Compound-Drought Vegetation Carbon Loss: Supporting Data and Analysis Code (2001–2023)

This repository provides the supporting research data, figure source datasets, and analysis scripts associated with the manuscript “Compound-drought Vegetation Carbon Loss: Asymmetric Stress Contributions and Limited Non-additivity” by Yangyang Li and Shuanggen Jin, submitted to Earth's Future. The study investigates the spatiotemporal characteristics of atmospheric drought (AD), soil drought (SD), and compound drought (CD), together with their impacts on vegetation gross primary productivity (GPP) and carbon uptake. The primary analysis covers 2001–2023, with additional meteorological and soil moisture analyses covering 1981–2023. The study uses a quasi-global land domain from 60°S to 90°N and a common 0.1° spatial analysis grid. The repository includes: Source datasets and statistical results supporting Figures 1–10 and Supplementary Figures S1–S2. Drought occurrence, duration, frequency, intensity, and exposure statistics. Vegetation GPP responses, carbon-loss estimates, and post-drought response analyses. Model evaluation results, feature importance, and SHAP-based analyses from spatially validated XGBoost models. Counterfactual estimates of VPD-only, RZSM-only, joint, and non-additive contributions to compound-drought-related carbon loss. Sensitivity analyses, figure-generation scripts, and supporting documentation. The repository is organized by figure and analytical workflow. A README, data dictionary, and file-integrity checksums are provided to facilitate interpretation and reuse. Original third-party datasets, including ERA5-Land, MCI-GPP, GLASS LAI, MODIS land cover, and other environmental products, are not comprehensively redistributed. Users should obtain the necessary original inputs from their respective data providers. Full end-to-end execution of some scripts requires external datasets, software dependencies, and adaptation of the original file paths. This repository is intended to support transparent inspection, independent evaluation, and reproducibility of the results presented in the associated manuscript.