Affiliations:
1 Qinghai Provincial Key Laboratory of Restoration Ecology for Cold Region, Northwest Institute of Plateau Biology, Chinese Academy of Sciences, Xining, Qinghai 810001, China
2College of Life Sciences, Luoyang Normal University, Luoyang, Henan 471934, China
3Key Laboratory of Adaptation and Evolution of Plateau Biota, Northwest Institute of Plateau Biology, Chinese Academy of Sciences, Xining, Qinghai 810001, China
4 University of Chinese Academy of Science, Beijing 100049, China
* Corresponding author: Xiaowei Guo and Guangmin Cao
E-mail:xwguo1206@163.com for Xiaowei Guo andcaogm@nwipb.cas.cnfor Guangmin Cao
Author contributions: L Dai and X Guo performed the research, analyzed data, and wrote the paper; Y Lan, F Zhang, X Ke, Y Li and Y Du analyzed data; G Cao conceived the study.
Conflict of interest: The authors declare no conflict of interest
Abstract: Accurate estimates of evapotranspiration (ET) are of great importance for water balance and energy exchange processes, as ET constitutes the key component of the terrestrial water cycle. Although many applicable reference evapotranspiration (ET0) models have been developed to estimate the ET, these are largely focused on low altitude regions, with little attention to alpine meadow. In this paper, we evaluate the performance of 13 ET0 models by comparison with large weigh lysimeter measurements. Specifically, we use three combination models, seven radiation-based models and three temperature-based models driven with data from 8 June 2017 to 18 September 2018 in a humid alpine meadow, northeastern Qinghai-Tibetan Plateau. The daily ET was also obtained by large weighing lysimeters located in an alpine Kobresia meadow. Results show that the performances of the 13 ET0 models, ranked on the basis of their RMSE (root mean square error), decreased in the order: DeBruin-Keijman>Priestley-Taylor> 1963 Penman> FAO-24 Penman>Hargreaves>Hargreaves2>Hargreaves1>IRMAK1>FAO-56Penman-Monteith>Makkink>Makkink (1967)>Makkink (1957)>IRMAK2. Overall, the radiation-based models performed best, with RMSEs ranging from1.03 to 1.47 mm d−1 and averaging 1.09 mm d−1, followed by the combination models (RMSE from 1.19 to 1.36 mm d−1 and averaging 1.26 mm d−1) and temperature models (RMSE from 1.28 to 1.32 mm d−1 and averaging 1.29 mm d−1). The best radiation-based model (DeBruin-Keijman) was more accurate than the best combination model (1963 Penman) and temperature model (Hargreaves) by 16.67% and 25.49%, respectively. The better performance of the radiation-based models over the other two types may be attributed to their inclusion of the dominant factors affecting ET, such as net radiation (Rn). All models tended to underestimate measured ET during periods of larger evaporative demand (i.e. growing season) and overestimate measured ET during lower evaporative demand (i.e. non-growing season). Our results could help in the selection of a suitable ET model for alpine ecosystems, thereby benefitting water irrigation management.
Keywords: reference evapotranspiration; alpine meadow; lysimeter measurement; combination models; radiation-based models; temperature-based models
  1. Introduction
Evapotranspiration (ET) is one of the key parameters in the simultaneous processes of heat and water transfer to the atmosphere via transpiration and evaporation in the soil–plant– atmosphere system (Sentelhas, Gillespie, & Santos, 2010), thereby playing an important role in water balance calculations, water allocation and water irrigation management. Thus, accurate estimates of evapotranspiration could improve water management strategies and promote the efficient use of water resources, especially in regions suffering water shortages (Ljungqvist et al., 2016; Sun et al., 2011).
