4.1 The performance comparison of combination models against lysimeter measurements
Many studies have shown that the Penman family models are generally the most accurate when evaluating ET across various climate scenarios and regions (Liu et al., 2017b; Qiu et al., 2002). Of the penman models for ET0, the Penman–Monteith FAO 56 has been considered as the standard equation for estimating evapotranspiration (Allen et al., 1998). Contrary to what we expected, the Penman–Monteith FAO 56 was not the best in our study: here, we found that the combination models did not yield the best performance, and instead Pen-63 and FAO-24Pen were more accurate (Pen-63 and FAO-24 Pen had smaller RMSE and higherR 2 than Penman–Monteith FAO 56 during both the growing season and non-growing season). Such similar results have been reported in many other studies (Berengena and Gavilán, 2005; Howell, 1999). For instance, Howell et al. (1998) operated a weighing lysimeter in Texas and found that the earlier 1948 Penman equation performed better than the FAO-56 PM (Berengena and Gavilán, 2005). A more recent study also reported the poor performance of FAO-56 PM when compared with data from 20 FLUXNET towers (Ershadi, Mccabe, Evans, Chaney, & Wood, 2014). Combining these results suggests that FAO-56 PM might not be the only standard model for evaluating ET, because it did not yield better accuracy than the other combination models or radiation-based models. The poor performance of FAO-56 PM may associate with measuring error of lysimeter and the lysimeter was not set in a reference environment. It should be noted that reference evapotranspiration calculated the FAO-56 PM was defined as crop height of 0.12 m, albedo of 0.23 and surface resistance of 70 s/m (Allen et al., 1998), such reference environment was often not satisfied in our study site. Therefore, the FAO-56 PM applied without local lysimeter data comparison lead to a great error. There is increasing evidence indicated that the Penman locally adjusted equation showed better performance than FAO-56 PM when comparing hourly and daily ET estimates against weighing lysimeter measurements (Berengena & Gavilán, 2005; Liu et al., 2017). Of course, given the better performance of Pen-63 and FAO-24 than FAO-56 PM in this study, we may apply old Penman family models to our study region, especially considering the Penman–Monteith FAO 56 requires many meteorological inputs, which limit its use in areas with sparse data, especially in harsh environment (Hossein Tabari et al., 2012).