Files

No attached file found for this publication.

Details

Authors
Abstract
Tropical moist forests face persistent threats from human activities, necessitating their continuous monitoring. As optical satellite assessments struggle with cloud cover, Sentinel-1 (S1) C-band SAR archive provides a reliable alternative for yearly-based forest loss evaluation. This study introduces the MeanB<P10 index, computed from S1 imagery, as a generic index for change detection and forest assessment. First, the index is presented comparatively to common median and 10th percentile composites. As a second step, the proposed method is calibrated and assessed on two distinct regions of the Democratic Republic of Congo to assess its transferability. Ongoing calibration results reveal a significant reduction in false positives, with a peak F-score of 76.46%, with balanced false positive and false negative, ensuring a relevant estimation of the forest loss surface area. Ongoing accuracy assessments of the index from independent optical data promise nuanced insights into metric performance, expanding the applicability of the model to diverse landscapes.
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Delhez, B. (2024). Computing the Best Sentinel-1 SAR-Based Annual Index to Provide Robust Forest Loss Assessment in Tropical Region. https://hdl.handle.net/2078.5/240786