Eucalypt plantations are expanding in Brazilian savanna, one of the most frequently burned ecosystems in the world. Wildfires are one of the main threats to forest plantations, causing economic and environmental loss. Modeling wildfire occurrence provides a better understanding of the processes that drive the fire activity. Furthermore, the use of spatially explicit models may promote more effective management strategies and support fire prevention policies. Machine Learning (ML) algorithms have shown their predictive accuracy to model wildfire occurrence. In this context, we implemented and compared Random Forest ML algorithm with Maximum Entropy (MaxEnt) and traditional methods like Logistic Regression (LR) to predict the ignition likelihood across the study area. The models were trained using several explanatory drivers related to fire ignition: accessibility, proximity to agricultural lands or human activities, among others. Specifically, we introduced the progression of eucalypt plantations on two-year basis. Fire occurrences in the period 2010-2016 were retrieved from the Brazilian Institute of Space Research (INPE) database. Results suggest that the use of ML algorithm leads to an improvement in the accuracy in terms of the AUC (area under the curve) of the model when compared to MaxEnt and LR outputs. The ML model denoted fairly good predictive accuracy (AUC ≈ 0.72) being useful for evaluating the current occurrence likelihood distribution and unravel fire occurrence driving forces across the studied area. Results suggested that fire occurrence likelihood was mainly linked to proximity agricultural and urban interfaces. Even though eucalypt plantations were not the main explanatory factor, it contributed to increasing wildfire likelihood.
No datasets are available for this submission.