Search for a command to run...
Broad-scale assessments often track forest productivity, yet they rarely quantify how soil conditions determine whether these gains persist as long-lived carbon and generate measurable economic value. This study focused on Northeast China, where forests include boreal coniferous stands dominated by Dahurian larch, temperate conifer–broadleaf mixed forests with Korean pine, and temperate deciduous broadleaf forests dominated by Mongolian oak. We combined GLASS net primary productivity and ESA CCI Land Cover to delineate forest pixels, used 2000 to 2005 as the baseline, and converted productivity anomalies into pixel level carbon economic value using a consistent pricing rule. Forest NPP increased significantly during 2000 to 2018 (slope = 1.57, p = 0.019), and carbon economic value also increased over time during 2006 to 2018 (slope = 2.24, p = 0.002), with the highest values in core mountain forests and lower values in the western forest–grassland transition zone. Correlation analysis, explainable random forests, and variance partitioning characterized spatial and temporal dynamics from 2000 to 2018 and identified environmental controls. Carbon value increased over time and showed marked spatial heterogeneity that mirrored productivity patterns in core mountain forests. Climate was the dominant predictor of value, while higher soil pH and clay content were negatively associated with value. The random forest model explained about 70% of the variance in carbon value (R2 = 0.695), and variance partitioning indicated substantial unique and joint contributions from climate and soil alongside secondary topographic effects. The automatable framework enables periodic updates with new satellite composites, supports ecological compensation zoning, and informs soil-oriented interventions that enhance the monetized value of forest carbon sinks in data-limited regions.