Assessing the environmental risk of lightning-induced forest fires under compound climatic extremes in Northeast China

Lightning-induced forest fires pose increasing environmental risks to boreal–temperate forest ecosystems under climate change, yet their impacts remain insufficiently incorporated into environmental risk assessment and planning. This study conducts a spatially explicit assessment of lightning-induced forest fire risk in the Da Hinggan Ling region of Northeast China using a 59-year fire record (1966–2024; n = 1047) integrated with climate reanalysis data. Fire occurrences were aggregated to a 5-km equal-area grid and analyzed within a GIS-based machine-learning framework using ERA5-derived climatic and environmental variables. An XGBoost classifier calibrated to probabilistic scales was applied to map ignition likelihood, while SHAP interpretation and partial-dependence analysis were used to identify dominant drivers and nonlinear response thresholds. Results indicate that lightning-fire risk is governed by compound interactions among thermal forcing, hydrological drying, vegetation stress, and atmospheric instability. Ignition probability peaks under moderate heat and dryness but declines under extreme aridity, revealing a self-limiting dryness regime that constrains further risk escalation. By linking long-term fire records with spatial risk modelling, this study provides a quantitative framework for integrating lightning-induced fires into environmental assessment and climate adaptation planning. The findings support risk-informed decision-making, early-warning strategies, and targeted management interventions for northern forest ecosystems under intensifying climatic extremes.