A systematic literature review on forest fire susceptibility mapping using geo-spatial technology and future research directions

Abstract Forest Fires pose escalating environmental and economic challenges, demanding accurate predictive modeling for effective mitigation. Recent advances in geospatial technologies and machine learning (ML) have improved Forest Fire-susceptibility mapping by integrating multispectral satellite imagery with advanced algorithms. This systematic review synthesizes studies published between 1992 and 2024 that apply ML and deep-learning approaches—such as XGBoost, LightGBM, CatBoost, and AdaBoost—to Forest Fire risk assessment. The review analyzes methodological evolution, data quality, and spatial–temporal modeling practices, highlighting the strengths and limitations of hybrid geospatial frameworks. Persistent challenges include data scarcity, model generalizability, and interpretability, while emerging opportunities lie in explainable AI (XAI) and climate-aware modeling. By consolidating cross-disciplinary insights, this review provides a conceptual and methodological foundation for enhancing the accuracy, scalability, and applicability of Forest Fire-prediction systems.