Abstract. As a result of climate change, climatic catastrophes, such as wildfires, are likely to increase. Wildfires continue to occur frequently and spread with greater intensity due to extreme weather conditions. In recent years, explosive fire growths have been reported in the United States, Australia, and other parts of the world. A combination of climate change and human activity has caused the semi-arid forestry areas in Iran's northern provinces to become more desiccated, leading to an increase in wildfires. The accuracy of the resulting fire susceptibility maps (FSMs) will directly be related to the performance of the method classifier. In this study, we use an ensemble classifier to model the FSM for a selected forestry case study area in one of the northern provinces of this country. Therefore, FSM is generated based on established criteria using the ensemble model. With Decision Trees, K nearest neighbor, and Logistic Regression, the ensemble model was created using the soft-voting method. A forest fire inventory data is created based on data collected over five years using GPS and the MODIS thermal anomalies product for training and testing the applied approach. The K-fold method was used for validation, and the resulting FSM was validated using five accuracy assessment metrics. The best result from the area under the curve (AUC) yields 93% for fold 9, and the mean AUC for ten folds yields 88%.

Abstract. This study aims to identify key fire factors via recursive feature elimination (RFE) to generate a forest fire susceptibility map (FSM) using support vector machine (SVM), and random forest (RF) models. The fire zones were derived from MODIS satellite imagery from 2012 to 2017. Further validation of these data has been provided by field surveys and reviews of land records in rangelands and forests; a total of 352 fire points were determined in this study. Seventeen factors involving topography, geomorphology, meteorology, hydrology, and anthropology were identified as being effective primary factors in triggering and spreading fires in the selected mountainous case study area. As a first step, the RFE models of the RF, Extra trees, Gradient boosting, and AdaBoost was used to identify important fire factors among all selected primary factors. The SVM and RF models were applied once on all factors and the second on those derived from RFE models as the key factors in FSM. Training and testing data were divided tenfold, and the model's performance was evaluated using cross-validation (CV). Different metrics were used to measure accuracy, including recall, precision, F1, accuracy, the area under the curve (AUC), Matthews correlation coefficient (MCC), and Kappa. The accuracy assessment process shows that the FSM results are further improved by leveraging RFE models to distinguish the key factors and not include unnecessary factors. The greatest improvement is for SVM, with more than 10.97 % and 8.61 % in the accuracy and AUC metrics, respectively.