Wildfires are responsible for alterations in biodiversity, soil degradation, climate changes and put risk to human beings. In the last years, greenhouse gasses achieved an alarming rate in the atmosphere. Deforestation is the second largest anthropogenic cause of this increase and wildfires are one of the main methods. Observing this scenario, researchers have been conducting studies that try to forecast wildland fire. However, many of these investigations use traditional forecasting models and data mining techniques. These studies also usually use local data and might miss important information for the prediction. In this project, we intend to use climatological data to study the relationship between climate and wildfires using complex network theory and data mining. The main advantage of this approach is the capability of studying the interactions and dynamics between the small parts that compose a complex system. It may lead to scientific discoveries that are hardly detected by traditional techniques. After the network construction, we will adapt some pattern detection methods to the context of the project. The discovered patterns will be studied aiming at finding explanations for the wildfire intensity variations. These patterns will also serve as the basis for the creation of more accurate wildfire prediction methods. These new methods can be used to create better prevention policies, management, and control of wildland fire.
News published in Agência FAPESP Newsletter about the scholarship: