Dataset for "A persistent soil carbon sink induced by land use and cover change in China".

We created an extensive SOC dataset through pairwise field sampling (with vs. without LUCC), integrating a large-scale field survey with a thorough literature synthesis. Leveraging this dataset, we developed an integrated extreme gradient boosting (XGBoost) and Shapley additive explanation (SHAP) approach to gain data-driven insights into the factors regulating the spatiotemporal variations of ΔSOCD. By coupling pixel-level predictions across China with reconstructed annual land conversion trajectories derived from high-resolution land-use datasets, we quantified the national-scale SOC budget induced by historical LUCC and evaluate its contribution to China's terrestrial carbon cycle.