SkyTruth has a mission to use remote sensing and digital mapping to educate the public and policymakers about the environmental consequences of human activity. Sensors & Systems (S&S) special correspondent Matteo Luccio recently spoke with John Amos, president of SkyTruth, about the organization’s motive and mission. Included in the conversation are details on imagery sources, the types of analysis and mapping products the organization produces, as well as details on some of their more impactful mission to make this kind of monitoring more commonplace.
S&S: What sort of things do you look for? How do you prioritize?
Amos: Our approach is to look for areas where our knowledge of environmental issues, combined with our knowledge of spatial technologies, comes together to identify issues or incidents where we think we can make a difference, generally within the realm of shedding light on these issues for the interested public. Some audiences within that for us are, obviously, policy-makers and decision-makers, the media, and other people who have more of a megaphone and that can help take whatever we produce and push it out to an even wider audience than our limited capacity allows.
S&S: There are a few steps between the satellite collecting the data and the public consuming the information. You collect it, analyze it, and feed it to the right sources for distribution?
Amos: Yes. In some of our projects, public interest groups—such as citizens’ groups, environmental groups, or humanitarian groups—come to us and they say, “Hey, look, we think we have a problem here.” I’ll give you an example: the flaring of natural gas related to oil operations in the Niger Delta.
“We think there’s a public health impact on the villagers living in those areas. We’ve been working with the big multi-national oil companies to try to reduce this problem and they say that they’re doing it and sticking to our agreements, but we have no idea how to independently assess whether they’re being truthful about that.”
They’ll come to us with a problem like that and we’ll say, “Well, it turns out that there’s a relatively new satellite called Suomi National Polar-orbiting Partnership (NPP) satellite that was launched, I think, last year, and it has an instrument on board called VIIRS that does very sensitive night-time imaging and is able to capture sources of illumination at night at a level of detail that we’ve never had available on the civilian space side.” We’ll take a look at that and see if we can measure the amount of flaring that’s happening right now in the Niger Delta, and pinpoint where it’s happening.
We have that kind of engagement, where people come to us directly with a problem, and if we think we can solve it, we do. Part of the beauty of being a non-profit is that we don’t have anything to sell, we don’t have to meet a monthly sales target. So, if we determine that the satellite approach isn’t really the best or that our tools really aren’t the best and that you’d be better off doing something else, we’re free to tell you that. Very often, we end up telling people, “Look, you know, we don’t have what you want for this problem, but these other folks over here do.” Or, “maybe you’d be better off just to rent a small airplane and get up in the air with a video camera. You’d spend less money, get better imagery, and get the job done.” So that gives us a certain amount of freedom in the way that we operate.
On the other side, we’re constantly following the news and environmental issues. My background as a geologist and ten years of work as a consultant to the oil and gas and mining industries gives us a leg up in expertise on fossil fuel-related issues. You’ll see from our work that we skew heavily toward addressing the fossil fuel extraction, transportation, and processing side of the environmental issues that we engage with. So, we choose to engage on our own, again based on our own internal expertise and understanding of what’s happening with the public dialogue on these energy-related and climate-related issues.
S&S: As demand for your services grows, are you growing as an organization? Are other organizations paying you for these services, allowing you to hire more staff? What’s your plan?
Amos: That’s a really interesting question. Eleven years ago, when I formed SkyTruth, I had to make this decision: did I want to incorporate SkyTruth as a for-profit business or as a non-profit business? That took a lot of thought and I occasionally still have to answer the question, “Why didn’t you just do this for money?” Right? Part of the answer comes down to being able to tell people, “Look, we’re not the best solution for the problem that you have.” When I was in private practice, my consulting firms were always the best solution for everybody’s problem. We’d take a job and figure out how to do it later. We don’t have to put ourselves in that box.
The other thing is, I’m not an empire-builder. That’s not what I’m interested in. What I’m interested in is enabling as many other people on the planet as possible to do as much of the stuff that we do here at SkyTruth as possible. In other words, growing skytruthing as a movement, as a verb, as an activity, the way Googling has become a routine, commonplace activity. You could call this the visual and geospatial equivalent of Googling. Googling is researching a topic, right? This is kind of a subset of that—visually researching a place or an issue.
SkyTruth as an organization has a bit more growing to do to be able to effectively accomplish that. Now, we’re four full-time people and a couple of paid interns. Four or five years out, I envision us being maybe about ten or twelve full-time staff and interns. I think that should be enough to do what we want to do: staying on top of changes with the technology, with geospatial technology, with satellite imaging systems, demonstrating the applications and the value that those new technologies bring to environmental issues and to environmental investigations. Then making it easy for other people to engage directly with those technologies and do skytruthing for themselves. We don’t have to cover the planet. It’s still a pretty big planet and, you know, it’s hard to cover it all from Shepherdstown, West Virginia.
