IMPACT OF CLOUDS ON FOREST DEGRADATION MEASUREMENT ERRORS USING THE LINEAR SPECTRAL MIXING MODEL IN SENTINEL-2A IMAGERY

Forest degradation affects ecological integrity and carbon storage, influencing climate regulation. This study evaluated degradation from wildfires and logging in the Gurupi Biological Reserve, eastern Amazonia, using Sentinel-2A imagery and a linear spectral mixing model (LSMM). The model decomposed pixels into vegetation, soil, and shade fractions to detect degraded areas. However, cloud cover reduced accuracy, especially in areas with high cloud probabilities. The error fraction showed overestimations where image fraction ratios ranged between 50% and 70%. As cloud cover exceeded 50%, both error amplitude and standard deviation increased, highlighting the impact of clouds on LSMM results. These findings stress the need for improved data or alternative methods to enhance degradation monitoring in cloud-prone regions.