Publicações

Fire has been an intrinsic ecological component of the ecosystems, affecting the public, economic, and socio-cultural policies of human-nature interactions. Using fire over grassland vegetation is a traditional practice for livestock in the highland grasslands and has economic and environmental consequences that have not yet been understood. A better description of the spatio-temporal biomass burning patterns is needed to analyze the effects of creation and application in these areas. This study used remote sensing techniques based on Sentinel-2 data and machine learning algorithms to identify burning scars and compare them with a national fire collection database for the highland grasslands in the Atlantic Forest Biome in Brazil. The aim is to evaluate public management tools and legislation evolution during the 35 years of the time series analyzed. The results indicated that 12,285 ha of grasslands were converted to other uses, losing about 24% of their original formation, with 10% occurring after banned this practice in 2008. The burned areas classification using the Random Forest algorithm obtained an AUC = 0.9983. Divergences in the burned area’s extent and frequency were found between the municipality’s authorized license and those classified as burned. On average, only 43% of the burned area in the Parque Estadual do Tainhas and its buffer zone had an environmental permit in the last 5 years. This research’s results provide subsidies for revising and creating public policies and consequently help territorial management.