Evaluation of CERES-Maize model for simulating maize phenology, grain yield, soil-water, evapotranspiration, and water productivity under different nitrogen levels and rainfed, limited, and full irrigation conditions

The CERES-Maize model performance was investigated in simulating maize phenology, grain yield, soil-water, evapotranspiration, and water productivity under different irrigation and nitrogen (N) levels under a variable rate lateral (linear)-move sprinkler irrigation system. The irrigation levels were rainfed, full irrigation treatment (FIT) and 75% FIT. The N levels were 0, 84, 140, 196, and 252 kg/ha. The field experiment was conducted in the form of split plots with the irrigation levels as the main treatment and N levels as a sub-main treatment. The root mean squared error (RMSE), normalized RMSE (RMSEn), R2, T test, and model prediction error (Pe) statistics were used to evaluate the performance and accuracy of the model. Calibration of the model was done using the data of 2011 and 2012 and validation of the model was conducted for 2013 and 2014 by considering days after planting to flowering (DAPF), days after planting to maturity (DAPM), grain yield, crop evapotranspiration (ETc), water productivity (WP), and soil water content (SWC). The DAPF simulations based on the average values of Pe (0%), RMSE (2 days), and RMSEn (3%) and the DAPM simulation results based on the average values of Pe (2%), RMSE (4 days), and RMSEn (3%) showed that the model had an acceptable accuracy. In calibration years, RMSE, RMSEn, and R2, respectively, were 0.57 ton/ha, 5%, and 0.91; and in validation years, the same statistics, respectively, were 0.86 ton/ha, 10% and 0.94, indicating good performance of the model in estimating the grain yield. Good accuracy was observed in the estimation of ETc and WP. In most cases, the model accuracy was greatest for 75% FIT and FIT treatments than the stressed conditions in the rainfed treatment. The model accuracy can be enhanced by improving the model coefficients in response to low levels of water and N supply. R2 values obtained in rainfed (0.83), 75% FIT (0.81) and FIT (0.67) treatments in calibration years and R2 values in rainfed (0.75), 75% FIT (0.77) and FIT (0.86) treatments in validation years showed that the model predicted the SWC relatively well. The comparison of ETc values with respect to N levels showed that there was no considerable difference between levels of N applications impact(s) on the ETc magnitude in the rainfed treatment. Comparison of different levels of N in rainfed and FIT showed that the application of 252 kg/ha of N resulted in 2.37 kg/m3 and 2.56 kg/m3 of WP, respectively, which was significantly different from other levels of N fertilizer applications. In general, CERES-Maize model can be a useful tool for predicting plant phenology, grain yield, ETc, WP, and SWC for the conditions similar to those presented in this research. The CERES-Maize model can provide valuable data and information for sustainable maize production by examining the long-term grain yield and WP, which can be beneficial to growers, advisors, and stakeholders to enhance the maize production efficiency by accounting for irrigation and N management strategies.