earticle

영어

Forest fires significantly impact forest ecosystems through tree mortality, necessitating scientific understanding of factors influencing post-fire survival for effective forest management. Using three predictive modelling approaches, this study identified critical factors affecting post-fire survival in Pinus densiflora (Korean red pine), a dominant species in Korean forest. We examined 734 fire-damaged pine trees across four burn sites, collecting data through field surveys, remote sensing, and meteorological observations. Three predictive models (logistic regression, Cox proportional hazards, and random forest) were compared to identify survival factors. The random forest model demonstrated superior performance with an AUC of 0.924 and sensitivity of 0.8919, due to its ability to capture non-linear relationships between variables. For tree-specific factors, diameter at breast height (DBH) and bark scorch index (BSI) emerged as key mortality indicators. Among topographic and meteorological factors, slope, elevation, isothermality, and temperature seasonality significantly influenced survival outcomes. Survival probability ranged from 16.9% to 77% depending on these variables. Notably, the influence of these factors on mortality became more pronounced 12 months after fire occurrence. This research provides a scientific basis for ecological management and restoration strategies in fire-affected P. densiflora forests, which constitute a significant portion of South Korea's forest ecosystem.