Forest Foresight — AI Radar-Based Deforestation Prediction Protecting Gabon's Rainforest

Forest Foresight combines Sentinel-1 radar imagery (via the RADD system) with a machine-learning model trained on patterns that typically precede illegal logging, mining and agricultural clearing, flagging high-risk forest areas up to six months before clearing happens. In Gabon it is jointly run by WWF, Wageningen University & Research and Gabon's Ministry of Water and Forest Affairs, with Cameroon's MINFOF as a partner in the wider Congo Basin pilot; the initiative was presented at COP28 in 2023.

The tool aims to give rangers and authorities an early warning window to intervene before illegal clearing occurs, rather than only detecting it after the fact, and ultimately to extend predictive monitoring to 15 landscapes across 12 countries by 2027.

Radar and ML-based alerts are generated for at-risk forest areas and passed to park rangers and authorities; the approach is piloted concurrently in Gabon, Kalimantan (Indonesia) and Suriname.

In one documented case, an alert let Gabonese rangers detect and stop an illegal gold-mining operation, saving an estimated 30 hectares of forest and leading to arrests. Across the Gabon and Kalimantan pilots combined, partners report alerts led to 34 deforestation-prevention actions within a six-month period.

The evidence comes directly from the implementing partners rather than an independent evaluation, and no published accuracy or false-positive rate exists for the prediction model; the division of operational responsibility between the Gabonese government and the NGO/university partners is also not clearly documented, so this is best read as a promising, real, but still NGO-anchored pilot.