OSINFOR's Machine-Learning Satellite System Helps Peru Seize $19 Million in Illegal Timber

Peru manages roughly 68 million hectares of tropical rainforest where illegal logging is typically selective -- a handful of high-value trees removed per hectare -- making it very hard to detect with standard satellite change-detection tuned to larger-scale deforestation.

Detect individual illegally logged trees in near-real time from very-high-resolution satellite imagery, cross-checked against Peru's official forestry registers, to target field inspections.

Researchers at the University of Sheffield's Grantham Centre for Sustainable Futures built a machine-learning method analysing very-high-resolution satellite imagery for individual tree removal. Detected logging events are cross-referenced against SERFOR (Peru's forestry service) and SISFOR (its national forest-tracking system) to determine if activity falls inside an approved logging plan. OSINFOR, Peru's independent forest and wildlife oversight agency, adopted the tool to target field inspections, formally inaugurating it in 2025.

Across an initial 1.8 million hectares of monitored forest, the system identified 37% of all officially reported illegal logging. Peruvian authorities used the resulting evidence to seize more than 41,000 cubic metres of illegal timber between 2023 and 2024, worth over US$19 million, according to OSINFOR chief Williams Arellano.

The tool represents a step-change for enforcement timing, detecting selective logging while it is still happening rather than after full clearance is visible, and is designed to be extensible to other Amazon basin countries, though deployment to date has been limited to Peru.