Modelling susceptibility to human-caused forest fire ignitions in Austria

In the European Alpine region forest fire ignitions are driven by changing environmental conditions and increased intensity of outdoor activities by man. Therefore, accurate assessment and prediction of the ignition danger specifically for the complex mountain terrain becomes crucial. This study develops an Austria-wide model for estimating the conditional susceptibility of human-caused fire ignition with a spatial resolution of 100 meters using ignition points from a national forest fire database and a logistic regression framework that integrates socio-economic and infrastructural predictors. Model performance was assessed using random train–test splitting, ROC/AUC analysis, and independent temporal evaluation with human-caused fire events from 2021 to 2025 and checked for plausibility in a regional case study. The resulting map of conditional susceptibility to human-caused fire ignitions was aggregated to the municipality level to provide a contextual comparison with an existing fire-probability model.Population density and transport related infrastructure such as railways, transregional road network, cable cars, and secondary roads emerged as significant predictors of human-caused ignitions, reflecting the importance of forest accessibility. The results reveal clear spatial patterns in human-caused ignition susceptibility, with most forest areas showing low susceptibility values at the national scale, while higher susceptibility values are concentrated around settlements and along major access corridors. These areas where human activity and forest accessibility may increase the likelihood for ignitions provide a sound basis for targeted fire prevention, monitoring, and planning, particularly in the wildland–urban interface (WUI).