Detecting deforestation with multitemporal L‐band SAR imagery: a case study in western Brazilian Amazônia

Abstract

Applications of L‐band SAR data to map deforestation are generally based on the assumption that undisturbed forests consistently exhibit higher radar backscatter than deforested areas. In this Letter we show that depending on the stage of the deforestation process (slashing, burning and terrain clearing), this assumption is not always valid, and deforested areas may display a stronger radar return backscatter than primary forest. The analysis of multitemporal SAR images, supported by several Landsat Thematic Mapper (TM) images and field knowledge, showed that wood materials left following the deforestation practices function as corner reflectors, causing an initial increase in the radar backscatter, which then subsequently decreases over time as the debris on these fields are removed.

Acknowledgments

The National Institute for Space Research (INPE) provided Landsat TM scenes and the JAXA Earth Observation Research and Application Center (EORC) provided SAR data, within the framework of the JERS‐1 SAR Global Rain Forest Mapping (GRFM) Project. The authors also thank three anonymous reviewers for their constructive criticisms of the manuscript, and colleague Ramon Freitas for his help in post‐processing the SAR data.