Assessing the performance of two models on calculating maize actual evapotranspiration in a semi-humid and drought-prone region of China

The two-step and one-step models for calculating evapotranspiration of maize were evaluated in a semi-humid and drought-prone region of northern China. Data were collected in the summers of 2013 and 2014 to determine relative model accuracy in calculating maize evaopotranspiration. The two-step model predicted daily evaoptranspiration with crop coefficients proposed by FAO and crop coefficient calibrated by local field data; the one-step model predicted daily evapotranspiration with coefficients derived by other researcher and coefficients calibrated by local field data. The predicted daily evapotranspiration in 2013 and 2014 growing seasons with the above two different models was both compared with the observed evapotranspiration with eddy covariance method. Furthermore, evapotranspiration in different growth stages of 2013 and 2014 maize growing seasons was predicted using the models with the local calibrated coefficients. The results indicated that calibration of models was necessary before using them to predict daily evapotranspiration. The model with the calibrated coefficients performed better with higher coefficient of determination and index of agreement and lower mean absolute error and root mean square error than before. And the two-step model better predicted daily evapotranspiration than the one-step model in our experimental field. Nevertheless, as to prediction ET of different growth stages, there still had some uncertainty when predicting evapotranspiration in different year. So the comparisons suggested that model prediction of crop evapotranspiration was practical, but requires calibration and validation with more data. Thus, considerable improvement is needed for these two models to be practical in predicting evapotranspiration for maize and other crops, more field data need to be measured, and an in-depth study still needs to be continued.