Last month, Taner Kodanaz, Director of DigitalGlobe’s Seeing a Better World Program, and Luke Barrington, Ph.D., Director of Product Management, joined us to share a comprehensive overview of DigitalGlobe’s advanced geospatial capabilities and of “Tomnod,” the organization’s crowdsourcing platform.
DigitalGlobe is a leading global provider in commercial high-resolution earth imagery products and services. They operate their own satellite constellation and their imagery offers solutions and support for a wide variety of industries.
The Seeing a Better World program is specifically focused on supporting humanitarian and environmental initiatives around the world by leveraging a unique set of transformational products and services to enable faster and better decision-making.
“The essence of Seeing a Better World is really focused on what Wade Davis calls the ‘fabric of existence’ on the surface of the Earth,” said Kodanaz. “Whether it’s how humans interact with the environment or what’s going on in terms of geopolitical discord, a lot of these issues and global challenges that we face need better information and better data for key players to make more informed decisions on the ground.”
Most development organizations know that better data, more timely data, helps saves lives and resources. The Seeing a Better World program is an effort to help improve situational awareness, provide tools to help with after-action impact, and enhance communication with stakeholders. The program framework focuses on four major themes: food and nutrition security, infrastructure development, sustainability and human rights. Within these areas DigitalGlobe is focusing on use cases around program efficiency, monitoring and evaluation, and human resource safety.
Kodanaz walked through examples of projects done in each of these areas in collaboration with clients and partners including the Bill and Melinda Gates Foundation, the World Bank, World Resources Institute and Walk Free. He also touched on relief work done in the wake of Typhoon Haiyan and efforts to support anti-poaching in Garamba National Park.
Barrington culminated the walkthrough of examples by telling the story of work done with The Nature Conservancy (TNC) to bring to life the opportunity in engaging large public audiences to help analyze satellite data. In 2014, TNC contacted DigitalGlobe for help in trying to understand the scale of a non-native Australian tree fern invasion on Kauai. TNC had already collected 44km2 of high-resolution aerial imagery, but needed support in sifting through all of this imagery to draw conclusions. The DigitalGlobe team uploaded the images onto the Tomnod crowdsourcing platform, effectively inviting thousands of volunteer contributors to help explore the problem.
Over the course of a few weeks there were 3.3 million map views by more than 11,000 participants, or “tree taggers,” and 2 million tags placed identifying the ferns. The geospatial information extracted from the data sets confirmed tree ferns as the biggest invasive problem – aggressively overtaking the island’s jungles and forests and requiring further TNC intervention.
To allay any issues around trust of the anonymous volunteer imagery analysts, DigitalGlobe has developed a statistical algorithm called “CrowdRank” to pinpoint overlaps in consensus. In the case of TNC and the Kauai tree fern invasion, the algorithm went through all 2 million points dropped on the map to reveal meaningful and reliable sources and information. The team was able to distill the 2 million points down to 22,000 locations and extract real information based on the statistical information afforded by CrowdRank. With this knowledge, TNC was ultimately able to better understand the nature and scale of the invasion and how their team could take action to halt the spread. [Read more about the TNC case story here >>]
DigitalGlobe continues to engage the Tomnod community around geospatial challenges like TNC’s. They’ve facilitated hundreds of campaigns and have a million subscribers who they reach out to as new challenges arise. This growing network of enthusiastic volunteers is invaluable in covering a lot of ground in little time.
“Eyes and brains are the best pixel processors,” said Barrington. “These are still the most advanced image analysis machines ever invented. It’s very hard for machines to be able to extract this information automatically.”
Kodanaz noted that clients nowadays are seeking actionable intelligence and are more interested in data layers than they are in the imagery itself.
“The pixels are interesting because they tell us information about the planet, but it’s the way we extract that information that’s really valuable to our customers and our partners,” said Kodanaz. “I’m finding that most global development organizations want the information and insight derived from the imagery. Imagery is still important because it helps you discern and contextualize what it is you’re addressing, but today it’s much more about information and insight.”