Do Land Surface, Vegetation, and Climate have Causality with Wildfire in Boreal Forests?

The increasing trend of wildfire events in higher latitudes is a growing threat to the boreal forest and global ecosystem under climate change. Land surface (e.g., albedo, snow), biophysical (e.g., leaf area index-LAI, gross primary production-GPP), and climate (e.g., evaporative stress index-ESI, precipitation, maximum temperature-Tmax, vapor pressure deficit-VPD) factors play a major role in occurrences of wildfire in the boreal forests. However, limited studies have focused on the long-term causal relationship between these covariates and wildfire and vice versa. In this study, we used the convergent cross mapping (CCM) approach to detect the bidirectional causality between these environmental covariates and wildfire. CCM accounts for time-delayed causality in both directions between the variables which is useful to study feedback. Results showed that a causal relationship exists between wildfire and environmental factors and vice versa. CCM identified a strong effect of Tmax, albedo, snow, ESI, and VPD on fire. Similarly, wildfire had a strong effect on ESI, temperature, LAI, albedo, VPD, and GPP indicating potential changes in surface energy balance and ecosystem processes after wildfire events in the boreal forests region. To understand the contribution of environmental factors to wildfire and vice versa, random forest models were used to calculate the variable importance for the causality strength across the North American boreal forests. Random Forest models informed that cross-map skills were mainly controlled by climate (e.g., drought, minimum temperature), soil (e.g., soil organic carbon), terrain (e.g., elevation), hydrological (e.g., soil moisture), and biophysical properties (e.g., GPP). These results can be useful for identifying the hotspots of fire events, understanding feedback mechanisms between wildfire and climate, and designing wildfire forecasting systems in boreal forests in the future.