Google DeepMind releases Forest Typology map distinguishing natural forests from plantations at 10-meter resolution
ForTy 2020 categorizes global forests by type and tracks conversions from 2017–2025, supporting EU deforestation compliance.
Last verified:
Forest classification closes a critical monitoring gap
Satellite data has long tracked where forests exist, but not what kind of forest occupies each pixel. According to the DeepMind Blog, Forest Typology (ForTy) 2020—released on October 22, 2026 by Google DeepMind, Google Research, the World Resources Institute, and the International Institute for Applied Systems Analysis—classifies every 10-by-10-meter patch of Earth’s forests into six distinct categories, from primary forest to tree-crop farm. Critically, the dataset includes a time series spanning 2017 to 2025, showing year-by-year changes and revealing not just where forest was lost, but what replaced it.
The distinction matters fundamentally for conservation and commerce. A primary rainforest and a single-species timber plantation may appear identical from orbit, yet the primary forest stores far more carbon and harbors greater biodiversity. The DeepMind Blog notes that most existing global forest maps indicate only whether trees are present—a blind spot for regulators and companies attempting to meet sustainability commitments.
EU deforestation regulation drives demand for forest-type data
The European Union’s deforestation regulation (EUDR) requires that imported commodities like coffee, cocoa, and palm oil originate from land that was not forest in 2020 or has not been degraded from primary to plantation forest thereafter. DeepMind Blog reports that wood products must also be certified as deforestation-free, prohibiting sourcing from primary forests converted into plantations or naturally regenerating forests converted into planted forests.
These compliance requirements created an urgent need for granular, global forest classification. A binary tree-cover map cannot answer whether a harvested area was formerly primary forest—ForTy 2020 can. The dataset provides a baseline to verify supply chains and support commodity producers in meeting conservation commitments.
Technical approach and deployment scope
The 10-meter resolution of ForTy 2020 represents a substantial leap in precision. Prior to this release, the collaborating organizations published Natural Forests of the World 2020, a 10-meter global map separating natural forests from tree cover broadly. ForTy 2020 extends that foundation by introducing forest-type granularity—primary, regenerating, planted, plantation, and farm categories—and adding temporal tracking. The ability to observe annual transitions year-over-year enables conservation organizations and regulators to detect conversions in near-real time and identify where forest protection mechanisms succeed or fail.
Why This Matters
ForTy 2020 shifts forest monitoring from a binary presence-absence paradigm to a categorical, temporally aware system. For companies in commodity supply chains, the map reduces compliance risk by enabling transparent sourcing verification. For conservation organizations, annual time series data reveals conversion patterns that inform prioritization of protection efforts. If the dataset proves accurate under independent validation and remains openly accessible, it will likely become the de facto standard for regulatory forest classification globally—similar to how Google-trained models have centralized satellite-derived land-use mapping in prior years. The ten-meter resolution balances monitoring precision against computational accessibility for developing-world regulators with limited satellite-processing infrastructure.
Frequently Asked Questions
What is the difference between ForTy 2020 and previous forest maps?
ForTy 2020 distinguishes forest type—primary, regenerating, planted, plantation, tree-crop farm—at 10-meter resolution with annual change tracking from 2017–2025. Prior maps showed only tree presence, not ecological value or conversion history.
Why does the EU Deforestation Regulation require forest-type mapping?
The regulation prohibits imports of commodities (coffee, cocoa, palm oil) and wood products from land converted from natural forest after 2020, or from primary forests degraded into plantations. Regulators and companies need to know forest type, not just tree cover, to verify compliance.
Who built ForTy 2020 and what institutions collaborated?
Google DeepMind and Google Research led development in partnership with the World Resources Institute (WRI) and the International Institute for Applied Systems Analysis (IIASA).