Abstract:

Objective: Global warming and the increasing frequency of extreme weather events have resulted in lightning-ignited forest fires with larger scale, higher intensity, and stronger destructiveness. This study investigates the spatiotemporal distribution and clustering evolution of lightning-ignited forest fires in the Daxing’anling Mountains, so as to provide a scientific basis for the prevention and management of lightning-ignited forest fires. Method: Based on historical records of lightning-ignited forest fires in the Daxing’ anling Mountains from 2013 to 2024, this study comprehensively applied statistical analysis, two-dimensional Gaussian kernel density analysis, and the geographical detector method to analyze the dynamic evolution patterns, spatial clustering characteristics, and explanatory power of key driving factors. Result: 1) A total of 791 lightning-ignited forest fires occurred during the past 12-year period, showing an overall fluctuating trend with a peak occurring in 2019. These fires were mainly concentrated between April and October, particularly during the spring fire prevention period and the summer growing season, with the earliest event on April 24 and the latest one on September 19. After the spring fire prevention period, the number of fires declined but rose again to a peak in mid-July. The period from 13:00 to 17:00 (excluding 17:00) was the high-incidence time, accounting for more than 50% of the events. 2) Spatially, lightning-ignited forest fires were significantly clustered in the northwestern of study area. Annual analysis revealed interannual variations in cluster locations. Using the natural breaks method, kernel density values were classified into five risk levels, with the southern region generally identified as very low-risk zones (only a few events occurred in 2022). The Getis-Ord Gi^* significance test confirmed that high-risk areas had stable spatial clustering characteristics. 3) The three-day mean temperature before a lightning-ignited fire, monthly maximum temperature, elevation, monthly mean temperature, 0?7 cm soil layer soil moisture, and normalized difference vegetation index (NDVI) were identified as the core driving factors with the strongest explanatory power across different years. Combinations such as relative air humidity and monthly mean temperature (which showed higher explanatory power nine times), as well as elevation and NDVI (seven times), generally exhibited strong explanatory power across the years, indicating that they have a significant influence on the occurrence of lightning-ignited fires. Moreover, the combinations of monthly mean temperature, monthly maximum temperature, and elevation with other factors also demonstrated high explanatory power. Conclusion: This study summarizes the recent occurrence trends and characteristics of lightning-ignited fires in Daxing’anling Mountains, which show overall fluctuating patterns over time. It is recommended to strengthen fire prevention efforts in summer in key forest areas, particularly around 14:00. Lightning-ignited fires in the Wuma, Yong’anshan, Tuqiang and Amur regions have exhibited persistent and significant clustering during the past 12-year period, although in certain years (e.g., 2015 and 2018) the fire events were more scattered and random, showing no significant clustering. The interactions vary among different factor types, with meteorological interactions being particularly prominent and playing a dominant role in the mechanism of lightning-ignited fire occurrence. This study reveals the spatiotemporal evolution patterns and key driving mechanisms of lightning-ignited forest fires, providing scientific basis for more refined and region-specific forest fire prevention and management strategiess.

Key words: lightning-ignited forest fires, driving factors, two-dimensional Gaussian kernel density, geographical detector, Daxing’ anling Mountains