Forest Preservation

AI maps, monitors, and prioritizes forest restoration from satellite and field data.

Business impact

• Average canopy-height prediction error — Meta reported an average canopy-height prediction error of 2.8 meters.
• Land under restoration — re.green reported more than 16,000 hectares under restoration.
• Seedlings to be planted — re.green reported at least 10.7 million seedlings to be planted.

Data requirements

• Satellite imagery (image) — Used to map canopy cover, height, and forest change.
• Drone imagery (image) — Adds higher-resolution observations for monitoring and validation.
• Field inventory data (tabular) — Ground data connects remote sensing outputs to on-site measurements.

AI methods and techniques

• Computer vision — Interprets satellite imagery to map canopy and forest conditions.
• Machine learning — Predicts canopy height and other forest metrics from remote-sensing data.
• Spatial prioritization — Ranks restoration sites by landscape and forest conditions.
• Remote sensing analytics — Combines satellite, drone, and field data for forest monitoring.

AI models and model families

DiNOv2

Industries

Real-world evidence

5 documented case studies on record.

Companies using this: Google, Ping An Insurance (Group) Company of China, Ltd., re.green, Meta and World Resources Institute (WRI), Anew Climate.