On Oct 22, 2026 Google DeepMind published Forest Typology 2020, a free global map that sorts land into six forest and land-use types. It was made with Google Research, WRI and IIASA; the paper (Neumann et al.) appeared in Scientific Data on Oct 21, 2026. From space an old-growth rainforest, a eucalyptus plantation and an oil palm estate can look like the same patch of green, and most global maps only say whether trees are present.
The number
90%. At 8,190 sites worldwide (the paper's count), two trained experts recorded what grew at each spot. The map agreed with them 9 times out of 10 (the paper gives 90.2% across six classes), so about 1 label in 10 was wrong.
What was found
- All land from 65 degrees south to 84 degrees north, in 10 by 10 meter squares (about 33 by 33 feet), as of 2020.
- Six labels per square: primary forest, naturally regenerating forest, planted forest, plantation forest (intensively managed tree farms), tree crops and agroforestry (oil palm, rubber, cocoa, fruit), other land.
- The model was trained on about 1.7 million labeled locations and reads years of satellite observations for each 10-meter pixel.
- An early-release yearly series, 2017 to 2025, shows where natural forests were replaced by plantations or farms.
- Free on Google Earth Engine under a CC BY 4.0 license.
What it does not mean
- The 90% is the makers' own check (vendor data), not an independent audit.
- The yearly series is an early release, not a finished product.
- The map describes 2020. By itself it does not prove any product deforestation-free.
Why it matters
An EU rule (EUDR) requires that products like coffee, cocoa and palm oil do not come from land that was forest in 2020 and has since been converted to agriculture. Google hopes governments, companies, conservation groups and scientists will use the map to help build deforestation-free supply chains. That is a hope, not a result yet.
Sources: Google DeepMind blog, Oct 22, 2026, deepmind.google/blog/beyond-tree-cover-mapping-the-worlds-forest-types; Neumann et al., Scientific Data, Oct 21, 2026, doi.org/10.1038/s41597-026-08299-5; Google Earth Engine data catalog, Forest Typology (ForTy) v1, developers.google.com/earth-engine/datasets/catalog/projects_nature-trace_assets_forest_typology_forest_typology_2020_v1_0_collection; checked Nov 9, 2026.
We are not affiliated with the institutions mentioned.