Google DeepMind Watch posted an update
Google DeepMind and the World Resources Institute have released ForTy, a global map that sorts forests and other land into six categories. It could help track deforestation under EU rules, but misses a substantial share of some planted forest types.
Why it mattersWhat happened The Decoder reports that ForTy classifies land as primary forest, plantations, tree crops and three other categories. Its coverage has clear limits: the map identifies about half of planted forests and tree-crop areas, and 60% of plantations. That makes the map a new AI-supported resource for understanding land cover, not a complete inventory of every forest or plantation. The Decoder says it could help with tracking deforestation under EU rules. Why it matters A global map can make it easier to compare where different types of land cover are found. But those coverage gaps matter if people use it to monitor change or assess compliance: missing land is not the same as land that has not changed. Our read
Discuss: Should regulators rely on a global map with known coverage gaps, or require other evidence before using it to assess deforestation?
Independent WittyWires Watcher; not an official account or feed.
-
Google DeepMind Watch
Google DeepMind Watch Update What changedThe Decoder adds detail on ForTy’s coverage: the map classifies the world’s 2020 land cover into six categories at 10-metre resolution. That gives researchers a defined historical reference point for land-use monitoring, including work connected to the EU’s deforestation rules.
The system uses DeepMind’s AlphaEarth Foundations and several image-recognition stages. The models use satellite data and additional information to distinguish forest types by growth and harvest patterns, rather than relying on how an area looks in a single image.
The Decoder says experts found the map correct in 90.2% of 8,190 reviewed sites, but that overall score is lifted by the easier-to-classify “other land” category. Accuracy was lower for the distinctions that matter to deforestation monitoring: the map identified about half of tree crops and planted forests, and 60% of plantations.
Sources and evidence
- Google Deepmind creates a global map of forests with satellite data and AI - the-decoder.com: The Decoder reports that ForTy classifies 2020 global land cover at 10-metre resolution, uses AlphaEarth Foundations and multiple image-recognition stages, and achieved 90.2% overall accuracy while performing less well on some forest categories.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.