There’s no mistaking the value of trees. They provide jobs and resources for people worldwide, and their carbon dioxide mitigation potential plays an important role in climate action. Fully understanding their value — and where more are needed — requires credible and up-to-date information on where and how many trees are growing.
Existing satellite technology, such as Global Forest Watch (GFW), relies on remote sensing algorithms that detect the percentage of tree cover in each pixel, rather than counting the number of trees within. When tree cover is below a certain threshold, the methodology used by GFW has trouble seeing those trees. As a result, this technology is best-suited for measuring tree cover in dense forests where thousands of trees are packed closely together.
However, this method can miss trees across dry forests, farms and other areas with sparse tree cover, which makes it even more difficult to measure change in these landscapes. Considering these areas make up more than 40% of Earth’s land area, neglecting to count their tree cover is a missed opportunity. This has serious implications for rural communities, which often rely on these trees outside forests for their livelihoods.
Though satellites have difficulty detecting trees outside forests, the human eye can. Collect Earth — a data collection tool developed by the Open Foris initiative of the UN’s Food and Agriculture Organization (FAO) that pulls imagery from Google Earth — uses the power of our eyes to visually analyze the imagery and collect data on land cover (type of land, like “forest”), land use (how people use that land) and tree cover at different points in time. By matching a high-tech tool with the low-tech but accurate human eye, local stakeholders can harness the power of this tool to see where billions of trees are growing in landscapes around the world — and where land is under restoration.
Monitoring Land Restoration With Collect Earth Mapathons
Since 2015, teams across El Salvador, Ethiopia, India and Rwanda have held participatory mapping workshops, which we call Collect Earth mapathons, to produce local data for monitoring restoration. Armed with the high-quality images from Google Earth that allow people to see objects as small as half a meter (about 20 inches), the teams recruited local residents to identify the different physical features of the landscape in a collection of sample plots.
These mapathons allowed people with first-hand knowledge of local landscapes to participate without any prior knowledge of remote sensing. The only expertise they need is basic computer literacy, familiarity with the landscape and knowledge on how people use it.
For each session, the teams recruited participants and met in a facility with the capacity and internet connection to host the event. Once trained, participants could analyze 50 to 80 plots per day, depending on the length of the survey and the amount of data to be collected. Then, analysts added up the results from each participants’ plots, creating accurate data for an entire landscape (or country). After the initial counting, the teams compared the identified land cover with the number of trees counted. For example, if a local participant counted just three trees in a large plot, but later categorized the same plot as a forest, the researchers knew there was human error and could correct it.
Although this method only requires basic expertise and minimal training, the results have major implications. The landscapes where mapathons occurred are home to some of the most striking examples of community-led landscape restoration, which this data can inform.
In Sodo Gurage district, the Ethiopian government used the program to report on progress toward the local target of reaching 19% forest cover by 2020. The data showed that forest cover increased from 7.5% in 2010 to 8.1% in 2015. This indicates a positive trend that falls short of the target, helping local leaders understand that they needed to tweak their approach and accelerate their effort.
A mapathon conducted in India’s Sidhi District worked with local farmers, students and other community members to identify where tree cover exists on healthy land, where farmland was already under restoration, and where growing more trees could benefit people environmentally and socially. Using this data, the participants made the case for restoration to the government and other partners — a $19 million opportunity by one calculation. Now, they’re starting to restore more land, faster.
Local Restoration Data, Local Benefits
Restoring land and growing trees can boost crop yields, stabilize the soil and hold back the ever-creeping desert. In landscapes where local experts hold mapathons, people can harness this data to achieve those local goals, including those laid out in five-year economic development plans. When local people gather data on restoration themselves, it brings benefits back to the community. Because they have personal knowledge of the surrounding areas, they are in the best position to put raw data in the right context.
Tapping into local knowledge of the landscape doesn’t only improve the quality of the data and maps. When local people see how Collect Earth data can help them evaluate the health of their landscape, they are more likely to invest resources to update the data on a regular basis, even after trainers and NGOs leave the landscape. Local data can also help global researchers better understand which restoration types and approaches work in which ecosystems.
In the long-term, tracking restoration progress in detail creates a win-win-win situation for all. Governments and investors can show that they are fulfilling their restoration pledges as early as one year into a project and more accurately than when they use algorithms to detect trees alone. Quick progress reports also allow funders to see that restoring land provides true return-on-investment. It also gives local communities the recognition they deserve for their hard work.
Harnessing the Value of Collect Earth
Hosting a mapathon sets a strong foundation for monitoring where trees are growing outside of forests. Training local mapping experts who can continue collecting and analyzing Collect Earth data for years is key to tracking progress effectively over time. In turn, they can use local languages to train dozens of users each year, building up a local knowledge base.
Mapping tree cover is just the beginning. Experts can customize the program to match local contexts, especially in unique ecosystems such as Niger’s tiger bush. It could also help show where restoration techniques that don’t involve growing trees, like digging rainwater collection ditches and regreening grasslands, are transforming the landscape.
An online version of Collect Earth, under development, hopes to further democratize access to the program. There’s also a promising possibility for new machine learning techniques that use data collected from mapathons and develop comprehensive maps of these trees outside forests with artificial intelligence models. This approach could provide high-quality, accurate baseline data for restoration projects — and entire landscapes — and then help them track progress over time.
At a time when many people are advocating for technological solutions to climate change and other pressing environmental challenges, it’s important to remember that even the most cutting-edge techniques need to build on local insights. Collect Earth mapathons show what’s possible when technology and local knowledge meet, creating new and exciting ways to count billions of trees that would otherwise go unseen.