source: Google DeepMind: Beyond tree cover: mapping the world’s forest types
level: technical
google deepmind, google research, the world resources institute, and iiasa released forest typology 2020, a global map that categorizes every 10-by-10-meter patch of land into six forest and land-use types. the map distinguishes primary forest, naturally regenerating forest, planted forest, plantation forest, tree crops and agroforestry, and other land. it serves as a baseline for the eu's deforestation regulation. an early time series tracks changes each year from 2017 to 2025, showing what replaced lost forest.
the model was trained on about 1.7 million labeled locations from expert surveys and datasets. it uses alphaearth foundations embeddings to capture subtle clues like planting patterns and harvest rhythms. validation against more than 9,000 randomly chosen sites showed 90% accuracy for forest type and 95% for forest versus non-forest, outperforming other global maps. the map provides probability scores for each class, letting users adjust thresholds for regulatory or research needs.
the time series reveals hidden changes, such as natural forest converted to timber plantation, which standard tree-cover maps often miss because the land stays green. the data is freely available on google earth engine under an open license. this work is part of google's ai for nature portfolio, which includes mapping deforestation drivers and forecasting risk. the map aims to support deforestation-free supply chains and direct forest finance where needed.
why it matters: for ai and data science, this map provides high-resolution, validated training data and a model architecture for land-use classification, enabling better monitoring of deforestation and forest degradation for regulatory compliance and conservation.
source: Google DeepMind: Beyond tree cover: mapping the world’s forest types