International Research Journal of Education and Technology

Forest fires pose significant threats to ecological systems, wildlife habitats, and human settlements. Timely and accurate prediction of forest fire occurrences is crucial for effective disaster management and mitigation efforts. This study presents a hybrid forest fire prediction model that leverages the spatial simulation capabilities of Cellular Automata (CA) alongside the predictive strength of Machine Learning (ML) algorithms. The model is designed to capture both the dynamic spread patterns of fire across geographical terrains and the environmental variables contributing to fire ignition and propagation.

By integrating historical fire data, meteorological parameters, and vegetation indices, the ML component of the system learns to identify fire-prone zones and trigger probabilities. Meanwhile, the CA mechanism simulates the fire spread behavior across the terrain using localized interaction rules influenced by slope, wind direction, and fuel availability. This dual approach offers a more holistic understanding of fire dynamics, providing early warnings not just of potential fire outbreaks but also their likely progression across landscapes.

Extensive experimentation using datasets from wildfire-prone regions demonstrates the model's superior accuracy compared to conventional methods. Performance evaluation metrics such as accuracy, F1-score, precision, and recall validate the effectiveness of the proposed system in predicting both occurrence and spread of forest fires. The results suggest that combining CA and ML offers a robust framework for proactive forest fire management and could serve as a decision-support tool for environmental and emergency agencies.

Keywords: Machine Learning (ML), Cellular Automata (CA)