Forest Fires Burned Area Monitoring with AI Machine Learning Algorithms

Abstract

Forest fires pose significant dangers and can have devastating effects. The major consequences of large-scale forest fires include but are not limited to, the injury or death of humans and wildlife, destruction of homes and infrastructure, loss of wildlife habitats, and the release of substantial amounts of greenhouse gases (such as carbon dioxide) into the atmosphere [1]. Due to factors like climate and weather conditions, completely preventing forest fires is often impossible, making it crucial to monitor burned areas and detect large-scale fires. Traditional methods for detecting and monitoring forest fires typically rely on local sensors [2], which can be both expensive and inefficient. Our aim is to develop an accurate AI machine learning model to predict the size of the burned area in a forest fire, thereby aiding in scale monitoring. The model is built using an open-access dataset from Montesinho Natural Park in Portugal [2]. The methodologies employed in building the models include multiple regression, SVR regression, ANN regression, KNN classification, and SVM classification. By using these ML models, forest fires can be detected and monitored more efficiently in real time, potentially reducing property damage, optimizing resource allocation, and mitigating environmental harm.