Work

Mapping large-scale ecosystem degradation using a sentinel-2-based browning vegetation index

Year: 2025

Type: article

Abstract: Browning vegetationan early-warning indicator of ecosystem degradation and environmental stress—poses particular challenges when monitored over large spatial extents. Conventional indices like the Normalized Difference Vegetation Index (NDVI) suffer from saturation effects and struggle to detect subtle or early-stage browning, while most stress-sensitive indices target single stressors and yield inadequate results across broad areas. To overcome these limitations, we first collected field spectra and applied linear spectral unmixing to characterize vegetation at multiple browning levels. Building on these insights, we introduce a novel Browning Vegetation Index (BVI) tailored for Sentinel-2 imagery, which leverages the red-edge band’s high chlorophyll sensitivity alongside near-infrared (NIR) and shortwave infrared (SWIR) bands to maximize spectral contrast among browning stages.We demonstrate BVI’s capability for large-scale mapping in the Tianshan Mountains, where—through Canny edge detection and Otsu adaptive thresholding—we traced the annual spatial transition from healthy green to browned vegetation. Validation with high-resolution Gaofen-2 data across four Tianshan subregions yielded precision, recall, and F1 scores above 0.85 and overall accuracy of 0.883. Against 11 established indices, BVI consistently delivered superior sensitivity to differing browning intensities. Finally, cross-continental tests in the United States, Afghanistan, India, and China confirm BVI’s robustness for fine-scale extraction of browning vegetation across vast and varied landscapes.

Source: Ecological Indicators

DOI: 10.1016/j.ecolind.2025.114083