Evaluations of Uncertainty and Sensitivity in Soil Moisture Modeling on the Tibetan Plateau

Abstract

To provide scientific support for improvements in land surface modeling on the Tibetan Plateau (TP) by reducing uncertainties in the physical parameters of models, comprehensive uncertainty and sensitivity evaluations were performed for the simulation of surface soil moisture (SSM). Five observational stations were selected for the study. The conditional nonlinear optimal perturbation related to parameters (CNOP-P) approach and the Common Land Model (CoLM) with 28 uncertain parameters were employed to evaluate the maximal uncertainty of the simulated SSM. The uncertainty analysis indicated that the parameter errors could induce large uncertainties. These uncertainties in the SSM generally fluctuated over the range from 0.33 to 0.64 m3 m−3 in terms of absolute changes and from 235% to 510% in terms of percentage changes. The uncertainty analysis addressed the necessity of decreasing the uncertainties in the parameters. When resources are limited, the most important and sensitive parameter set should first be identified in order to reduce uncertainties. To find this parameter set, a sensitivity analysis framework based on the CNOP-P approach was applied. The results showed that the most sensitive and important combinations of 4 parameters changed slightly among the study sites and consisted of soil texture-related parameters. Although the most sensitive and important parameter combinations only had 4 elements, the uncertainties that they could induce accounted for a large proportion of the uncertainties caused by all 28 uncertain parameters. Furthermore, the decreases in parameter errors, which were derived from the CNOP-Ps of the most sensitive and important parameter combinations, led to the maximal reductions in the uncertainties of the simulated SSM. The above results imply that we should prioritize reducing the uncertainty of sensitive parameters or parameter combinations in order to improve prediction and simulation abilities.