S&S: To what extent are you developing or do you envision developing tools or applications that people can use to do for themselves what you’re now doing for them?
Amos: The advent of Google Earth was a great step forward in the democratization of spatial technologies and it really laid the foundation for skytruthing by anybody. Anytime anybody goes to Google Earth and enters an address to see what’s going on there, they’re doing a little bit of skytruthing. However, we need to take it farther than that, because there are multiple satellite-imaging systems that are producing imagery on a daily basis that is free to the public, if only they knew about it. And slowly it’s getting easier to find these images and to use them. They’re starting to be produced for the public in a relatively simple format to interact with, but there’s still a long way to go.
We see a little bit of toolkit building is necessary—for example, to bring a high-resolution aerial survey photography that you can commonly interact with on Google Earth, but to bring the latest versions of it and historical archive versions of it to your desktop in an easily-managed form, applied to a specific issue of interest. That’s the key. You can build all kinds of tools—desktop tools, Web tools—to make stuff available to people. But if you don’t frame it within the context of a specific issue or subject that they’re interested in, most people will never engage. If you say, “Hey, we’re going to bundle imagery for you, and give you the tools to help us study this problem of fracking for natural gas across a wide swath of your state,” then people will come, and they’ll start to engage.
We’ve seen that with our very first crowd-assisted image mapping and analysis project called, Frack-finder. We launched that product right about the time an article came out on us in The Washington Post, a few months ago. We packaged high-resolution imagery, delivered it to people in small, consumable bites, and asked them to do a very simple thing: tell us if they saw a drilling site in the middle of that little chunk of imagery. We wanted to map all of the drilling sites across the state of Pennsylvania, over three years of imagery, and we wanted at least ten people to look at every image of every drilling site. That worked out to about 90,000 individual image-analysis tasks. The project has been completed. In three weeks we had more than 200 volunteers engaged and do these 90,000 image-analysis tasks with a surprising degree of quality, as measured by the level of agreement among the analysts.
We’re encouraged by that. We want to see how far we can push that and whether we can get people to do more sophisticated things. Ultimately, can we do that for multiple different environmental issues, in multiple different geographies, harnessing different streams of imagery depending on what spatial resolution, what spectral resolution, what other attributes of the imagery we need to have to answer the questions we want to answer?
S&S: Which satellites do you use?
Amos: For very wide-areas daily reconnaissance— this is usually looking out across the oceans for large-scale pollution events—we’re using the MODIS sensor on the Aqua and Terra satellites operated by NASA. That’s a great tool. It’s extremely low resolution, but has twice daily coverage of pretty much every place on the globe, which is very handy. Going to higher resolutions, Landsat 8 is operating spectacularly. We are thrilled. It is sending down very high quality datasets, hundreds of images scattered across the planet every day. We can’t tell it where to point and shoot, but there’s a lot of serendipitous value just in analyzing what it is collecting on a daily basis. It has a 25-meter resolution, so it is able to do a lot of the mapping that we’re interested in—such as mapping oil and gas drilling sites, mining activities, especially coal strip-mining here in Appalachia and mountain-top removal mining.
We’re also looking at ocean events in more detail. There are some chronic pollution sources that we’re able to see in detail with the Landsat imagery to try to keep people honest in terms of what they’re reporting to the public. Those are common, but we also bought imagery from commercial systems—such as the French SPOT system and Canada’s Radarsat system. We’re using Radarsat on a year-long illegal fishing monitoring project that we’re doing for Pew Charitable Trusts. We use SPOT to map the growth of drilling infrastructure in Wyoming for some scientists who are correlating that drilling infrastructure to the movement, behavior, and survival of mule deer out in Wyoming. That’s a multi-year project that we’ve been doing, I think for nine years now. We also capture imagery from a variety of other satellite systems, depending on what’s available and what the needs of the project are.
S&SL What types of data do you use and what types of analysis do you do?
Amos: We’re kind of a two-headed beast here. One head produces imagery, purely for communications purposes. Typically, that’s visible satellite imagery, sometimes combined with infrared images to enhance the vegetation or other aspects of the image that you’re interested in showing people. Then there is also scientific imagery, where typically we’re acquiring imagery in infrared and thermal infrared, as well as visible parts of the spectrum.
