As major components of terrestrial ecosystems, forest ecosystems play an important role in sequestering carbon and hence mitigating climate change. Canopy height is a crucial factor characterizing the structure and function of forest ecosystems, yet the driving mechanism of forest canopy height receives less attentions in China. Here, we utilize the satellite-based forest canopy height product with several environmental and climate factors (e.g. forest age, temperature, etc.) to delineate the spatial distributions of forest canopy height and its drivers in China at 1 km spatial resolution during the period of 2014 to 2018. The random forest is employed for identifying the dominant factors at province level, while Shapley additive explanations (SHAP) analysis is further incorporated at pixel-level to dig into the specific contributions of each driver. The results show that forest age primarily dominates the spatial distributions of forest canopy height across different forest ecosystems of China, followed by mean annual precipitation, soil type, and aspect. SHAP analysis further indicates that other factors, such as soil moisture and wind speed, also play critical roles to shape the spatial patterns of forest canopy height in China, which could not be revealed from province-level random forest analyses. Such results emphasize the priority of incorporating SHAP analysis with random forest to advance our understanding of forest canopy height distributions and benefit future projections. Our study highlights the necessity to characterize the spatial heterogeneity of forest canopy height, which is critical for accurate estimations of forest and even terrestrial carbon sink in China, facilitating the achievement of the goal of carbon peak in 2030 and carbon neutrality in 2060.

Aboveground biomass removal and canopy opening by selective logging modifies soil moisture in the main root zone, impacting soil aeration and various biogeochemical processes in tropical production forests. This study investigated the relationship between canopy damage and topsoil (10 cm) moisture in two logged forests in Malaysian Borneo, while simultaneously controlling for logging intensity, time elapsed since historical logging, and spatial autocorrelation. Volumetric soil water content (VSWC), canopy height model (CHM), leaf area index (LAI), and historical logging data were collected from 84 transects placed subjectively in 15 sites exhibiting varying canopies. We generated an index (PC1) quantifying the magnitude of canopy structural degradation from canopy structure metrics (CSM) combining CHM and LAI data within a 20-meter buffer for each transect. PC1 was analyzed for its impact on VSWC across logging periods, and contrasted with topography. Spatial autocorrelation of VSWC was examined regarding to canopy conditions. VSWC was significantly higher in all logged forests (over 0.4 m(3) m(-3)) comparing to non-disturbed forests (0.27 m(3) m(-3)). The immediate wetting could be a result of extracting mature individuals of late-successional species holding large biomass, while the persistent wet condition may be due to retarded canopy and biomass recovery. In the study area, canopy structure was a stronger predictor of soil moisture than topography. The high soil moisture underneath the most degraded canopies presented the largest spatial extent of autocorrelation. This study revealed soil wetting after selective logging in humid tropical forests, driven by reduced transpiration from biomass loss rather than increased evaporative demand resulting from canopy opening. The elevation in soil moisture could have disrupted biogeochemical processes in the below-ground system, which in turn impede forest succession and put stress on the overall vulnerability of disturbed tropical rainforests.