Abstract:

To clarify the spatiotemporal distribution patterns of major and catastrophic lightning-caused forest fires in the Greater Khingan Mountains and the main meteorological drivers,a three-temporal scale (daily,monthly,and annual) was set up.Using forestry bureaus as the basic statistical units to quantitatively analyzed the spatiotemporal distribution characteristics from 2002 to 2023.Logistic regression models and random forest models were used to identify the main meteorological drivers factors for major lightning-caused fires in the Greater Khingan Mountains and to investigate the influence of meteorological elements in different periods (T₀—T₇) on the occurrence of major lightning-caused fires.The results showed that:1) A total of 77 major and catastrophic lightning-caused fires occurred from 2002 to 2023,mainly from May to September,with June as the peak period.Spatially,they were concentrated in Huzhong,Qiqian and Mangui,especially in Huzhong.2) Mean temperature is the dominant meteorological driver,followed by evaporation,mean relative humidity,and sunshine duration.3) Both the logistic regression and the random forest model performed well (AUC values were all greater than 0.85),and the random forest showed stronger discriminatory ability (AUC=0.964),and higher prediction accuracy (ACC=0.891).4) Meteorological variables 5 days before the fire occurrence (T₅) had the greatest impact on the occurrence of major catastrophic lightning-caused forest fires in the Greater Khingan Mountains.The results provide theoretical basis and reference for the prevention and control of major and catastrophic lightning-caused forest fires in the Greater Khingan Mountains.

Key words: Greater Khingan Mountains, major and catastrophic lightning-caused forest fires, spatiotemporal distribution, meteorological drivers