Analyzing forest fires in a brazilian savannah conservation unit using remote sensing and statistical methods: spatial patterns and interaction

The objective of the study was to analyze the occurrence of forest fires in a conservation unit (CU) of the Brazilian savannah using remote sensing techniques and statistical methods developed for spatial punctual processes. To conduct the spatial analysis of fires, fire polygons mapped using Landsat 8 satellite images were used. The fires were considered into size classes to better illustrate the spatial patterns. The analysis of the spatial distribution of fires utilized Ripley's K-function, in addition to the Kcross function to verify spatial interaction. The results show that the year 2015 had the highest number of fires and burned area. Smaller fires represent a greater number of occurrences, located mostly on CU boundaries. The spatial distribution of forest fires is not random and can cluster on a scale of approximately 6 km. There is a strong spatial interaction between forest fires and traditional communities, particularly with fires smaller than 100 hectares. However, these communities are not responsible for large fires. These results contribute to better-targeted forest fire prevention and combat policies, serving as management tools for the protected area.

Graphical Abstract