Determinants of Physical Processes and Their Contributions for Uncertainties in Simulated Evapotranspiration Over the Tibetan Plateau

Abstract

AbstractEstimation of the Evapotranspiration (ET) over the Tibetan Plateau (TP) still keeps uncertain due to inaccurate physical parameters and processes in numerical models. In this study, the simulated ET is evaluated for 13 sites over the TP during the period of 2001–2017 using the conditional nonlinear optimal perturbation related to parameters (CNOP‐P) approach. To find key physical parameters and processes, the sensitivity analysis (SA) method based on the CNOP‐P is employed. The traditional SA method (one‐at‐a‐time, OAT) is also employed to compare to the SA method based on the CNOP‐P for the identification of the key parameters and parameters subset. The numerical results show that the average estimation of the ET for 13 sites could be improved by 37.7 mm year−1 and 62.5 mm year−1 for two data sets using the CNOP‐P‐type parameters perturbations, respectively. Further, the sensitive physical parameters subset includes more vegetation parameters using the SA method based on the CNOP‐P method, which are different to those using the OAT method. This implies that the vegetation transpiration plays a vital role for simulation of the ET over the TP. It is found that the improvement degree (24.6%) of simulation ability of the ET over the TP by reducing more sensitive vegetation parameters identified by the CNOP‐P sensitivity analysis method is larger than that employing the OAT method (21.9%).