Spatial modeling of fire in the atlantic forest considering future climate change scenarios in Rio Grande do Sul state – Brazil

Abstract

Several biomes worldwide have a long history of conflict over fire use and management. The Atlantic Forest Biome (AFB) is a Brazilian biome that suffers the most from land use and cover changes that cause environmental degradation. The impact of climate change is expected to exacerbate this situation, with extreme weather events potentially leading to a higher frequency and intensity of fire. The main goal of this study is to understand the spatiotemporal distribution of fires for future climate scenarios obtained from CMIP6 climate simulations in the AFB in Rio Grande do Sul (RS) state, Brazil, using machine learning algorithms. This study selected several environmental and anthropogenic variables as factors associated with the cause, occurrence, and spread of fire. The results showed an uneven fire density distribution in the study area. An extensive fire cluster was found in pasture areas located northeast of RS state, reaching more than 1500 fire foci per km2 on average per year. The final model had a training R2 value of 0.99 and a test R2 value of 0.93. The most significant variable identified by the model was the average maximum temperature during the warm period, while livestock is the most influential economic activity. Regarding the simulated fire densities, the period between 2021 and 2040 in the SSP 5.8–5 scenario displayed maximum values that were equivalent to those observed in 2018, with an expansion in the occurrence region observed for the same scenario. However, unexpectedly, between 2081 and 2100, fire density decreased across all areas under the SSP 5.8–5 scenario. This study provides useful insights into climate change context scenarios, offering valuable insights into the intricate relationship between natural processes and human influences, ultimately contributing to informed decision-making and sustainable environmental management.