Country-Level Modeling of Forest Fires in Austria and the Czech Republic: Insights from Open-Source Data

Forest fires are becoming a serious concern in Central European countries such as Austria (AT) and the Czech Republic (CZ). Mapping fire ignition probabilities across can be useful tool for risk mitigation. This study was conducted to: (i) evaluate contribution of variables obtained from open-source datasets (i.e., MODIS, OpenStreetMap, WorldClim) modeling probability at country level; (ii) investigate how well Random (RF) method performs one to another. The importance predictors evaluated using Gini impurity method, RF ROC-AUC confusion matrix. most important were topographic wetness index AT model slope CZ model. AUC values validation sets 0.848 (AT model) 0.717 (CZ model). When respective models applied entire dataset, they achieved 82.5% 66.4% accuracy. Cross-comparison revealed that may successfully dataset (AUC = 0.808, Acc 82.5%), while showed poor explanatory power when 0.582, 13.6%). Our provides insights into effect accuracy completeness data on reliability national-level forest assessment.