Comparison of three modified models in evapotranspiration and its components over a rainfed spring maize cropland on the Loess Plateau, China

Abstract

Accurately estimated evapotranspiration (ET) and its components in the growing season of cropland are required to optimize agricultural practices and improve water use efficiency. In this study, we developed a modified Shuttleworth–Wallace (SWm) model, a modified Priestley-Taylor (PTm) model, and a modified dual crop coefficient (DKm) model based on the characteristics of a rainfed spring maize cropland on the Loess Plateau. The estimated ET and its components by the three models were compared with the measured ET and its components using an eddy covariance system and microlysimeters. The results indicated that the three models performed well in estimating the 30 min ET and daily ET and its components during the growing seasons of the rainfed spring maize field in 2015 and 2016. The diurnal course of the measured 30 min ET was closest to that of the estimated 30 min ET from the DKm model, and this model performed best to estimate 30 min ET with a regression coefficient (b0) of 1.01 (R2 = 0.89) in 2015 and 0.97 (R2 = 0.88) in 2016. According to statistical analysis, the PTm model outperformed the other two modified models on estimating daily ET with b0 of 0.97 (R2 = 0.89) and 0.94 (R2 = 0.89) and daily plant transpiration (Tp) with b0 of 1.07 (R2 = 0.93) and 1.11 (R2 = 0.93) during the both growing seasons. The performance of the three modified models on daily soil evaporation (Es) was not as good as that of daily ET and Tp, and the DKm model performed best to reproduce daily Es with a b0 of 0.91 (R2 = 0.37) and 0.88 (R2 = 0.54) in 2015 and 2016, respectively. As the PTm model required the fewest input variables among the three modified models, this model was finally recommended for rainfed spring maize on the Loess Plateau, China.