Wildfires are becoming a more relevant problem in Sweden, especially in forest-dominated regions where climate, vegetation, topography and human activity influence where fires can occur. Gävleborg County is a suitable study area because of its large forest areas and previous wildfire events. GIS-based methods can therefore be useful for identifying areas with higher wildfire susceptibility.

The aim of this study was to create a wildfire susceptibility map for Gävleborg County using the Frequency Ratio (FR) method and to compare two model versions: one where only water bodies were excluded and one where both water bodies and urban areas were excluded. The analysis was based on twelve environmental and anthropogenic factors together with VIIRS active fire detections from NASA FIRMS.

The results showed that the water-only model achieved the highest validation performance, with a point-based ROC-AUC value of 0.8336 and a class-based prediction-rate AUC of 0.9446. The model where both water and urban areas were excluded produced slightly lower values, with a point-based ROC-AUC of 0.8077 and a class-based AUC of 0.8367. The results also showed that factor importance changed between the two models. Distance to urban areas, forest type, precipitation and wind speed were among the most influential factors, while slope and aspect had a relatively small influence.

Overall, the study shows that the FR method can be used for wildfire susceptibility mapping in Gävleborg County and that constraint choices can have a clear impact on both model performance and susceptibility patterns. The study contributes by demonstrating how different constraint approaches influence wildfire susceptibility results, and the resulting maps can support wildfire risk assessment, preparedness planning and resource prioritisation.