UN Climate Summit 2014: LIVE BLOG

WRI will be liveblogging during the 9/23 summit, tracking major announcements and offering expert commentary throughout the day.

You are here

FORMA: A Near-Real Time Alert System for Tropical Forest Loss

If you were the administrator of a protected forest, you would ideally have a staff of rangers on patrol, backed up by law enforcement and high-resolution satellite images to keep you up-to-date on the forest’s condition. In this ideal world, illegal activity would quickly be halted.

Yet the reality is that many protected areas lack these resources. Updated maps of forest damage can take years to produce, making it difficult to know where to focus limited enforcement efforts.

But it doesn't have to be this way, thanks to advances in technology. Global Forest Watch (GFW), the near-real time monitoring application released by WRI and partners last week, is powered in part by FORMA, a monitoring system that issues monthly forest loss alerts for the humid tropics in Asia, Latin America, and Africa. FORMA (FORest Monitoring for Action) is designed to help people managing forests respond more quickly to unwanted forest loss. Learn more about the system through our new issue brief, which compares FORMA’s performance to systems specific to the Brazilian Amazon, PRODES and DETER.

Rather than hoping forest managers can keep an eye on vast swaths of forest using scarce rangers or expensive high-resolution imagery, they can use GFW and FORMA to quickly highlight areas of interest that can be targeted for more intensive investigation. You don’t need a PhD in remote sensing to use the data—you can browse the maps for free at GlobalForestWatch.org.

By analyzing rapidly updated imagery from NASA, FORMA generates alerts of likely forest-clearing activity every 16 days at 500-meter resolution. The FORMA alerts data go back as far as January 2006, allowing us to see where forest loss has happened over the past few years—and help us understand why it’s happening. For example, check out the animated gif below, which highlights tree cover loss alerts in Indonesia’s Tesso Nilo National Park from 2000-present.

The History of FORMA

The FORMA team, which started at the Center for Global Development and moved to the Data Lab at the World Resources Institute, was inspired by the DETER and SAD forest monitoring systems covering the Brazilian Amazon. These systems quickly generate maps of forest loss hotspots. They have made it possible for law enforcement, civil society organizations, and the media to react to illegal activity quickly and reduce Brazil’s rate of deforestation. The Terra-i system covering Latin America aims to do the same beyond the official borders of the Brazilian Amazon.

But really curbing deforestation in forests around the globe means making data like this available even more widely. So, building on the work of numerous forest scientists and remote-sensing experts—and leveraging the power of cloud computing—we developed the FORMA system to cover all of the world’s tropical forest regions.

How Does FORMA Work?

The FORMA system is made up of several data components: vegetation intensity and fires data from the MODIS sensor aboard NASA’s Terra satellite, precipitation data from the National Oceanic and Atmospheric Administration (NOAA), and historical data on forest clearing. A statistical model uses the individual history of each pixel to identify meaningful signs of forest cover loss (as opposed to drought or other seasonal variation—see figure below). The end result is a map that highlights areas of concern based on the most recent satellite imagery. We update the GFW map with new data every month, providing a near-real-time look at deforestation and forest cover change in the tropics.

FORMA’s analysis of vegetation intensity looks at change in NDVI (Normalized Difference Vegetation Index). This graph shows a single pixel’s NDVI history and analysis of its rate of change, a key component of the FORMA algorithm.

We strive to keep false positive alerts (those that mistakenly highlight forest loss) to a minimum, so some potential clearing activity does not appear on the map. The dataset driving the map is downloadable in CSV format, and we are working to make the complete dataset available in an easily digestible form for bulk download and API access.

For detail on the methodology of the FORMA system, please visit the source code repository on GitHub.com, or keep an eye out for our forthcoming methodology paper.

Using FORMA Data to Track Forest Loss Hotspots

As an alerting system, FORMA is intended to identify hotspots of activity that can then be verified using other data, including information from people on the ground. For example, the Jane Goodall Institute (JGI) is exploring potential uses of FORMA alerts in its work in Africa. Partners like JGI might use FORMA to help local communities and protected area managers detect likely tree cover loss in chimpanzee habitat, and then dispatch ranger patrol units using mobile devices to investigate or alert local authorities. The system can play a vital role in curbing illegal deforestation and preserving important biodiversity.

Opening Up Forest Data

As big believers in open data, we're releasing FORMA data with the Creative Commons Zero license—no restrictions on use—as a step toward bringing transparent forest monitoring to the rest of the world. While highly detailed maps of forest loss – at 30-meter Landsat-scale or better – are the final word on the state of the world's forests, rapidly updated systems like FORMA, Terra-i, and SAD have a crucial role to play in improving forest management practices, aiding law enforcement efforts, and understanding the economic drivers of land cover change. Through Global Forest Watch, FORMA can help decision-makers and others keep an eye on the world’s forests.

SEND US YOUR FEEDBACK: As you explore FORMA data yourself, please send us feedback! This is a living dataset, so you can expect further refinements, including higher spatial resolution.

Share

Comments

Great system Robin. Question about the decision to "keep false positive alerts (those that mistakenly highlight forest loss) to a minimum". If you are providing an early warning system, might you want to err in the other direction?

Thanks for the question, Dave. The raw output of the FORMA system is a confidence, 0-100%. This confidence corresponds to the intensity of the spectral signals, as they relate to historical forest clearing activity. There is a tradeoff between confidence/intensity and timeliness. The single (purplish pink) visualization on GFW is at a relatively high confidence, which yields an early warning system -- albeit not as early as it could be. We chose a specific, global confidence for the visualization, but we release the full array of intensities (and code, found here: github.com/wri/forma-clj). Power users can explore the intensity data set and the generating code, and make planning decisions based on the lower confidence alerts. Our plan is to surface a new visualization of the data set soon, one that includes the various signal intensities. This is not an easy task. Stay tuned.

Add new comment

Stay Connected

Sign up for our newsletters

Get the latest commentary, upcoming events, publications, maps and data. Sign up for the biweekly WRI Digest.