A methodology for investigating trends in changes in the timing of the fire season with applications to lightning-caused forest fires in Alberta and Ontario, Canada

Abstract

Lightning-caused fires account for approximately 45% of ignitions and 80% of area burned by forest fires in Canada. Investigating the seasonality of these fires and the extent to which it may be changing over time is of interest to both fire managers and researchers. In this project, we develop flexible models for describing the temporal variation in the risk of lightning-caused fires. Generalized additive models are first used to obtain smooth estimates of fire risk by Julian day for each year. Inverse calculations are then employed to obtain point and interval estimates of the start and end of the fire season annually; these are defined by the crossing of fire risk thresholds. Finally, permutation-based methods are used to test for significant linear trends in the start and end of the fire season. This methodology is applied to historical forest fire records in Alberta, Canada, and the western and eastern subregions of Ontario, Canada. Our results suggest significant changes to both the start and end of the fire season in Alberta and a significant change to the end of the fire season in western and eastern Ontario.