Small-scale forest fire risk zoning based on bivariate statistics and multi-criteria decision analysis

Fires pose serious threats to human life, forest environments, and biodiversity. Small-scale regional forest fire risk mapping is crucial for fire management. We combined bivariate statistics (weight of evidence, WOE, statistical index, SI) with multi-criteria decision analysis (analytic hierarchy process, AHP, analytic network process, ANP) to construct new WOE-ANP and SI-ANP comprehensive models to conduct forest fire risk zoning in Wangmo County, Guizhou Province. The results showed that most areas in the southern, western, and northern parts of Wangmo County were highly prone to forest fires, with fire risk of regions classified as level 4 or above accounting for 39.2%. Fire risk was severe in the county. The comprehensive models effectively enhanced the predictive ability of single bivariate statistical models. Compared to AHP, ANP provided more reliable assessments for the weight of forest fire risk factors. The WOE-ANP and SI-ANP comprehensive models demonstrated high accuracy (84.3% and 83.8%, respectively) in assessing forest fire risk, offering more reliable decision support and reference for forest fire management.