A good example are the flaring studies that we’re doing, where we’re looking at illumination in the visible spectrum, but we’re also looking at temperature information coming out of the thermal channels to discriminate natural gas flaring from wildfires, brushfires, and normal electrical lighting illumination. So, in terms of imagery, that’s the kind of data that we’re collecting.
We also collect a lot of complementary geospatial data to help our analyses. With oil and gas drilling, we’re collecting on a daily basis from several states. We’re collecting information on drilling permits that they’ve issued and other information on stages of drilling activity, such as SPUD reports, each of which indicates that a drill bit has just started to cut the ground, completion reports indicating that they’ve done a fracking operation at a drilling site, mining permits, and other things that help us provide context to what the stages of construction activity are that we’re seeing in the imagery.
One of our big motivating factors here at SkyTruth is that we think that datasets should be more easily accessible to the public and that when you make them more easily accessible, all kinds of good stuff that you couldn’t predict happens as a result of that. So we’ve been routinely collecting data on the chemicals used in hydraulic fracturing operations at oil and gas drilling sites and compiling that into a database that we made available to the public, which has allowed researchers to start using the data. Researchers from Yale and Harvard have started analyzing all kinds of things in the patterns of chemical use, water use, other attributes of drilling operations that we couldn’t do on our own. So, we’ve worked to make datasets more easily available to the public purely to enable that kind of scientific research to take place that wouldn’t happen otherwise.
S&S: How much analysis do you do?
Amos: Lots of our project work is simple overlay work, Map-making 101. Are there datasets of interest that you can lay on top of an image and have the image provide visual context to those data points and polygons? A lot of it is very simple and a lot of that work is done within Google Earth. It’s a very easy-to-use platform for creating that kind of simple map and publishing it.
However, we do more sophisticated analyses. For example, we do land-cover analyses, to measure the amount and rate of mining activity throughout Appalachia and to map the direct landscape and habitat footprint of drilling activity. Out in Wyoming and in other places where it’s happening, we do fairly sophisticated analyses of this thermal data to try to identify natural-gas flaring versus all other sources of illumination that are appearing in that type of imagery. So, it runs the gamut from no analysis to some relatively sophisticated image processing and image analysis.
S&S: What are some surprising things you’ve found out?
Amos: Probably the most surprising thing we saw was our BP oil-spill story. This is another example of the value of making this data more widely available to the public, because there are experts out there on various issues, who may not be experts in image processing and remote sensing, but can provide value in terms of analyzing what you’re seeing.
I think the BP spill was a great example of this. When the spill happened, we immediately started collecting all the imagery we could get our hands on to show the oil slick and to make maps. Our original motivation was a desire to do something to help. By producing maps from these images showing where the oil slick was, relatively how thick it was, where it was moving day to day, the clean-up responders might be able to use that information and do something with it.
Very early on, we got a call from one of our colleagues, Ian McDonald, an oceanographer who spent his career working in the Gulf of Mexico. He said, “Hey John, I’ve been looking at your oil-slick maps and the size of the oil slick on these images just doesn’t jibe with how big they’re saying the rate of oil flow from the well is.” That was a surprise! I really didn’t expect that you’d be able to deduce that from satellite imagery and that the company that owned the well that had failed could possibly be so far off in their estimates of how much oil could be coming out of it.
That didn’t really make sense to me, as a geologist who’s hung out with petroleum reservoir engineers, who live and die by the numbers of reservoir pressure and permeability and all kinds of data that you’d want to know if you’re going to drill a well in that reservoir. So that was a surprise. As it worked out, with Ian’s help, imagery made a pretty valuable contribution in this case, grappling with the size of this spill and coming up with a much more accurate estimate of how bad it was. Even working with imagery, it was a surprise to me that imagery could be that useful on an issue of this magnitude.
S&S: In the future, they will no longer be able to assume that nobody would know one way or the other.
Amos: Absolutely. That’s a really positive-impact story. Getting that stuff out into the public domain, so that people like us could publish it and people like Ian could see it, leads to a game-changing leveling of the playing field. Even a hundred miles off shore or a thousand miles off shore you can’t hide what you’re doing. You now have to operate with the expectation that somebody like SkyTruth is going to see what’s happening and is going to go public with it. I would like to think that’s a very positive game-changer in terms of how people are becoming much more scrupulous about how they’re operating and much more honest when things go wrong. Because we’re humans, so we’re going to make mistakes, we’re going to have accidents. But let’s be honest and forthright about them and get to work fixing them rather than arguing about whether it happened or how bad it was, or any of that trivial stuff.
S&S: In what ways and formats do you make the data available?
