Learning ensembles of process-based models for high accurately evaluating the one-hundred-year carbon sink potential of China’s forest ecosystem

China’s forests play a vital role in the global carbon cycle through the absorption of atmospheric CO2 to mitigate climate change caused by the increase of anthropogenic CO2. It is essential to evaluate the carbon sink potential (CSP) of China’s forest ecosystem. Combining NDVI, field-investigated, and vegetation and soil carbon density data modeled by process-based models, we developed the state-of-the-art learning ensembles model of process-based models (the multi-model random forest ensemble (MMRFE) model) to evaluate the carbon stocks of China’s forest ecosystem in historical (1982–2021) and future (2022–2081, without NDVI-driven data) periods. Meanwhile, we proposed a new carbon sink index (CSindex) to scientifically and accurately evaluate carbon sink status and identify carbon sink intensity zones, reducing the probability of random misjudgments as a carbon sink. The new MMRFE models showed good simulation results in simulating forest vegetation and soil carbon density in China (significant positive correlation with the observed values, r = 0.94, P