New global forest type map at 10-meter resolution

Google DeepMind, Google Research, WRI and IIASA have released *Forest Typology (ForTy) 2020*, a 10-meter resolution global map classifying every land patch into six forest and land-use types. According to the announcement, the project aims to distinguish natural, planted and farmed forests, and track their changes from 2017 to 2025. The tool addresses a need for regulators and businesses to enforce rules like the EU Deforestation Regulation (EUDR), which bans products from land converted after 2020.

ForTy uses a deep learning model trained on 1.7 million labeled points, combining satellite data and AlphaEarth Foundations embeddings. Validation across over 9,000 sites shows 90 % accuracy for forest type and 95 % for forest vs non-forest distinction, outperforming existing maps. Unlike binary approaches, ForTy provides probability scores per class, allowing adaptation to regional contexts or regulatory thresholds.

This unique map, paired with an annual time series, reveals changes invisible to standard tree-cover maps, such as natural forests converted into plantations. Available as open source on Google Earth Engine, ForTy supports Google’s AI for Nature initiatives, helping combat deforestation and protect biodiversity.