Evaluating the performance of spectral indices on high-resolution satellite data to map low-intensity forest management

Exploring alternatives for analyzing spatial patterns of low-intensity forest management in a Brazilian Amazon forest under sustainable exploration, we evaluate the combination of high-spatial resolution imagery and spectral vegetation indices as a potential approach for detecting disturbances in the forest canopy. In this study, we utilized Planet NICFI mosaics alongside the Normalized Difference Vegetation Index (NDVI), Soil-Adjusted Vegetation Index (SAVI), and Visible Atmospherically Resistant Index (VARI) in a forest management unit of Jacundá forest. We calculated each vegetation index and defined thresholds to create a binary mask representing two classes: (1) disturbed forest and (2) undisturbed forest. Subsequently, we masked and calculated the managed area captured by each vegetation index and compared these results with validation ground data to assess the performance of each index. It was possible, using binary thresholding, to observe which index presented greater accuracy in detecting low-intensity forest management patterns.