MATTEO LUCCIO/ PALEBLUE DOT LLC / WWW.PALEBLUEDOTLLC.COM
Open source geospatial software—spatial data management, GIS, and related developer tools and end-user applications delivered with an open source license—has co-existed with proprietary geospatial software for decades. Software developers and users often mix and match the two. The Open Source Initiative developed the definition of open source license, which consists of ten criteria. Two of these criteria are:THAT THE SOFTWARE MUST BE FREE TO REDISTRIBUTE, AND
THAT EVERY REDISTRIBUTION MUST INCLUDE THE SOURCE CODE.
Unlike proprietary software, which is written by a development team from a single software company, open source software is written by a community of developers from all over the world, some doing this as part of their paid jobs and some as volunteers. Their contributions and bug lists are managed by a steering committee. Local groups can immediately make changes to the code, but it may take a while for those changes to be accepted by the steering committee and placed into the current version. Like proprietary software, open source software is commercial in that companies can profit from it by offering services (such as installation, customization, or training) or related software.
I discussed the interplay between open source and proprietary geospatial software with:Anne Hale Miglarese, Founder and CEO, Radiant.Earth Andrew Dearing, CEO, Boundless Spatial, Inc. Andrew Turner, Director & CTO, Esri DC R&D Center, Esri
Radiant.Earth uses open source software and data to enhance and expand the geospatial market for the global development community. Miglarese has been in the geospatial industry for 30 years, working at the federal and state government levels, and with commercial enterprises. Prior to starting Radiant.Earth, she was the CEO of an airborne mapping, remote sensing, and GIS services company called EarthData (now known as Fugro EarthData), as well as the CEO of PlanetIQ, a company that is fielding commercial weather satellites. “What I’ve never done before, she says, “is work in the not-for-profit sector, which is what Radiant.Earth is.”
The idea behind Radiant.Earth emerged about two years ago in discussions with Peter Rabley, an investment partner at Omidyar Network. “Peter had an idea about how impactful it would be to organize the world of open data,” says Miglarese, “because the data are fragmented and also very costly, difficult, and time-consuming to acquire. Often, satellites are built at great expense, but primarily for principal investigators in the science community, and the data they collect does not find its way outside of that community.”
Radiant.Earth was publicly launched at the Thought Leaders Forum, hosted by the Gates Foundation in Seattle in February 2017. Its vision is to aggregate the world’s open Earth imagery (collected by satellites, aircraft, and drones), make it easily discoverable and accessible, and allow people to compute on it either within its cloud or by downloading it to their environment. “While our user community may be very broad, our primary focus will be on supporting the global development community, in line with the mission of our primary investors,” says Miglarese. While an increasing number of global development specialists use geospatial tools, there are few resources to help that effort grow. For this reason, Radiant is also helping to develop this community.
Radiant.Earth will not house commercial data with a restrictive license because it cannot currently monitor and regulate what happens with it downstream. It does, however, enable its users to discover relevant commercial imagery. “This model is a win-win for everybody because we understand that sometimes the problem is best solved with commercial sources of imagery,” Miglarese says. It will be building its platform with open source tools, in partnership with Vizzuality, while Azavea is building its imagery pipeline and analytics tools. “While our platform will be built with open source technology, we anticipate that many of our users will continue to use their commercial software workflows to analyze the data discovered on Radiant.Earth.”
Radiantís primary focus is on the global development community and organizations with a mission to deliver their services and products at the local community levels. Its platform will be open for anyone to register and use and Miglarese anticipates seeing some commercial traffic on it. “We’ll certainly see education and academic research institutions, as well as traffic from a broad swath of the conservation, environmental, and global development community,” she says. Innovation in satellite imagery and analytics enable faster and more accurate change detection, she points out. “It is going to meet many needs for the international and global development sector.”
Radiant has developed a partnership with Amazon Web Services, which already hosts Landsat and Sentinel data, and aims to expose additional open datasets. It has advanced formal relationships with the European Space Agency and the United Nations Institute for Training and Research (UNITAR) through its Operational Satellite Applications Program.
