Performance of AquaCrop model in simulating maize growth, yield, and evapotranspiration under rainfed, limited and full irrigation

The performance assessments of AquaCrop model using long-term field-measured data are rare. In this study, the model was evaluated relative to maize growth, yield, and water use parameters/variables under different water stress conditions over six years (2005-2010) in Nebraska, USA. The model was calibrated and validated for full irrigation treatment (FIT), limited irrigation (50, 60, and 75% of FIT), and rainfed. Model default parameters provided very poor estimates for mostly all variables. After calibration, the model adequately simulated daily canopy cover (CC) for both 2009 and 2010 (NRMSEs <= 15.6%), except for slight discrepancies in 2010. However, the model overestimated the final biomass in 2009 due to overestimation of biomass development in the late growth stages. The simulation of final biomass was better in 2010 (NRMSE = 5.3%) than in 2009 (NRMSE = 31%). The model simulated grain yield quite well (NRMSE of 7.7 and 12.1%, EF of 0.8 and 0.7) during both calibration and validation, respectively, except for rainfed treatments in most years. The model was not able to simulate total soil-water accurately in most cases, except for 2009 growing season. The crop evapotranspiration (ETc) was simulated with good accuracy during 2007, 2009, and 2010 and the higher prediction error (up to 16.5%) was observed for dry years (2005 and 2006) and wet year (2008). The model performance declined substantially in conditions of water stress, excess water, and high evaporative demand. In conclusion, the AquaCrop simulated yield and ETc (with slight underestimation) quite well in some cases, but encountered substantial difficulties in simulating biomass and soil-water, especially during years of low precipitation and high evaporative demand as well as in wet year. Further evaluation of the model is needed using field measured evaporation and transpiration data to determine the accuracy of ETc partition by the model to better simulate soil-water and ETc, which are critical for the estimation of in-season irrigation requirements, biomass production, and yield.