Abstract :
[en] Tropical forest degradation results in severe biomass loss and biodiversity decline. However, fine-scale natural and logging-related forest disturbances remain difficult to trace, both from the ground as well as remotely. Comprehensive, landscape scale characterization of anthropogenic forest degradation requires accurate accounting of baseline canopy disturbance rates and patterns. This paper has evaluated the feasibility of radar data for detecting canopy gaps created by natural and anthropogenic mechanisms at large spatial scale by assessing the extent to which the Sentinel-1 C-band radar signal can be used to map fine-scale disturbances in both naturally disturbed and logged landscapes. Our physical-based method detects disturbances based on changes in backscatter resulting from radar shadow and/or layover. We apply various detection thresholds to explore the trade-off between detection and false detection and validate our method in study areas for which spatially exhaustive drone-based canopy gap maps are available for validation, namely Barro Colorado Island nature reserve (median gap area: 39 m2) and five logging concessions in the Congo Basin (median gap area: 237 m2). With a moderate threshold (2.5 dB backscatter reduction), we reach detection rates above 65 percent for disturbances above 200 m2 in both naturally disturbed and logged areas. Detection rates were primarily driven by gap area; gap depth had a smaller, yet significant, influence. These results significantly improve on operational forest disturbance products and previous studies on fine-scale disturbance detection using Sentinel-1 radar. Moreover, the improved insight in detection accuracies of anthropogenic disturbances fosters a move towards monitoring forest dynamics across large scales at which we cannot be certain whether the disturbance driver is anthropogenic or natural.
Funding text :
The authors would like to thank Ethan Belair and Peter Ellis from The Nature Conservancy and Bart Slagter from Wageningen University for inspiring discussions and thoughtful questions. We would like to thank the two anonymous peer reviewers for their time, effort, and excellent suggestions. A.W. received funding from the European Space Agency (ESA) Earth Observation for Sustainable Development (EO4SD) initiative (ESA ESRIN/Contract No. 4000131862/20/I-DT). A.W. and J.R. received funding from Norway's Climate and Forest Initiative (NICFI); the US Government's SilvaCarbon program; L.C.R and J.R. received funding from the Open Domain Science project Forest Carbon Crime (Project Number: OCENW.M.21.203) of the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO). S.W. and J.R. received funding from the Open-Earth-Monitor Cyberinfrastructure project (grant agreement No. 101059548) and The ForestWard Observatory to Secure Resilience of European Forests (FORWARDS) project (grant agreement No. 101084481). M.P.C. was supported by Apasia Grant 015.014.006. M.H. was supported by the CGIAR MITIGATE+ project. K.F. was supported by the Bezos Earth Fund.A.W. received funding from the European Space Agency (ESA) Earth Observation for Sustainable Development (EO4SD) initiative (ESA ESRIN/Contract No. 4000131862/20/I-DT ). A.W. and J.R. received funding from Norway\u2019s Climate and Forest Initiative (NICFI); the US Government\u2019s SilvaCarbon program; L.C.R and J.R. received funding from the Open Domain Science project Forest Carbon Crime (Project Number: OCENW.M.21.203 ) of the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO). S.W. and J.R. received funding from the Open-Earth-Monitor Cyberinfrastructure project (grant agreement No. 101059548 ) and The ForestWard Observatory to Secure Resilience of European Forests (FORWARDS) project (grant agreement No. 101084481 ). M.P.C. was supported by Apasia Grant 015.014.006 . M.H. was supported by the CGIAR MITIGATE+ project . K.F. was supported by the Bezos Earth Fund .
Scopus citations®
without self-citations
1