Satellite-powered deforestation risk assessment and supply chain traceability for EU Deforestation Regulation compliance across a cocoa importer's supply network.
A major European cocoa importer with supply chain operations spanning multiple continents, requiring compliance with the EU Deforestation Regulation (EUDR) for cocoa sourcing.
47 direct supplier farms across 3 continents, sourcing cocoa beans and cocoa derivatives with complex traceability requirements.
Full compliance with EU Regulation 2023/1115 (EUDR) requiring deforestation-free due diligence for all supply chain actors.
Hansen Global Forest Watch, Copernicus Sentinel-2, and ESA WorldCover integrated for multi-source deforestation monitoring.
Machine learning models trained on multi-spectral satellite imagery to detect and classify deforestation risk at farm-level resolution.
The EU Deforestation Regulation (EUDR), entered into force on 9 June 2023, requires operators and traders to exercise due diligence to ensure that products placed on or exported from the EU market have not been produced on deforested land after 31 December 2020. Regulation (EU) 2023/1115
The challenge is compounded by the need for high-resolution forest cover data. Bourgoin et al. (2024), in the EU JRC's global forest cover mapping for 2020, demonstrated that deforestation detection at sub-hectare resolution requires integration of multiple satellite sources and careful calibration against ground reference data. Bourgoin et al. (2024) — EU JRC
The foundational work of Hansen et al. (2013) established the first high-resolution global maps of 21st-century forest cover change, providing the baseline data infrastructure that underpins modern EUDR compliance efforts. Their annual forest loss dataset at 30m resolution remains the primary reference for temporal deforestation analysis. Hansen et al. (2013) — Science
Our client faced a critical gap: while EUDR mandates deforestation-free supply chains, existing tools provided only regional-level risk indicators. They needed farm-level traceability, integrating satellite imagery with GPS boundary data, supplier questionnaires, and field verification audits.
Downloaded Sentinel-2 composites for 47 farm polygons. Integrated Hansen GFW, ESA WorldCover, and historical MODIS vegetation indices. Established cloud-based data pipeline.
Trained Random Forest classifier on labeled reference data. Applied BFAST change detection. Generated preliminary risk scores and identified 8 farms requiring field verification.
Deployed field teams to 12 locations across 3 countries. Ground-truthed 450 reference plots. Validated GPS boundaries and documentation. Reclassified 3 farms as high-risk.
Compiled due diligence statements for all 47 farms. Delivered GIS dashboard with risk monitoring. Trained client team on continuous monitoring protocols.
Time-series analysis of forest cover change across supply chain regions using Hansen Global Forest Watch data (2000–2023).
Risk scores derived from Hansen GFW annual forest loss data, normalized per supplier region. Lower scores indicate reduced forest cover integrity.
Three.js globe showing supplier farm locations with color-coded risk levels. Green = low risk, Yellow = medium risk, Red = high risk.