Amos: Since our main audience is the general public, we try to make this stuff available as simply as possible, in easily-digestible JPEG images. In the mid-1990’s, when I first started thinking about starting SkyTruth, I was seeing satellite images that were showing these fascinating environmental stories, such as the vast amount of clear-cutting that we were doing in national forests out west. I saw the amount of forest destruction from clear-cutting compared to the amount of forest destruction from the Mt. St. Helens eruption. That was an eye-opener for me. I realized that we were doing much more damage to the forest with this deforestation program that taxpayers were funding than Mt. St. Helens had done. Everybody was crying about what Mt. St. Helens had done to the forest and it was truly apocalyptic-looking, but when you looked at a satellite image that was overwhelmed by what the US Department of Agriculture was doing to our forests.
I started seeing images like that—for example, images of very intensive drilling on public lands, giving the lie to the myth that more environmental-friendly drilling techniques were being used—and I was saying to myself, wow, these images should be on the front page of the New York Times and The Washington Post. You know, they tell the story without you having to say any words, and this is a very powerful thing. So, why am I not seeing these things scattered all over in the media? I started researching why that was so. In an effort to change that, you know, we want to make everything we do available, basically for free, in an easy-to-use form. So, Web-friendly graphics and print-friendly graphics are our main product.
Beyond that, for our more analytical projects—like land-cover analysis for mountaintop mining over the decades, land-cover analysis for drilling footprint impact over the decades—that is hard-core GIS data, so we make that available to people in GIS formats: shape file, Arc export format, whatever it is they want to use to do actual geospatial research with the data. That’s how we will provide it. In terms of imagery, when there are no licensing restrictions, we’ll provide it in pure binary raster format, if that’s what they want. We’ll provide them with contrast-enhanced, spatially-rectified GeoTIF format, if that’s what they want. It depends on who we’re working with and what they need, what their software is and how they want to use it, but we can provide it in pretty much every portable raster image or GIS-compatible format that people need.
S&S: And KML, since you were talking about Google Earth.
Amos: Increasingly KML and KMZ, yes, absolutely.
S&S: What kind of technical assistance do you provide?
Amos: With a full-time staff of four we don’t have really the bandwidth to do very systematic training in a formal sense. On a daily basis we are providing technical consultation and training to people—through email, over the phone, even by Twitter. It’s kind of astonishing, but sometimes it’s as simple as just pointing people to data sources that they weren’t aware of. Sometimes it’s telling them a more advanced sequence of steps, such as how to download a satellite image from a NASA website and get it into Google Earth Pro.
In terms of the more sophisticated stuff, we’re just pointing people to the online tutorials and courses that already exist, that do a pretty fine job actually. So I’d say that in an informal sense, probably anywhere from 15 to 20 percent of our time is hand-holding—Geo 101, Remote Sensing 101, Mapping 101, all the way up to what I would characterize as a university undergraduate level sophistication of consultation and training.
S&S: What is it worth for activists to invest the time to learn themselves, given the free tools available, and what should they leave to experts, such as SkyTruth? If you were to put together an image analysis toolkit for activists, what would it include?
Amos: That’s a work in progress. At a bare minimum, I would encourage almost everybody in the non-profit world to apply for a Google Earth Pro license. That is a fundamental starting place, to get a very easy entry into exposure to imagery and geospatial data. The Pro license offers you the ability to take data sets provided by people like SkyTruth, in Shapefile or CSV or other common, portable formats, and overlay those within Google Earth and interact with them within Google Earth. So, if I were to build a one-size-fits-all toolkit, that would be, for us, an indispensable item in that toolkit.
Additionally, the toolkit is a list of links to online tutorials on how to interact with imagery and basic explanations of what satellites do, and what kind of satellites are up there and how they’re accessible right now. There are some good online tutorials. NASA has a decent online tutorial for that stuff. Then it would have specific links to the main, free public image data sources that I’ve just mentioned, with basic information on how to find those images that you want, download them, and get them into Google Earth Pro, so that you can then interact with them and make simple maps.
Beyond that, it gets tough. You get into the realm of custom image processing and image analysis and that’s always a function of specifically what questions you want answered. What problems are you trying to solve? That’s where the generic processing that’s applied to imagery in Google and Bing only gets you so far. They’re trying to make an overall, cosmetically good-looking product, which means that they are necessarily truncating and throwing out a certain amount of information that they would consider noise but that might be exactly what I was looking for.
