Automated detection of commodity-driven forest loss in tropical regions using Sentinel-1 C-SAR data

Details

Supervisors

Defourny, Pierre ; Delhez, Baptiste

Faculty

Faculté des bioingénieurs

Degree label

Master [120] en sciences agronomiques et industries du vivant, à finalité spécialisée: Geo-Information Science and Earth Observation for Environmental Modelling and Management

Abstract

enTropical forests are being lost at an alarming rate, primarily driven by the expansion of agriculture and commodity-related land use changes, posing a serious threat to biodiversity, ecosystem services, and overall global climate regulation. In response, the European Union Deforestation Regulation (EUDR) was enacted to guarantee products entering the EU market are not linked to deforestation, emphasizing the importance of developing effective, timely, and scalable tools for monitoring and compliance. This research explores Synthetic Aperture Radar (SAR) data from Sentinel-1 to develop an automated system for detecting forest loss in tropical regions between 2021 and 2024. The approach is applied across seven study areas, each representing a different deforestation-linked commodity: cattle, timber, soy, palm oil, cacao, rubber, and coffee. The methodology applies a threshold-difference technique based on quantile segmentation to identify deforestation patches on a quarterly basis. The workflow includes the selection and validation of forest masks, in addition to accuracy assessments through confusion matrices using optical reference data. The results show a strong detection performance, with an overall F1-score of 0.92 for identifying crop-induced forest loss. Temporally, the system detected changes with an average delay of only 14 days. Likewise, high deforestation events were observed to align with dry seasons, indicating the system’s sensitivity to seasonal patterns. These findings emphasize the potential of SAR-based monitoring to support the EUDR. The proposed method allows for scalable, regular, and near real-time detection of deforestation, contributing to more transparent commodity supply chains.