Higher-Res Data Set Now Shows Trees Inside and Outside Forests Across the Tropics

Mapping tree cover is vital for monitoring deforestation, restoration and trends in global land cover change. Remote sensing makes it possible to observe forests on a global scale by analyzing imagery from satellites to detect where trees are — and aren’t — over time.

It’s not just dense forests that are important; open canopy forests — which are common in drylands and in non-forested regions like in cities and on farms — are also important to map and monitor. However, doing so has proven challenging: Patches of trees may cover a smaller area than individual satellite pixels. This means trees in open canopy forests are often excluded by lower-resolution data sets, resulting in an inaccurate inventory of tree cover.

Global Forest Watch (GFW) is home to three remote sensing tree cover data sets widely used for quantifying forest trends and creating timely snapshots of the state of global tree extent. Two of these data sets — the University of Maryland GLAD Lab’s Tree Canopy Cover and Tree Cover Height — have 30-meter resolution and are derived from Landsat satellite imagery. These data sets are widely used for many applications, and they have the benefit of Landsat’s long historical archive to provide a consistent understanding of global tree cover trends over recent decades. However, one limitation of these data sets is that their resolution of 30 meters is usually not granular (high resolution) enough to map open canopy forests or small disturbances in closed canopy forests.

But a newly updated data set available on GFW, Tropical Tree Cover (TTC), uses higher-resolution data to map trees inside and outside of dense forests. Using 10-meter global imagery from Sentinel-2 satellites launched in 2015, the TTC data set maps tree cover across the tropics with greater granularity, improving our ability to quantify tree cover on non-forest lands, such as urban areas and cropland, and monitor trees at small spatial scales. TTC was originally launched on GFW in 2021 and is now updated to expand from non-forest land use to total coverage of the tropics in high resolution. The updated TTC data set is now fully available to download and analyze on GFW.

What’s the Tropical Tree Cover Data Set?

WRI’s Global Restoration Initiative (GRI) team developed the TTC data set (formerly called Trees in Mosaic Landscapes) to help land restoration and conservation stakeholders assess tree cover distributions at a local and national scale, and they published data for 1.3 billion hectares of non-forested land on GFW in 2021.

TTC uses artificial intelligence and monthly composites of Sentinel-1 and Sentinel-2 imagery to map tree cover distributions at a highly granular scale across the tropics. A key methodological improvement of TTC over other tree cover data sets is the use of video-based artificial intelligence to model how regions of pixels change over time to better differentiate trees from their background. All code for the project is open source and the code is freely available online.

What’s New With Tropical Tree Cover in 2023 and How Can I Use It?

In 2023, the GRI team updated the TTC data set to expand from solely focusing on non-forest land uses — like cropland, urban areas and areas with mosaic land cover — to include both forested and non-forested areas for 4.3 billion hectares of the tropics. Compared to other tree cover data sets on GFW, TTC’s relative strength is quantifying and understanding trees in non-forest areas, which requires detecting and monitoring trees at a small spatial scale. Learn more about the similarities and differences between TTC and other tree cover data available on GFW.

This data update coincides with new research and the ability to download and conduct analyses on the data from GFW. Users are also able to interact with the data directly on GFW to create visuals or select unique tree cover percent thresholds to best match their local forest definitions. For example, government agencies can use TTC to support the validation of their own national data sets on land cover or forest structure, or to inform forest and agriculture policies. Private sector organizations could use TTC for purposes such as assessing baseline characteristics of an area before implementing restoration practices.

Next Steps For Tree Cover Monitoring

The current data set shows tropical tree cover in the year 2020. The GRI team at WRI is now building on the method to monitor annual changes in tropical tree cover (loss and gain) at a 10-meter scale for 2017 onwards. These change maps will be useful for monitoring both degradation and restoration trends in the tropics, such as measuring progress toward the adoption of REDD+ sustainable forest management practices. GRI is testing applications of this data set to model aboveground biomass and carbon change on tree planting sites to more accurately monitor carbon removals from restoration and regrowth, and methods to differentiate between planted and natural tree cover. Results of these research initiatives will help to better understand tree cover dynamics in the tropics, monitor the progress of smallholder restoration projects and estimate carbon sequestered by these efforts.

To stay up to date on the latest TTC news, register for the TTC mailing list.