Q&A: What could cheaper satellite imagery do for the SDGs?

DENVER, Colorado — When nonprofit leaders enter the headquarters of DigitalGlobe, they often wait in the lobby beneath a glass ceiling shaped like a satellite dish, then walk into their meeting asking what can be offered for free. “More often than not, development folks will approach DigitalGlobe or other companies expecting us all to act in the same way — that we are a funder, or that data philanthropy is our approach to the community, and everything should be free and this is why,” said Rhiannan Price, director of DigitalGlobe’s Global Development Program, a commercial vendor of satellite imagery and geospatial analysis based outside Denver, Colorado. “They [NGO representatives] tee up the conversation with: ‘This could be really great for you guys for PR.’ And it’s like, ‘I completely understand where you guys are coming from. Let me explain to you our theory of change and why we do it the way we do.’ I don’t think they’re used to working with companies that have dedicated teams that work with not-for-profit organizations.” DigitalGlobe owns and operates high-resolution earth imaging satellites, and donates imagery and analytics to just three nonprofit organizations: The Jane Goodall Institute, the Amazon Conservation Team, and Team Rubicon. The company prefers to give discounts rather than donations, offering discounts to organizations working in areas such as disaster response, agriculture, and global health. Price says the arrangement means these global development partnerships are neither public relations nor corporate social responsibility, but rather core to business and a priority for the company. Price spoke to Devex about how she works with organizations to leverage satellite imagery for global development. The conversation has been edited for length and clarity. Can you tell me more about your role and how you balance the potential for revenue with potential for impact? My role at DigitalGlobe is leading our global development program. What that actually means is, I Iead a team of folks who are really dedicated to supporting global development and humanitarian partners in using DigitalGlobe’s products and services. My team is seven people at this point, but we also tap into our other teams who support business writ large. A lot of what I personally do is, at the highest level, bridge the gap between what’s technically possible and what is needed by development practitioners. What DigitalGlobe does is enabling what we call our purpose, which is “seeing a better world.” For us, that’s not just a tagline. It really is part of how we make business decisions, and interact with our customers and our partners, and how we approach the transformational change that we hope to enable across the world. We act as a geospatial adviser and a real partner to organizations so that we can guide them in terms of best solutions for their use case. There are times when DigitalGlobe products and services can play a part and there are other cases when we’re not the fit-for-purpose solution. We try to really be an honest broker in terms of what makes sense and what is right by our development organizations because they don’t have a ton of capital to invest, particularly in cutting-edge technologies. We want to make sure we are being good stewards of that investment. There are conversations when we say, “DigitalGlobe is overkill.” A lot of folks jump to the fact that they think they want 30-centimeter resolution imagery even if a 2-meter product will do … There are a lot of examples where there are vector data [available] off-the-shelf we point people to. A good example of that is night time lights data. There’s a publicly-available night time dataset, so when people come to us for electrification products to try to understand demand within a community, rather than [directing] them to try to understand it via optical, we’ll point them to that as a good proxy. The GovLab at New York University has this term “data stewards.” It’s this really cool idea of these individuals you find — particularly within tech companies and companies that have data or content that would be useful for the development world — who for some reason have a social impact mandate, or work in the global development space, or somehow, we come across these types of use cases and networks; they think of me as one of these data stewards. DigitalGlobe donates imagery and analysis to three nonprofit organizations. But most work with global development organizations centers on discounts on imagery and analysis, for example through your GBDX program. Can you tell me about one of the partnerships you’re most excited about, and how it demonstrates the way DigitalGlobe likes to work with global development organizations? Our partnership with the Gates Foundation started five years ago, and it was born out of an initial need from their polio team to understand where remote populations were located in places with new cases of polio. We were looking at remote and insecure areas of the world, such as Northeast Nigeria, Somalia, rural parts of Afghanistan, and Pakistan. They had no alternative in terms of understanding where populations were because the statistics were off — there were no current maps over those areas, and a lot of times things were changing because of displacement, because of conflict, because of the usual drivers of change we see in those parts of the world. At the time, we were just starting to do more on the analysis side, so we were evaluating with them: “How can we use cutting-edge machine learning AI techniques to try to get to your solution a little bit faster?” What we started with was just a detection question which was — “yes or no, in this imagery grid cell, is there a building or a potential settlement?” And we trained the model over remote areas of the world to do this detection. What we did at the time, in terms of brokering our partnership with Gates, was to say, “let us take on parts of this workflow for you to see if we can accelerate it, and do it in a way where we develop the product in partnership with you guys. And we can do it as something that can be crosscutting and helpful across your programs, not just for polio.” “We recently did all of Tanzania in a matter of three weeks. And that’s 18.5 million building footprints that we automatically extracted over an image mosaic of the entire country, which is nearly a million square kilometers.” --— Rhiannan Price, director of DigitalGlobe’s Global Development Program So their malaria team came on board, and their neglected tropical disease team, and it was not only detecting where buildings were but also extracting the footprints of those buildings and being able to do it rapidly. There were all sorts of iterations of the approach over time. After the first model, where we were doing the search reduction step, or the detection step, we tried crowdsourcing the footprints. We partnered with Humanitarian OpenStreetMap Team for a project for the malaria team, and again it was successful, but it took us eight months to map 500,000 square kilometers and 4 million buildings across seven different countries. We realized we could do even that piece of the workflow much quicker and then free the crowd to do more in terms of the attribution, validation, and field work. That takes us to where we are today with the Gates Foundation, which is, we can deploy these machine learning algorithms across entire countries and regions for them. We recently did all of Tanzania in a matter of three weeks. And that’s 18.5 million building footprints that we automatically extracted over an image mosaic of the entire country, which is nearly a million square kilometers. I have been following efforts such as Radiant.Earth, which advocates for more open geospatial data for global development. But I also know that neither you nor your clients want this imagery to get in the wrong hands. How do you think about licensing with global development organizations now and moving forward? For development partners, licensing is such a tricky subject, particularly because it’s not something development folks have to think about very often, and most often folks see it as either open or not open but the reality is there is a huge spectrum of data licensing. There is a spectrum of open. There is open but for non-commercial use, which more imagery providers feel comfortable with, particularly because we acknowledge that these organizations are not trying to make a profit with the data [but] that they really are impact-oriented. And there are opportunities in which we can go fully open. We just take them by use case and try to understand: What is the ultimate outcome that our partners are trying to have and how can the license enable that? The reality is there are a lot of folks who are, for a lack of better term, bad actors — who will abuse even a permissive and development-friendly license — and we see groups who will either blatantly ignore or abuse licensing terms. We have to be cognizant of that and anticipate it in terms of how we provide the data, or who is making the data available. If there is a platform where imagery is available we have to make sure that they are also taking on the regulatory risk, and make sure that security is accounted for and that bad actors aren’t accessing the imagery. One thing that is really mission critical in terms of enabling some of the development outcomes is the ability to share data and work against a common operating picture. On any given development project, there are many organizations involved, from the funding to the implementation to the oversight. To only provide data to one of those organizations is, in some ways, detrimental to the overall mission. Licensing is a really tricky subject, but it’s also not in a lot of people’s wheelhouse if you don’t have to deal with this on a daily basis, so for us trying to make the license as straightforward and intuitive as possible is important. We’ve heard, particularly in times of disaster response, if governments haven’t already included imagery into their workflows and they come across “publicly available” imagery, they will be hesitant to use it if there’s any risk whatsoever around that license. So having something that is transparent and obvious in terms of the usage or the attribution that is required is really important, and I think often-time overlooked by providers, because they’re really just looking out for the risk. Read more about the impact of satellites in development work on our Satellites for Sustainability site.