“Clearly, there are valuable commercial archives out there,” Miglarese says. “We have selective interest in older satellite archives, which we may be able to add to our platform and expose to users for greater use. I think they’ll be particularly interesting for conservation, climate change, and development efforts.” Additionally, she hopes that Radiant’s effort can make it more cost effective for drone operators to collect data for global development and conservation projects.
“We have a wide portfolio, whether it’s for agriculture, property rights, government transparency, global health, financial inclusion, or basic cartography for the developing world,” Miglarese explains. “We are also setting up a sandbox for people to be able to experiment and run models with the data. We are developing a very collaborative relationship with the commercial sector, with both software companies and imagery providers.”
Dearing has worked in and around the geospatial industry for almost 15 years. He began at a small startup using GIS to help support the defense industry, then spent almost ten years at Esri, working with a variety of industries. In early 2015, he joined Boundless as the VP of professional services. “We had a variety of customers and initiatives that were looking into deploying open GIS tools,” he recalls, “including the traditional OpenGeo stack that Boundless maintained, then transitioning it from an Esri environment to an open source platform.” In late 2015, he was promoted to CEO.
Boundless (known in its early days as OpenGeo) started in 2001 with a New York City nonprofit called OpenPlans. Its goal was to enable the free flow of communication and data between communities and the government, specifically as related to city planning. “The city wanted to provide an easy way of sharing geospatial information, without being tied to proprietary software platforms,” Dearing says. Out of these efforts emerged a server technology able to disseminate geospatial data and information in applications and maps. In 2002, the team built a project known as GeoServer. It was donated to OSGeo and a powerful open source geospatial community formed around the project. “We continue to support GeoServer today through a variety of community efforts with OSGeo.”
In 2008, to fulfill the need for a complete supported open source geospatial enterprise platform, Boundless created OpenGeo Suite, which bundles PostGIS, GeoServer, GeoWebCache, and OpenLayers. By 2010, there were more than 20,000 downloads of OpenGeo Suite.
OpenGeo Suite’s release was followed by a significant uptake in downloads from the federal government, which had realized that open source GIS could help enable its enterprise and reduce dependencies on single vendors of proprietary software. Additionally, the federal government is required to establish commercial support for their open source implementations. “OpenGeo (Boundless) is the only commercial support provider for the open source geospatial platform, but at that time, we were still a non-for-profit,” Dearing recalls. In 2012, In-Q-Tel (the federal government’s investment arm), Vanedge, and several angel investors invested in OpenGeo to transform it into a for-profit organization known today as Boundless.
MISCONCEPTIONS OF OPEN SOURCE
With open source, Dearing points out, the developer community and the customers, not a single company, shape the software platformís vision and roadmap. Among the misconceptions about open source geospatial software are that it is insecure, buggy, or unmonitored. “That is far from reality,” says Dearing. “There are many great gate checks out there and well-established steering committees. Core contributors control the code base to ensure high quality, that there are no malicious code or security holes, and that the technology performs at the level it should. Then you have organizations like OSGeo and LocationTech that help promote, sponsor, and maintain these communities around the open source projects.”
Up until five to ten years ago, open source geospatial web applications and desktop products were not widely adopted, says Dearing. However, he pointsout, ìas many organizations saw the performance, scalability, and the cost advantages of open source platforms, the gap in usability and functionality between them and proprietary solutions narrowed significantly. There are desktop tools, such as QGIS, as well as web mapping technologies, such as Leaflet and OpenLayers, that now are widely adopted to help GIS professionals, and developers create, share, and visualize geospatial information.î There is an explosion in the adoption of the open GIS platform in many federal government entities, he adds, and agriculture, insurance, and healthcare are also huge growth sectors.
Often, Dearing points out, Boundless’ customers are also Esri’s customers. The fact that Esri is now pushing code out on GitHub is “a nice first step,” he says, but the core of its business is to provide proprietary software to its customers. By contrast, Boundless offers a hybrid approach: “Many organizations are seeing success in implementing open source in a proprietary way. There is a progression to implementing and realizing the value in open source, as you minimize your dependency on closed source solutions.”