Depending on the problem, you have to have a fairly customized approach at some point. That’s where I think we’re going to make headway in terms of building skytruthing projects for specific issues that get the crowd to engage and produce a public product that all the crowd can use at the end with some measure of quality assurance built into it, so that individual organizations won’t have to build their own in-house expertise to create that product. That’s the trick. At some level, there are diminishing returns for small non-profit organizations to bone up on the ins and outs of land-cover classification and image segmentation and feature extraction. Those things start to get too out in the weeds for a small group to dedicate time and effort to. However, if they let us know what projects they want to have done, we may be able to combine our own expertise with the assistance of the crowd to get that work done and out into the public domain where everybody can use it.
S&S: Has shutter control ever been a problem? Do you expect it might be?
Amos: Shutter control is a non-issue. Anybody with a credit card now can buy what ten years ago was considered highly-classified spy satellite data. You could just go to the Digital Globe Web site, break out a credit card, and buy 40-centimeter detail imagery of pretty much any place on the planet—including US military facilities, government facilities, and other countries’ facilities. It’s more or less all there for the taking. There still are some exceptions, but in most of the world where we care to engage on environmental issues, it’s really a non-issue.
Two commercial drone operations were licensed in the United States just a couple of weeks ago. That is the beginning of a new era. Drone prices are dropping and capabilities are increasing. You can buy a pretty simple drone for yourself for a few hundred bucks and launch it out of your backyard and have a camera on board to collect imagery of your neighborhood or wherever else you’re flying it. As commercial licenses proliferate, the common use of drone systems for a whole range of applications is going to increase. Society will have to have a debate about this. It certainly raises privacy issues about what’s going to be licensed and how, but the writing is on the wall.
This technology is going to become more and more ubiquitous, not only for commercial uses, but for use by hobbyists, by the casual person who’s got a few hundred bucks to spend and wants to fiddle around with it. In that sense, shutter control becomes even more difficult.
In the rest of the world, people are going to have access to this technology, too. We just had a crew out in North Dakota that did a high-altitude balloon launch over the Bakken oil drilling and flaring out there. We had four cameras—a couple of GoPro cameras and a couple of still cameras—hooked up to a high-altitude weather balloon that flew up to 100,000 feet and collected imagery that we’re just starting to work through. So, all kinds of alternative imaging platforms are becoming available and that just makes shutter control an almost impossible thing for governments to exercise.
S&S: Do you ever worry about what you are not able to see? Will skytruthing lead to efforts by governments and corporations to conceal some of their activities?
Amos: I’m already so worried about what I can see that I can’t afford to be worried about what I can’t see! Google Earth and Bing Maps and all the weather satellite imagery on your nightly newscast combine to give the public a common perception that we’re constantly under surveillance and that satellite imagery of every spot on the planet is ubiquitous on a daily basis in high resolution. That is absolutely not the case. There are still huge gaps in coverage. It surprises people when they dig into Google Earth and look at a high-res image and realize, “Hey, that car parked in front of my house is red, but I’ve had a blue car for three years now.” Then, “Oh, that image wasn’t taken yesterday. It was taken three years ago! Oh, I didn’t know this stuff was that old!”
We get requests from people all the time that are predicated on their assumption that we can snap our fingers and download a high-resolution image taken today of that protest on the mall in D.C. and it’s just not the case. There are many things that could be happening out there that we are not aware of and that, even if we were aware of, have already happened and nobody broke out the credit card to order a satellite image of that event, so that satellite image never came into being. That event was missed.
We are very concerned, especially in terms of routine monitoring or areas where we know we have a lot of industrial activity happening. For example, in the Gulf of Mexico there are some 3,600 off-shore oil and gas platforms and tens of thousands of miles of pipeline on the sea floor, sprawling across a huge area ranging from the shoreline to more than 100 miles off shore. Nobody is monitoring that with imagery on a systematic basis. The only way our government knows what’s going on in terms of pollution out there is what the polluters themselves report to the government. That’s it. We’ve seen in the past that that’s not a very reliable source of data. Our colleagues at Florida State University did a study recently comparing what they saw in satellite images of the Gulf to what the polluters reported. They were able to conclude that the size of the slicks that we saw in images ranged from about 10 to 19 times larger on average than the size that was reported through official channels by the polluters to the federal government.
What we’d like to have is routine imagery, because we learn about these pollution events usually after the fact. If you had a satellite image in an archive that you could compare to that event, you’d have some way of enforcing accountability and credibility on the polluters, and they would know that. I guarantee that they would start reporting things more consistently and more accurately. It would be great for everybody to have solid information upon which to build regulation, inspection programs, and public policy. We don’t have that simply because we don’t have the threat of ubiquitous, routine satellite imagery.