Citation:
Abstract:
Recent climatic shifts and the growth of agricultural land have escalated the incidence of wheat field fires, presenting severe risks to both food security and local economies. This study aims to develop advanced predictive models to effectively forecast significant wheat fires in Barika, Algeria. We utilized a comprehensive dataset spanning from 2015 to 2023, which includes information on fire incidents and meteorological factors like temperature, humidity, precipitation, and wind speed. A sophisticated ensemble machine learning model was crafted, combining Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), and Random Forest (RF) in a stacked configuration to predict wheat fire events. Our analysis indicates that the ensemble model significantly outperforms traditional single-model approaches in terms of both accuracy and reliability. Employing these cutting-edge predictive techniques significantly bolsters firefighting measures, enhances resource management, and reduces the adverse effects of fires in agricultural zones. The employment of ensemble learning highlights its utility as a formidable tool in environmental management and crisis response. With more precise forecasts, this model facilitates improved emergency preparedness and strategic intervention plans, aiming to safeguard essential agricultural assets and support rural communities against the backdrop of mounting environmental pressures.