Published 29 October 2026 | Publisher: NewsPro | AI-assisted reporting based on the sources below.
A satellite map that marks land simply as tree-covered can blur meaningful differences: naturally regenerating forest, timber plantation and tree-crop farm may all appear green from above. Google DeepMind, Google Research, the World Resources Institute and the International Institute for Applied Systems Analysis say their Forest Typology (ForTy) project distinguishes those categories at 10-metre resolution.
The collaborators announced the map on 22 October. Their project overview describes a global 2020 baseline and an annual series tracking changes from 2017 through 2025. It labels six classes: primary forest, naturally regenerating forest, planted forest, plantation forest, tree crops and agroforestry, and other land cover. This can help researchers examine what land use follows forest loss.
Pixels and confidence
The underlying paper in Scientific Data describes a dataset derived from satellite imagery and expert classification. Its 10-metre pixels offer local detail, but resolution is not the same as certainty: the dataset includes class-confidence values that users can apply when assessing ambiguous areas.
For validation, the authors used a global sample of 9,100 sites. They report that 8,190 had been resolved through expert consensus and were included in accuracy reporting; other locations were pending consensus or marked uncertain at the time of analysis. This documents a review process, but does not mean every pixel is correctly labelled.
A map for investigation
ForTy is intended to make forest types and transitions more visible to researchers, conservation groups and policymakers. It does not itself determine whether a product meets a legal standard, prove why a landscape changed, or replace field checks. Remote-sensing classifications can be difficult where imagery is sparse or land uses overlap.
The annual series can guide investigation, not make legal judgments.
Sources
- Google DeepMind, “Beyond tree cover: mapping the world’s forest types” — 22 October 2026.
- Scientific Data, “Global forest typology at 10-meter resolution for forest and land-use monitoring” — 2026 paper.
Illustrative photo: Geranimo / Unsplash.