From "are there trees?" to "what kind of forest?"
From a satellite, an old-growth rainforest, a eucalyptus plantation and an oil palm estate can look like the same green, yet they hold very different amounts of carbon and species. Most global maps only answer whether trees are present. Forest Typology (ForTy) 2020, released on October 22 by Google DeepMind and Google Research together with the World Resources Institute (WRI) and the International Institute for Applied Systems Analysis (IIASA), goes a step further. It sorts every 10-by-10-meter patch of land on Earth into 6 classes based on definitions from the UN Food and Agriculture Organization (FAO):
- Primary forest: native forest with no visible signs of human disturbance;
- Naturally regenerating forest: forest that grew back on its own, for example after logging or fire;
- Planted forest: forest that people planted or seeded and manage over long cycles;
- Plantation forest: fast-growing, intensively managed tree farms, often of one species;
- Tree crops and agroforestry: farms growing things other than wood, such as oil palm, rubber, cocoa or fruit;
- Other land: grassland, cropland, cities, water and everything else.
Alongside it comes an early time series of yearly changes from 2017 to 2025. When a natural forest is logged and replanted as a plantation, the land soon looks green again and standard tree-cover maps often miss the change; ForTy's time series is meant to answer what replaced the forest.
How the model sorts land, and how accurate it is
The difficulty is that these classes depend partly on how land is used, not just how it looks. The team assembled about 1.7 million labeled locations from expert surveys, national and regional datasets, and targeted reviews of places where the model struggled. A key input is AlphaEarth Foundations embeddings, compact summaries of years of satellite observations for each 10-meter pixel, which help the model pick up details such as planting patterns, growth cycles and harvest rhythms.
For validation, the team randomly chose more than 9,000 sites worldwide, and at each one two trained experts used high-resolution imagery and years of satellite records to decide what was really there. ForTy picked the right forest type 90% of the time and told forest from non-forest 95% of the time, which the team says beats other global forest maps. Rather than a single label, ForTy gives a probability for each class: a company doing compliance checks can accept only high-confidence pixels, while a researcher estimating total forest extent can use the full probability distribution.
Who will use it
The most direct demand comes from the EU Deforestation Regulation (EUDR). Products such as coffee, cocoa and palm oil must not come from land that was forest in 2020 and has since been turned over to agriculture, and wood products also can't come from primary forest converted into plantations. Companies and regulators therefore need to know not just where trees are, but what kind of forest it is and what it later became. The ForTy 2020 map (described in Scientific Data) and the time series are free on Google Earth Engine under an open license. The time series is still an early release, and the team says it will keep improving it with partners over the coming months.
via: Google DeepMind: Beyond tree cover: mapping the world's forest types