Abstract
This study was conducted to gain a better understanding of factors influencing regional-scale wildfire probability and its spatial distribution. The random forest model was employed to investigate the spatial association between historical fire events in the Zagros Mountains (Iran) and different geo-environmental factors to characterize regional-scale trends of wildfire occurrences. Among the factors considered, proximity to human settlements (Gini = 9.38), annual rainfall (Gini = 8.85), altitude (Gini = 7.41), and proximity to roads (Gini = 7.40) were identified as the most effective factors. To predict future ignitions and to delineate the study area to different levels of wildfire probability, the support vector machine model was applied and proved to be an effective model with satisfactory accuracy (success rate = 0.81; prediction rate = 0.75). The insights obtained from this research can be applied to spatially explicit assessment of fire-prone landscapes and decision-making for wildfire management.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have