Google DeepMind Releases ForTy 2020: A 10-Meter Global Forest Typology Map

Google DeepMind and Google Research, with the World Resources Institute (WRI) and the International Institute for Applied Systems Analysis (IIASA), released Forest Typology (ForTy) 2020, a global 10-meter resolution map that classifies every 10-by-10-meter patch of land on Earth into six forest and land-use types, alongside an early time series covering 2017 to 2025. The release matters because it goes beyond binary "tree cover" mapping to show what kind of forest exists on a given patch of land — primary forest, timber plantation, tree-crop farm or other land — and what replaced forest after it was lost, a distinction the EU's deforestation regulation requires.

What was announced

ForTy 2020 is a global map published in the journal Scientific Data that categorizes land at 10-meter resolution into six distinct forest and land-use types. It is described by its authors as a key baseline for the EU's deforestation regulation.

Alongside the 2020 map, Google DeepMind shared an early-release time series that tracks how these forests changed each year from 2017 to 2025, including where natural forests were replaced by plantations or farms. The announcement states that this is the first time users can see at 10-meter detail, worldwide and year by year, not just where forest was lost but what replaced it: a timber plantation, a tree-crop farm, or other land.

The work was co-developed by Google DeepMind and Google Research, together with the World Resources Institute (WRI) and the International Institute for Applied Systems Analysis (IIASA). The post is attributed to Maxim Neumann and Charlotte Stanton and was published on October 22, 2026.

Why "tree cover" is not enough

Most global forest maps only indicate whether trees are present, which is insufficient for protecting natural forests. From space, an old-growth rainforest, a eucalyptus plantation, and an oil palm estate can look like the same patch of green, yet they hold very different amounts of carbon and biodiversity.

A forest's ecological value depends on its history: a primary forest that has never been cleared holds far more carbon and biodiversity than a single-species timber plantation harvested every decade or two.

The European Union's regulation on deforestation-free products (EUDR) requires that products such as coffee, cocoa, or palm oil do not originate from land that was forest in 2020 and has since been converted to agriculture. Wood products such as furniture must also be free of forest degradation: they cannot come from primary forests converted into plantation or planted forests, or from naturally regenerating forests converted into plantation forests.

As a result, companies and regulators need to know not just where trees are, but what kind of forest they form — primary or regrowing, planted or plantation, forest or farm — and what replaces that forest after a harvest.

The six land classes

ForTy sorts land into six categories based on definitions from the UN Food and Agriculture Organization (FAO):

1. Primary forest: native forest with no visible signs of human disturbance.
2. Naturally regenerating forest: forest that grew back on its own, for example after logging or fire.
3. Planted forest: forest that people planted or seeded, and managed over long cycles.
4. Plantation forest: fast-growing, intensively managed tree farms, often of one species.
5. Tree crops and agroforestry: farms that grow things other than wood, such as oil palm, rubber, cocoa, or fruit.
6. Other land: everything else, from grassland and cropland to cities and water.

The published Forest Typology map for 2020 shows the most common forest or land type in each square kilometer, and can be explored at 10-meter detail in a Google Earth Engine app.

How the ForTy model works

Training the ForTy model was challenging because the six categories depend in part on how land is used, not only on how it looks. To teach the model the difference, the team built a large-scale collection of forest-type examples with about 1.7 million labeled locations drawn from expert surveys, national and regional datasets, and targeted reviews of places where the model struggled.

The model uses a custom architecture designed for the challenges of remote sensing data. Its key input is AlphaEarth Foundations embeddings — data-compact summaries of years of satellite observations for each 10-meter pixel — which allow it to spot subtle clues such as planting patterns, growth cycles, and harvest rhythms.

The five forest and tree-crop classes can look alike from above. ForTy identifies subtle differences in structure and seasonal patterns that set them apart, categorizing them in a global map database.

Validation and accuracy

Working with IIASA and WRI, the team validated ForTy against more than 9,000 randomly chosen sites around the world. At each site, two trained experts studied high-resolution imagery and years of satellite records to determine which type of land was actually there.

ForTy picked the right forest type 90% of the time, and told forest apart from non-forest land 95% of the time, outperforming other global forest maps.

Probability scores instead of yes-or-no labels

Rather than giving a yes-or-no label for a specific patch of land, ForTy provides a probability score for each class. This lets users adapt the classification to regional contexts and to application-specific precision trade-offs.

A company assessing deforestation risk under EUDR, for example, might set a strict threshold and accept only pixels the model is highly certain about, to avoid regulatory risk. A researcher estimating total forest extent might instead use the full spectrum of probabilities to get a fuller picture of the potential landscape.

The 2017–2025 time series

The new early-release time series applies a similar approach to every year from 2017 to 2025. It shows not only where forests disappear and where they grow back, but also what kind of change happened — for example, a natural forest turned into a timber plantation, or abandoned land slowly regenerating into forest.

This differentiation matters because some of the most important changes are hard to see from space. When a natural forest is logged and replanted as a plantation, the land soon looks green again, so standard tree-cover maps might not pick up the change. ForTy's time series makes these hidden changes visible, covering both losses and gains.

Availability and licensing

The Forest Typology 2020 map, described in Scientific Data, and the new time series are freely available on Google Earth Engine under an open license. The stated goal is for governments, companies, conservation groups, and scientists to use them to build deforestation-free supply chains, direct forest finance where it is needed, and protect the natural forests that matter most.

The team says it will keep improving the time series with its partners and users over the coming months.

How ForTy fits into Google's broader nature and Earth AI work

ForTy follows Natural Forests of the World 2020, a 10-meter resolution global map jointly released by the same organizations in the previous year that separated natural forests from all other tree cover. That earlier map answered one question for every patch of land: was it natural forest in 2020 or not. ForTy goes further, showing what kind of forest is present and, with the time series, how that changes each year.

ForTy is part of Google DeepMind's AI for Nature portfolio, which includes mapping the drivers of forest loss, forecasting deforestation risk, modelling global species distributions, and monitoring biodiversity in the field. It is also the latest addition to Google Earth AI, Google's collection of geospatial models and datasets for environmental challenges.

The post credits co-authors Anton Raichuk, Peter Potapov, Myroslava Lesiv, Matthew Overlan, Mélanie Rey, Ravindran Rajakumar, Michelangelo Conserva, Radost Stanimirova, Michelle Sims, Sarah Carter, Elizabeth Goldman, Yuchang Jiang, Linus Scheibenreif, Ivelina Georgieva, Maria Shchepashchenko, Steffen Fritz, Nicholas Clinton, Dan Morris, and Drew Purves, as well as trusted testers who helped improve the map.