Item

Abstract

Uncertainties of physical parameters are important sources of uncertainties in numerical simulations and predictions. It is a key issue to identifying important and sensitive physical parameters for uncertainty reduction in numerical models. This study proposes a novel approach: conditional nonlinear optimal perturbations sensitivity analysis (CNOPSA). The CNOPSA method fully considers the nonlinear synergistic effects of parameters in the whole parameter space and quantitatively estimates the maximum effects of parameter uncertainties, prone to extreme events. Numerical results of the theoretical five-variable grassland ecosystem model demonstrate that the CNOPSA method can effectively identify the sensitive and important physical parameters and parameter combinations. Based on the Community Land Model (CLM5.0), the key physical processes and physical parameters of the net primary productivity (NPP) and soil organic carbon (SOC) simulation uncertainties in carbon-nitrogen-water cycle over the Tibetan Plateau are explored by using the CNOPSA method. Some carbon, nitrogen, and hydrological parameters are relatively sensitive and important to the simulation uncertainties in carbon cycle. However, the variance-based approach, based on the possibility of a limited parameter samples from a statistical point of view, only recognizes the importance of the carbon and nitrogen parameters. Additionally, the improvement of carbon cycle simulation abilities caused by eliminating the error of sensitive parameter combination identified by the CNOPSA method is also higher than the result of parameter combination identified by variance-based method. This study suggests that the CNOPSA method is effective and feasible for improving the simulation abilities in land surface models.