Data and Model Supplement: Driver Analysis of Subarctic Wildfire Severity over a 35-year Period

Large subarctic wildfires are causing environmental damage, releasing stored carbon, and forcing residents to relocate. Subarctic ecosystems are experiencing earlier, longer, and more intense wildfire seasons due in part to factors such as warmer winters and the broader spread of damaging insects. There is limited agreement within existing research on the type of drivers and degree of influence that climate, vegetation and topographic factors have on wildfire severity. As a result, existing research has been limited through using small datasets and applying a limited number of drivers. This study aims to address these gaps by improving upon existing methodological limitations and quantifying the influence of a more comprehensive list of climate, vegetation, and topographic variables on wildfire severity. An XGBoost regression model with an r2 of 0.7 was trained, and shapely values were used to investigate the combined variable contributions to predictions of wildfire severity. This research identified variable importance trends unique to predicting subarctic wildfire severity in rugged regions with cold annual temperatures and short growing seasons. Important variables not previously identified include, skin reservoir content, evaporation from vegetation transpiration, wind exposition index, soil temperature, and visible sky percentage, in addition to variables found by existing research, namely pre-fire vegetation, wind speed, topographic position index and land cover. This research helps to build consensus on the factors driving severe wildfires in subarctic ecosystems, and the methods developed could become the basis for future study.