Modeling actual evapotranspiration with routine meteorological variables in the data-scarce region of the Tibetan Plateau: Comparisons and implications

Authors

Ning Ma, Chinese Academy of Sciences, Beijing, China
Yinsheng Zhang, Chinese Academy of Sciences, Beijing, China
Chong-Yu Xu, Chinese Academy of Sciences, Beijing, China
Jozsef Szilagyi, Budapest University of Technology and EconomicsFollow

Document Type

Article

Date of this Version

2015

Citation

Ma, N., Y. Zhang, C.-Y. Xu, and J. Szilagyi (2015), Modeling actual evapotranspiration with routine meteorological variables in the data-scarce region of the Tibetan Plateau: Comparisons and implications, J. Geophys. Res. Biogeosci., 120, 1638–1657

Comments

©2015. American Geophysical Union. All Rights Reserved.

Abstract

Quantitative estimation of actual evapotranspiration (ET_a) by in situ measurements and mathematical modeling is a fundamental task for physical understanding of ETa as well as the feedback mechanisms between land and the ambient atmosphere. However, the ETa information in the Tibetan Plateau (TP) has been greatly impeded by the extremely sparse ground observation network in the region. Approaches for estimating ET_a solely from routine meteorological variables are therefore important for investigating spatiotemporal variations of ETa in the data-scarce region of the TP. Motivated by this need, the complementary relationship (CR) and Penman-Monteith approaches were evaluated against in situ measurements of ET_a on a daily basis in an alpine steppe region of the TP. The former includes the Nonlinear Complementary Relationship (Nonlinear-CR) as well as the Complementary Relationship Areal Evapotranspiration (CRAE) models, while the latter involves the Katerji-Perrier and the Todorovic models. Results indicate that the Nonlinear-CR, CRAE, and Katerji-Perrier models are all capable of efficiently simulating daily ETa, provided their parameter values were appropriately calibrated. The Katerji-Perrier model performed best since its site-specific parameters take the soil water status into account. The Nonlinear-CR model also performed well with the advantage of not requiring the user to choose between a symmetric and asymmetric CR. The CRAE model, even with a relatively low Nash-Sutcliffe efficiency (NSE) value, is also an acceptable approach in this data-scarce region as it does not need information ofwind speed and ground surface conditions. In contrast, application of the Todorovic model was found to be inappropriate in the dry regions of the TP due to its significant overestimation of ET_a as it neglects the effect of water stress on the bulk surface resistance. Sensitivity analysis of the parameter values demonstrated the relative importance of each parameter in the corresponding model. Overall, the Nonlinear-CR model is recommended in the absence of measured ET_a for local calibration of the model parameter values.