90-m Resolution Mapping of Black Soil Organic Carbon in Heilongjiang: Integrating Meta-Analysis with XGBoost

Soil organic carbon (SOC), a component that is essential to the global carbon cycle, is highly concentrated in Heilongjiang Province's black soil. To reduce the greenhouse effect and quickly achieve the "Dual carbon" goal, it is crucial to intensify research efforts regarding the storage and spatial distribution of SOC and implement appropriate carbon sequestration techniques. Through meta-analysis of published domestic and international literature from 2005 to 2023, a database of SOC samples in Heilongjiang Province was created for this study. For the purpose of conducting the correlation analysis, a total of 175 soil samples and 13 environmental factors were used to construct the database. The extreme gradient boosting method (XGBoost), support vector machine (SVM), and random forest (RF) were adopted as the prediction models to depict the spatial distribution pattern of SOC in the 0–20 centimeter surface layer of the soil in Heilongjiang Province at a resolution of 90 meters.Model performance was validated via ten-fold cross-validation (MAE = 1.35 kg/m², RMSE = 1.80 kg/m², R²=0.82, E _var =0.82). XGBoost outperformed other models in capturing nonlinear SOC-environment relationships. Climate, topography, and Soil type variable were key drivers of SOC spatial differentiation. Total SOC storage was calculated as 4.04 Gt using ArcGIS. The high-resolution distribution map provides critical insights for targeted carbon sequestration strategies in black soil regions.