Turner has worked in the technology domain for 20 years, the Web domain for 15, and the geospatial domain for a little over ten years. “I come from the open source community,” he says. “I began learning LINUX in 1997.” When he started to work at Esri, he was surprised at how much open source was part of its culture. “Internally, all the engineers have access to all the different projects. So, if I see that there is a feature that I want to add to ArcGIS Pro or Enterprise or Online, I could make a pull request against that, much as you can do in the open source world. We might even bring open source projects into our technology, improve them, and release the results back out. We also take things that we built ourselves and open source those. There are currently 400 open source projects on Esri’s GitHub repository.”
GIS is not an app, says Turner, but a system of many interconnected components. “That’s fundamentally the way that ArcGIS has been designed. We’ve been doing open source, kind of quietly, for the last several decades, while also adopting other open source projects. We’ve been heavy supporters of GDAL and other libraries.” A lot of popular open source geospatial software, he adds, came out of Esri’s research.
“One of Esri’s crown jewels is the open source code we use to manage and manipulate spatial operations on core geometry. We use the Apache 2 reusable license. You can use it and modify it and may share those modifications but donít have to.”
Starting around 2010, Turner points out, GitHub made it very easy to distribute, discover, and access open source code. “We want to make a lot of our technology open source because we want to make it as accessible as possible for any of our 23,000 government customers to find on GitHUB, then download, deploy, and modify.”
When he joined Esri in 2012, Turner says, Esri had two or three open source projects that were part of its product line, including Geoportal, a catalog and distribution tool, and one for editing OpenStreetMap data with Esri tools. Then, Esri decided to re-engage with the open source community. “It was a tactical shift in terms of where we were doing it. Geoportal Server is a metadata catalog and search tool. Another one is an extension for ArcMap that allows people to download and analyze the map data or upload it to ArcMap.”
Esri is interested in the broader context of open source, says Turner. “That it is not just available, but also accessible, so that it is easy to discover, use, and integrate into existing workflows.” People bring their own ideas and perspectives into open source projects and are able to influence them. ìIn the end, this will make better technology and better solutions for everyone.
SUPPORTING OPEN SOURCE
Esri supports the larger open source community through collective funding as well as code changes to various open source libraries, says Turner. “We open sourced our geometry engine and are now working with the University of Minnesota to integrate that into different libraries and tools. Now we’re working with other communities to adopt their tools and improve them. Esri Leaflet, for example, is a popular Web mapping library. We’ve improved that core concept code, and we’ve built extensions to that, that we open source. We adopt existing open source tools, share them back out, and sponsor open source-focused conferences as well.” For example, he points out, Esri has been the longest continuous sponsor of the FOSS4G (Free and Open Source for Geospatial) conference.
Esri recently launched ArcGIS Hub, to try to put open source data and software “into the perspective of what communities are actually trying to accomplish to improve lives, safety, health, and happiness,” says Turner. “How is Los Angeles going to accomplish its own Vision Zero pedestrian death reduction program? How is New Orleans going to do that? How is Rio going to do that?” ArcGIS Hub enables them to share their plans and methodology as they develop them, as well as their data, he says.
Location, Turner points out, is as fundamental an aspect of all software as time, but implementing geospatial software requires particular expertise. “There are particular complicated mechanisms and algorithms that make geospatial possible,” he says. “They include how to code data, handle precision, handle accuracy, make it scalable, make it fast, make it capable to do the different kinds of analytics that we understand today but also that we might implement in the future.”
Geography, he points out, is interesting because it’s fractal. “You can never measure perfectly. Geospatial has tools, mechanisms, and expertise that we have acquired through centuries of measuring and coding geography. I think that’s the unique aspect of geospatial as we see it.”
he roots of open source GIS can be traced back to the Map Overlay and Statistical System (MOSS) developed by the U.S. Department of the Interior in 1978, followed shortly by the Geographic Resources Analysis Support System (GRASS). In a 2005 article introducing open source geospatial tools, Tyler Mitchell wrote, “The development of open source geospatial software is an exciting part of the new geospatial landscape.”
In 2013, Boundless became the first company to provide commercial support and maintenance for the world’s most popular open source GIS applications at the database, server, desktop, web, mobile, and cloud levels. Evidence of Esri’s commitment to the open source community is the launching of ArcGIS Hub. As the developer and user communities including Radiant.Earth continue to grow, open source geospatial software holds great potential for increased collaboration, the sharing of valuable data, and access to key resources.