Comparison of three evapotranspiration models in a rain-fed spring maize filed in the Loess Plateau, China

Accurate estimates of the rain-fed field evapotranspiration (ET) in the Loess Plateau of Northwest China is critical for predicting hydrologic processes, crop yield formation, and climate change. We simulated the ET variation in a rain-fed spring maize field in the eastern Loess Plateau using the Penman-Monteith (PM), Shuttleworth-Wallace (SW), and modified Priestley-Taylor (PTm) models. Then we compared their results with the observed ET using the eddy covariance (EC) method. Generally, the diurnal variation in estimated ET from the three models was similar to that of observed ET by the EC method. However, the PM model significantly underestimated ET. The ET estimates obtained from the SW and PTm models were approximately equal to the observed ET by the EC method. Considering the PTm model’s simplicity, we finally recommend it for rain-fed spring maize fields in the Loess Plateau. After precipitation, the estimated ET from the three models was significantly smaller than measured ET by the EC method, especially the PM model. After a frost, all three models failed to reflect the dramatic decrease in crop transpiration, and thus significantly overestimated ET.