Analyzing the Prominent Environmental Factors that Cause Forest Fires in the Algerian and Montesinho Forests

Abstract

Forest fires are becoming increasingly common due to rising temperatures from global warming. According to the National Interagency Fire Center (NIFC), in 2022, 66,225 fires in the United States burned 7,534,403 acres of land. Data from the Global Forest Watch (GFW) indicates that in the last 10 years, around 82 million hectares of forests have been destroyed by wildfires worldwide. In this paper, I use a random forest classifier to predict the occurrence of a forest fire given a set of environmental conditions (Fine Fuel Moisture Code, Duff Moisture Code, Drought Code, Initial Spread Index, temperature, relative humidity, wind speed, rain, and the month in which the data was collected). A random forest classifier uses an ensemble of decision trees—which ask a series of binary questions to split the data—to determine whether a fire is likely, and displays which factors are most important in making this decision. My model achieves 100% accuracy in predicting whether a forest fire occurs given a set of environmental conditions from the Algerian Forest Fires dataset, and achieves 55.78% accuracy when given the Montesinho Forest Fires dataset. Knowing the prominent factors responsible for wildfire formation is useful for devising measures to offset the conditions and prevent harm. In the future, we can compare environmental data from more forests around the world to form a holistic view of the environmental factors that cause forest fires.