Spatio‐Temporal Analysis of Key Physical Processes Driving Uncertainty in Net Ecosystem Productivity Prediction Over the Tibetan Plateau

Net ecosystem productivity (NEP) on the Tibetan plateau (TP) owns complex physical processes influenced by climate, vegetation, soil, and human activities. The study of NEP on the TP is important for understanding the carbon cycle in the region, predicting the impacts of climate change on the ecosystem, and formulating ecological conservation policies. In this study, the most sensitive parameter combinations were determined to cause the maximal uncertainties in estimated NEP on the TP using a sensitivity analysis method based on the conditional nonlinear optimal parameter perturbation (CNOP‐P) at the spatial and temporal scales. The numerical results showed that the most sensitive parameter combinations using the CNOP‐P method, which included vegetation and hydrology parameters, were more important than the single‐parameter sensitivity of the top‐ranked parameter using a traditional method to cause the uncertainties in estimated NEP on the TP. Sensitive parameter combinations exhibited spatio‐temporal features in predicting uncertainties in NEP. Moreover, the improvements in estimated NEP through reducing the errors of the most sensitive parameter combinations identified by the CNOP‐P method were better than those by the traditional method. This study demonstrated how identifying relatively important and sensitive combinations of physical parameters was essential to improve the simulation and prediction of the NEP on the TP.