Fire in the Atlantic Rainforest: an analysis of 20 years of fire foci distribution and their social-ecological drivers

Abstract

The Atlantic Rainforest biome, where fire events were originally less frequent, is now increasingly at risk of burning due to anthropogenic influences on the climate and landscape. Identifying fire patterns and their leading causes in tropical forests is essential to improve prevention and fire management. We applied a filtering algorithm to remove redundant heat signals from the National Institute for Space Research (INPE) fire foci database, generate fire density maps after validating the data, and build Random forest (RF) regression models to identify fire risk and the variables that best explain the fire patterns. Results showed high redundancy of heat signals and that the minimum detection size for fire scars was lower for forests than other land covers. Mean annual temperature higher than 22ºC was associated with a high risk of fire and tree cover greater than 80% with low fire risk. Our study reveals a fire pattern that can expose the last remnants of the Atlantic Rainforest in Rio de Janeiro to fire, including inside Protected Areas. Highlights The redundance in the fire foci dataset was ∼46%; Protected areas have alarmingly high densities of heat signals compared to the unprotected landscape; The minimum fire scar size detected was lower for forests than agriculture or grasslands land cover; Temperature higher than 22 °C was an important threshold to determine fire risk; High tree cover (>80%) characterized areas with the lowest densities of heat signals.