Abstract The manuscript addresses an existing gap in research on ignition probability conditions in rural and non-forest environments. A new rural fire risk index alternative to the Canadian Fire Weather Index (FWI) is introduced, integrating machine learning and fuzzy logic to improve the assessment of ignition potential in various vegetation types in rural settings. The work presents a computational algorithm to calculate the potential for fire occurrence in heterogeneous rural environments, considering static and dynamic variables, and differing from the FWI by focusing on the flammability risk of fuel on the ground based on vegetation cover. A detailed methodology is presented, including the development of the input dataset, the development of the flammability index for each area, and the visualization of the results. The satellite data used, the land cover classification by machine learning, and the flammability index definition are described. Our findings reveal that during a three-month evaluation period, the FWI underestimated risk in 20.70% of cases, overestimated in 34.90%, and aligned accurately in 44.69%. This model enhances assessment accuracy and equips natural resource managers and local authorities with effective tools for informed decision-making regarding fire prevention and mitigation strategies. The research demonstrates that the proposed index significantly improves the prediction accuracy by incorporating the flammability of various vegetation and localized climatic variables. The index was validated in Misiones, Argentina, a region characterized by diverse vegetation and frequent rural and forest fires. By refining fire risk evaluations, this research aids in protecting natural ecosystems and communities while advancing wildfire management practices through localized assessments. This research offers a novel perspective on fire risk assessment models, providing valuable insights that broaden the current understanding of flammability. The implications of this work are profound, offering a customized approach to wildfire risk assessment that can improve prevention and mitigation strategies. The study’s findings contribute to preventing rural fires, highlighting areas for future research and possible policy or practical applications.
Brys et al. (Tue,) studied this question.