Prioritizing the Right to Environment: Enhancing Forest Fire Detection and Prevention Through Satellite Data and Machine Learning Algorithms for Early Warning Systems

Forest fires pose significant threats to ecosystems, human settlements, and biodiversity, necessitating advanced and effective detection systems. Traditional methods of fire detection, such as ground-based observations and aerial patrols, are often limited by their scope and response time. In this study, we present a comprehensive approach for forest fire detection utilizing satellite data, image processing techniques, and advanced machine learning models. Our proposed forest fire hybrid detection model (FFHDM) combines random forest (RF), support vector machine (SVM), and convolutional neural networks (CNN) to enhance detection capabilities. Landsat satellite images serve as the primary data source, offering high spatial resolution crucial for detailed land cover analysis and long-term monitoring. We enhance image quality using Gaussian filtering to suppress noise, thereby improving data accuracy. Min–max normalization is employed to standardize images, ensuring consistent brightness and contrast for comparative analysis. Image segmentation via k-means clustering isolates forested areas, refining the focus on relevant regions. For feature extraction, the normalized burn ratio (NBR) is used to detect fire-affected areas, supported by texture analysis using the gray level co-occurrence matrix (GLCM) to differentiate land cover types. Our FFHDM achieves an impressive accuracy of 98%, providing a robust and reliable system for early fire detection and efficient forest management.