PARTNER POST: Introducing the new data GLADiator of forest monitoring in Brazil
By Camellia Williams We don’t normally like to Crowe, but we’re really proud to announce the launch of Brazil’s most advanced forest monitoring tool — GLAD Brazil.
Updated weekly and available at a high resolution of 30 meters by 30 meters, the satellite-based GLAD (Global Land Analysis and Discovery) alert system lets anyone with an internet connection monitor deforestation as it happens, in almost real-time.
Brazil already has two of the most sophisticated forest monitoring tools: PRODES and DETER. However, GLAD offers more frequent updates, at a higher resolution, for the entire country. Created by the University of Maryland, Google and World Resources Institute (WRI), GLAD alerts have an important role in stopping deforestation before it spreads too far. GLAD alerts are already available for Peru, Republic of Congo and Kalimantan, Indonesia, and as we prepared to add to Brazil to that list, we knew we had to ensure the analysis remained fast and accurate. The biggest challenge in achieving this was the sheer amount of information the system has to process and deliver. Every week, new Landsat satellite images reveal where forest cover might have been lost. The appearance of roads are often the first sign that deforestation is happening and these can be spotted by the high-resolution Landsat images. But with higher resolution comes greater amounts of information that has to be processed. Working with WRI to process the GLAD data, we start up a 40 core Amazon server in the cloud that downloads raw rasters from the University of Maryland (UMD). For each area of interest, the UMD produces a confidence and a date raster for 2015 and 2016 — sixteen rasters in total. We then use the GDAL open source raster processing library to combine them and encode the data using the RGB colour scheme we developed. Once this data is ready, we use Mapnik to tile each raster into png images and upload them to Amazon S3 to be pulled into GFW. If you are interested in looking at the github repo, you can find it here. In the front end of GFW, we associate a colour with each date and confidence value so it can be drawn on the map, displaying alerts as pink pixels on the screen. Unlike some solutions, which have to make multiple calls to the server, our approach allows the front end to simply hide the pixels that don’t match the date range you are interested in. The result is a faster, more satisfying experience.
Another new feature that’s been introduced is the option to receive deforestation alerts for Brazil in either English or Portuguese. This means there’s no need for anyone to translate anything before they share it with their team for investigation. We’ve also written a tutorial on how to set up alerts for anyone that’s using them for the first time. Like the emails, this will also be available in Portuguese. Knowing that it can be tricky to get people to sign up for new things, we’ve updated the login process so people can login once and get access to a whole range of databases and saved subscriptions. By making the experience as painless as possible, we aim to create a rewarding experience for people that want to use a range of Global Forest Watch features. GLAD alerts are available to anyone who wants to use them — forest managers, protected area managers, local communities, students, and researchers — and we’re looking forward to seeing how people use it. Explore them here and if you find something interesting, why not share your story with Global Forest Watch.
VIDEO: Patrol the Leuser Ecosystem with Forest Defenders at HAkA
It’s the last place on Earth where orangutans, Sumatran rhinos and elephants and tigers still roam wild together, and it’s under threat. The Leuser ecosystem in Indonesia is a hotspot for biodiversity and a vital carbon sink, but encroachment from logging and oil palm and rubber plantations is eating away at the forest. Forest, Nature and Environment of Aceh (HAkA), an organization dedicated to protecting the environment in Indonesia’s […]