Uso de múltiplas plataformas orbitais na detecção e validação de focos de queima em uma área do cerrado
Data
Autor(es)
Orientado(es)
Título da Revista
ISSN da Revista
Título de Volume
Editor
Abstract
The study of wildfires in the Cerrado is essential for the preservation of this biome, as it enables faster and more precise actions in monitoring and combating forest fires. This study aimed to evaluate the accuracy of satellite data in detecting wildfires and to verify whether the convergence of multiple satellites on a single hotspot indeed represents a real fire event. The research was conducted at the Uruçuí-Una Ecological Station, located in southern Piauí, a region characterized by high fire incidence and significant anthropogenic pressure resulting from agricultural activities. Data on fire hotspots were obtained from the Brazilian National Institute for Space Research (INPE), while the burned areas (burn scars) were delineated based on the MapBiomas project, both referring to the year 2021. The analyses included a comparison between active hotspots and false alerts, applying the kernel density technique to all identified hotspots. The results highlighted the importance of integrating information from multiple satellites to identify real fire events and demonstrated the potential of kernel density as a complementary indicator to distinguish genuine wildfires from false alerts. The Terra and Aqua satellites were more efficient in detecting smaller fires, while NOAA and GOES satellites were more effective in identifying fires in larger areas. The NPP-375 satellite exhibited greater accuracy in detecting active hotspots, despite its lower revisit frequency. The application of kernel density, considering the maximum density value, revealed a correlation above 60% in differentiating active hotspots from false alerts. This method allowed the identification of areas with higher fire concentration, highlighting more intense wildfires and contributing to the optimization of firefighter allocation in fire combat efforts.
