they calculated that, at a minimum, 26,500 barrels of oil were escaping from the well daily. A 2012 report by top scientists at several federal agencies concluded that the blowout, at its start, dumped more than 60,000 barrels a day into the Gulf. The media exposure opened doors—and checkbooks—for Amos. The Pew Charitable Trusts, Friends of the Earth, and other environmental NGOs soon started paying SkyTruth to surveil illegal fishing around Easter Island and oil flaring around Nigeria, among other projects.
The fracking pond–finder mission got started when Brian Schwartz, a professor of epidemiology at the Johns Hopkins Bloomberg School of Public Health, read about SkyTruth. In a meeting with Amos and Woods in Shepherdstown, he explained that his research group had collected complete digital medical records on more than 400,000 patients in Pennsylvania, many of them in the fracking-intensive parts of the state. They wanted to know whether air pollution coming off these well pads and ponds was sickening adjacent residents—in particular, exacerbating their asthma, promoting cardiovascular disease, or causing problems with pregnancies. But they needed one more piece to the puzzle.
The researchers had paid the state to obtain maps showing where unconventional gas production was happening. But they didn’t have locations, sizes, and dates of active use for all the fracking wastewater ponds. The conversation, in October 2013, ended with a handshake agreement. Just over a month later, the mission was nearly ready to begin.
When I arrive at the SkyTruth office on the morning of November 20, Woods is huddling with his colleague Teri Biebel over a screen showing the tracks of large ships across a swath of the South Pacific. “That’s a good one,” Woods says, pointing to a blip on the screen. Biebel clicks on the dot to pull up identifying details: it’s the Seiwa, a 95-meter cargo ship flagged to Kiribati. Like all large ocean vessels, it is broadcasting a collision-avoidance transponder known as AIS, for automatic identification system. Satellites operated by exactEarth are monitoring AIS signals from the Seiwa—along with those from about 100,000 other ships—and funneling the data to paying customers in nearly real time. Thanks to backing from Pew’s program to counter illegal, unregulated, and unreported (IUU) fishing, SkyTruth is among them.
The software shows that the Seiwa pulled up next to other ships four times within the past six days. “This is a reefer, a refrigerated transshipping vessel,” Woods says. Fishing boats, most of which are invisible to AIS because they are permitted to turn off their transponders, sometimes offload their catch to reefers. The tactic enables them to remain longer at sea—and in some cases to evade catch limits or to slip into waters for which they lack permits. “We’re trying to target some of those events with radar,” Woods says. “As soon as the reefer stops, we’ll place an order for a satellite image there—and you hope that by the time it is taken, the ship is still in the frame.”
When SkyTruth scanned nine months of radar and AIS data in an attempt to estimate the amount of “dark fleet” activity around Easter Island, they turned up 42 boats running AIS-silent, about half of them suspicious, Woods says. According to Matt Rand of Pew’s Global Ocean Legacy program, that was enough evidence to persuade the Chilean navy to start running air patrols around Easter Island. Pew and others have been lobbying to get the surrounding waters set aside as a preserve. That legislation passed but has yet to be signed into law. Nevertheless, Rand says that SkyTruth’s work in Chile “has been instrumental, strategic, and extremely important to advance the largest marine protected area in the world.”
But because IUU fishing is such a massive problem, says Tony Long, who recently retired from 27 years of running surveillance operations in Britain’s Royal Navy and now leads the IUU project at Pew, surveillance must scale up dramatically to put a real dent in it. With hundreds of thousands of vessels plying the oceans daily, catching cheaters requires automation.
Amos teamed up with Analyze, a company founded by ex–defense contractors, to see whether their engineers could work out ways to automate the detection of interesting behavior. Militaries use such software to monitor suspicious movements of aircraft. By processing all AIS data collected from the entire South Pacific over the course of a year, Analyze has now developed algorithms that flag poachers or the cargo ships that collude with them.
This past summer, Amos says that work has now begun on “a big-data project with Google and Oceana—code-named Pelagos—to map all of the commercial fishing activity in the ocean that can be detected using satellite-collected data. We’re hopeful that this visualization will inspire people to ask questions about how and where our seafood is being caught, and how effectively we’re managing our vital ocean resources for the future.”
The morning of the pond-finder mission launch, Amos and Woods arrive at their office to find that their Internet link is dead. The landlord had neglected to pay the cable bill. Amos dispatches Biebel to the cable company with a wad of cash, but she gets into a fender-bender en route. So they spend the day working at the Mellow Moods café, elbow to elbow around a small table jam-packed with three laptops. In the early afternoon, after more debugging, Woods presses the go button and releases the pond-finder into the wild.
Amos confesses that he doesn’t know how many volunteers will show up to do the work. “Our dream is that SkyTruthing becomes common, but we don’t know whether that is realistic,” he says. “How much are people really going to be able to deal with imagery like this? How much will they want to do?”
Over the next two months, 131 citizen spies chip in. The most dedicated participants analyze more than 4,000 photos each; a few hundred is more typical. Together, the volunteers drop 35,196 red X’s, which pinpoint 7,835 ponds around active well pads. Some fraction of those are farm ponds or swimming holes. So within days of completing the first phase, SkyTruth kicks off the second, in which the amateur analysts view close-up photos of each pond and classify it as frackwater or not. By April the results are complete, and Woods sets his crew to updating the database with fresh images collected by USDA planes in 2013.
When the classification is finished later this year, Amos will hand off a thoroughly vetted set of digitized frack-pond boundaries to Schwartz for incorporation into his air pollution study. While that analysis is under way, Amos and Woods plan to extend their fracking surveillance—crowd willing—to the well-pocked hills of West Virginia and Texas.
The bigger the target, Amos hopes, the larger the crowd that will muster. “In the long run,” he says, “the goal of SkyTruth is to motivate people to use satellite imagery, geospatial data, and digital mapping tools to personally investigate what’s happening—good and bad—in the place that they care about. If they share what they find, there is then the potential for people to self-organize and do something that wasn’t even on our radar.”