Global Forest Watch-Fires: Improving Remote Sensing Through Community Engagement

Sunday, 15 February 2015
Exhibit Hall (San Jose Convention Center)
Lexine T. Hansen, US Agency for International Development, Washington, DC
Forest Fires in the southeast Asia region contribute to deforestation, produce a toxic haze that is detrimental to the environment and human health, and result in millions of dollars in damages. Because fires are in areas often remote from government officials, it is difficult to locate them in time to bring them under control and to hold fire-starters responsible for damages. Building upon the detailed remote sensing data sets and online platform of Global Forest Watch, Global Forest Watch-Fires (GFW-Fires) combines real-time satellite data from NASA’s Active Fires system, high resolution satellite imagery, detailed maps of land cover and concessions, weather conditions, and air quality data to track fire activity and related impacts in the southeast Asia region. GFW-Fires can alert people to fires quickly, to better combat harmful fires before they burn out of control and to hold accountable those who burn forests illegally. Moreover, iGFW-Fires allows global activist communities, including neighboring countries which suffer the pollution effects, to locate the source and hold land-holders responsible for downwind impacts. While GFW-Fires is free to use and follows an open data approach in putting decision-relevant information in the hands of all, the reliance on remote-sensing data alone distances data control from local communities and activists who are also concerned with damaging fires and deforestation.  Therefore, we developed crowdsourcing to validate and improve the data set underlying the GFW-Fires platform. We partnered with Tomnod, a web application that uses satellite images to explore the earth and solve real-world problems, encouraging volunteers to tag satellite images as fires, burn scars, or neither, allowing people to validate or dispute what satellite images and other data suggested from afar.  Statistical comparisons of data accuracy showed which remote-sensed data were most vulnerable to errors.  We used this crowdsourced data to improve GFW-Fires’ accuracy and to build engagement with local communities and the platform. As a result, not only is GFW-Fires more accurate, but we also find that increasing crowdsourcing access allows us to triangulate and improve accuracy across the Global Forest Watch family of platforms. Moreover, the Tomnod campaign increased GFW-Fires traffic and spurred interest in addressing a complex global challenge.