To-date, direct measurements of ET have been achieved by a variety of methods such as the Bowen Ratio Energy Balance System (A. Irmak & Irmak, 2008; S Irmak, Allen, & Whitty, 2003; S. Irmak, Howell, Allen, Payero, & Martin, 2005; Si et al., 2005), lysimeters (Jia, Dukes, Jacobs, & Irmak, 2006; Liu et al., 2017) and the eddy covariance technique (Gu et al., 2008; Novick et al., 2009). Alternatively, ET can be indirectly assessed by applying various reference evapotranspiration equations. Several models of reference evapotranspiration have become widely used for the calculation of ET, and can be classified into three types: radiation-based models (Doorenbos, 1975; Hargreaves and Samani, 1985), temperature-based models (Trajkovic, 2005; Valipour & Eslamian, 2014), and combination models (Monteith, 1965; Penman, 1963).While the development of these models has undoubtedly benefited the calculation of ET, it still difficult to choose the optimal one due to the availability of the observed data the together with large number of models also adds confusion and arbitrariness to model selection, because most models have not been evaluated against lysimeter measurements across a range of regions and climates. To select the best performing models, many studies have been conducted to assess model performance under various climates. For instance, the Food and Agriculture Organization of the United Nations (FAO) recommend the Penman–MonteithFAO-56 combination equation (PM-56) as the standard equation for estimating reference evapotranspiration (Allen, Pereira, Raes, & Smith, 1998), and this has been widely used worldwide when compared with other equations (Cai, Liu, Lei, Pereira, & Luis, 2007). The advantages of the Penman–Monteith equationare that it is does not require any local calibration because it incorporates both physiological and aerodynamic parameters, and it has been well tested by a variety of lysimeters (Trajkovic, 2009). Although many models have been widely used to estimate ET, it should be noted that most previous models have only been evaluated with reference to FAO-56 PM (Cao, Yu-Zhong, Liu, Zhong, & Zhao, 2015; Martı́Nez-Cob & Tejero-Juste, 2004), with few being tested against lysimeter measurements (Jensen, Burman, & Allen, 1990; Ventura, Spano, Duce, & Snyder, 1999; Yoder, Odhiambo, & Wright, 2005). Furthermore, the application of the PM-56 equation needs many meteorological inputs, such as wind speed, temperature, humidity and solar radiation, that are often not available in regions with harsh environments (Hossein Tabari, Kisi, Ezani, & Talaee, 2012), especially in developing countries where have limited number of meteorological stations. Thus, it is essential to develop an relatively accurate reference evapotranspiration equation that requires fewer meteorological parameters, to allow more simplified estimates of ET than those of PM-56, applicable across a range of climatic conditions (Hossein Tabari, 2010; H. Tabari & Talaee, 2011). So far, many models have been developed: for example, Tabari (2010) assessed four reference evapotranspiration models in an arid climate, and found that the Turc model performed the best. Meanwhile, the Hargreaves equation performed best in semiarid regions (Sabziparvar, Tabari, Aeini, & Ghafouri, 2010; Hossein Tabari, 2010). Liu et al. (2017) compared 16 models for reference evapotranspiration against weighing lysimeter measurements, and found that the combination models performed best for estimating ET in semiarid regions. Overall, most previous studies have been conducted in low-humidity conditions at low altitude (i.e. arid and semiarid regions) (Liu et al., 2017; Sentelhas et al., 2010), with few studies in humid climates, particularly in alpine ecosystems.
The Qinghai-Tibetan Plateau (QTP),with an average altitude of 4000 m, is the world’s highest alpine ecosystem and is also known as the ”Asian tower”, playing an important role in ensuring the safety of water resources in China and southeast Asia (Dai, Guo, Zhang, et al., 2019; Zou et al., 2017). The alpine meadow and alpine grassland across the QTP account for almost 60% of the plateau area (Dai, Ke, et al., 2019)); therefore, accurate assessment of ET in an alpine ecosystem are not only provides new insights into the water cycle, but also benefit the formulation of water resource management strategies. Furthermore, given the uncertainty and confusion in the selection of ET equations across different regions and climates, it is critical to thoroughly understand the performance of the various models in a humid alpine meadow (Liu et al., 2017). In this study, we compared 13 reference evapotranspiration models against weighing lysimeter measurements on the northern Tibetan Plateau, with the aims of selecting the best fit model in applications in this region to estimate the ET.
2 Materials and methods