Assessment of active fire detection in Serra da Canastra National Park using MODIS and VIIRS sensors

Abstract. The geographic data utilization for the occurrence of wildfires and forest fires for monitoring fire usage in vegetation has become increasingly important for generating information that aids in decision-making and policy development regarding climate change and its impacts. This study aims to validate the active fire data processed by INPE’s Queimadas Program at Serra da Canastra National Park during the first semester of 2023. This period is characterized by lower intensity of fire thermal radiation due to meteorological and climatic factors such as lightning, precipitation, and temperature. We compared active fire data from the MODIS and VIIRS sensors with 220 fire scars mapped by the Sentinel-2 satellite. Additionally, we analyzed the distribution of active fire data across different classifications of fire scar size. The validation rate was 99.4% for VIIRS and 95% for MODIS, with few commission errors (false positives). However, omissions (false negatives) were observed in detecting scars smaller than 1 km2, related to both sensor characteristics and fire behavior. The results were satisfactory, indicating that future studies can further refine this research.