Researchers have released “ForestCast: Forecasting Deforestation Risk at Scale with Deep Learning”, along with the first publicly available benchmark dataset dedicated to training deep learning models to predict deforestation risk. The work shifts the focus from measuring forest loss after it happens to forecasting where forests are most at risk in the future.
The announcement comes as forest loss remains a major global problem. Last year alone, the world lost the equivalent of 18 soccer fields of tropical forest every minute, totaling 6.7 million hectares — a record high and double the amount lost the year before. Habitat conversion is also described as the greatest threat to biodiversity on land.
For years, satellite data has been used to measure forest loss. In collaboration with the World Resources Institute, the team previously mapped the drivers of that loss — including agriculture, logging, mining and fire — for the years 2000–2024. Those maps were produced at an unprecedented 1km2 resolution.
The new approach is designed to forecast deforestation risk using pure satellite data rather than patchily available input maps such as roads and population density, which can quickly go out of date. The researchers say the method can be applied consistently in any region and updated as more data becomes available.
They also said the approach could match or exceed the accuracy of previous methods. To support reproducibility, they are releasing all input, training and evaluation data as a public benchmark dataset.
Source: research.google.
Companies can share verified announcements through Newz9’s international press release submission page.
