Wildfire Risk Forecasting Using Weights of Evidence and Statistical Index Models

The risk of forest and pasture fires is one the research topics interest around world. Applying precise strategies to prevent potential effects minimize occurrence such incidents requires modeling. This was conducted in city Sanandaj, which located west province Kurdistan Iran. In this study, fire assessed using weights evidence (WoE) statistical index (SI) models. Information about Sanandaj (2011–2020) divided into two parts: educational data (2011–2017) validation (2018–2020). Factors considered for rangeland included altitude, slope percentage, direction, distance from road, river, land use/land cover (LULC), average annual rainfall, temperature. Finally, order validate models used, receiver operating characteristic (ROC) curve used. results WoE SI showed that 62.96% 52.75% study area, respectively, were moderate very high classes. addition, ROC analysis had area under (AUC) values 0.741 0.739, respectively. Although input parameters both same, model a slightly higher AUC value compared model, can potentially be used predict future area. help decision makers managers take necessary precautions and/or damage.