Modeling aspect-associated microclimate variation and time-lag effects on surface fine fuel moisture dynamics in subtropical forests

Background Surface fine fuel moisture content (SFFMC) strongly influences ignition potential and wildfire behavior, yet its sub-daily dynamics and terrain-related variability remain poorly quantified in humid subtropical forests. We used a year-long bihourly dataset from Masson pine ( Pinus massoniana ) stands in southern China to characterize diurnal and seasonal SFFMC variation across slope-aspect classes and to evaluate meteorological, topographic, and antecedent controls using Random Forest, XGBoost, SHAP interpretation, and exploratory SEM-based path analysis. Results SFFMC showed pronounced diurnal and seasonal variation. Time-of-day mean moisture was lowest at 10:00–16:00 (35.27 ± 0.56%) and highest at 04:00–06:00 (49.08 ± 0.47%). Seasonal mean SFFMC was highest in spring (62.0 ± 30.8%) and winter (52.8 ± 28.2%) and lowest in summer (39.2 ± 33.5%) and autumn (34.6 ± 17.6%). Shady-slope plots maintained higher SFFMC than sunny-slope plots (52.1 ± 31.9% vs. 41.6 ± 27.1%), and their daily moisture minima occurred 4–6 h later in autumn and winter. Random Forest and XGBoost showed strong predictive performance within monitored seasonal regimes, whereas temporally blocked validation indicated stronger weekly than monthly transferability. SHAP analysis showed seasonal shifts from precipitation-related wetting effects in spring and winter toward stronger temperature- and aspect-related drying in summer and autumn. SEM-based analysis identified a lag-indexed daily-precipitation feature as the strongest positive association with SFFMC and temperature as the strongest negative association, with the principal association structure retained in sensitivity analyses. Conclusions SFFMC dynamics in subtropical Masson pine forests reflect interacting hydrothermal, temporal, and terrain controls. The persistent moisture advantage of shady slopes and the earlier daytime drying of sunny slopes indicate that operational fire-danger assessment should consider both slope exposure and time of day. Combining real-time SFFMC monitoring with recent meteorological conditions can help fire managers identify daylight low-moisture windows, evaluate overnight moisture recovery, and improve terrain-sensitive warning and preparedness in subtropical forests.