We investigated the vulnerability of Chinese forests to fires during the period 2001-2020 by integrating automatic machine learning with fire disturbance data and satellite products. The proposed methodology is purely data-driven and therefore reproducible, applicable at large scales, and in line with the measurement/reporting/verification process of UNFCCC. Part of pre-processed data, codes of vulnerability models developed in the afore-mentioned work are made publicly available here and briefly described to facilitate reproducibility and applicability.