Wildfire risk assessment in Sichuan Province, China: hazard modeling approach considering different combinations of classification criteria and connection values of factor attributes

Background Current wildfire risk research has primarily focused on hazard assessment, lacking a comprehensive framework that integrates vulnerability and adaptive capacity. Moreover, the influence of different statistical connection methods and classification criteria of factor attributes on hazard assessment has been overlooked. Aim Taking Sichuan Province, China, as the study area, a comprehensive wildfire risk assessment model was constructed based on the hazard-vulnerability-adaptive capacity framework, with special focus on the effects of differences in connection methods and classification criteria of factor attributes on the modeling performance of wildfire hazard. Method The impact of six connection methods integrated with logistic regression (LR) on wildfire hazard assessment was explored using wildfire samples/whole region as classification criteria. Vulnerability and adaptive capacity were analyzed using techniques for ranking preferences by similarity to ideal solutions (TOPSIS), coupled with combination weights and integrated with the optimal hazard model, resulting in an integrated risk assessment framework. Key results and conclusions Significant differences between hazard assessment results based on different classification criteria and connection methods were found. The Point-IV-LR model, constructed using wildfire samples as classification criteria and utilizing information value (IV) coupled LR, performed the best. The risk assessment highlighted southwestern mountains as critical high-risk areas. Implications These findings provide targeted wildfire prevention strategies tailored to different risk levels in Sichuan Province.