December 23, 2021 9:54 pm
A data-driven prediction model for Fennoscandian wildfires
Sigrid Jørgensen Bakke, Niko Wanders, Karin van der Wiel, and Lena Merete Tallaksen
Nat. Hazards Earth Syst. Sci. Discuss., https//doi.org/10.5194/nhess-2021-384,2021
Preprint under review for NHESS (discussion: open, 0 comments)
In this study, we developed a machine learning model to identify dominant controls of wildfire in Fennoscandia, and produce monthly fire danger probability maps. The dominant control was shallow soil water anomaly, followed by air temperature and deep soil water. The model proved skilful with a similar performance as the existing Canadian Fire Weather Index. We highlight the benefit of using data-driven models jointly with other fire models to improve fire monitoring and prediction.