Simulation of crop production and nitrogen uptake of rice under nitrogen management using the CERES-Rice model

Abstract

Abstract To investigate CERES-Rice model, a field experiment was conducted from 2012 to 2015 at the Rice Research Institute - Deputy of Mazandaran, Iran (Amol) to evaluate the CERES-Rice model. The experiment was designed and implemented as a factorial design based on a randomized completely block design with three replications. In the model calibration, nitrogen amounts were evaluated at four rates (40, 80, 120, and 160 kg ha− 1), and nitrogen splitting was evaluated at four levels. For model validation in 2014 and 2015, nitrogen rates were evaluated at three levels (60, 120, and 180 kg ha− 1), and nitrogen splitting was evaluated at four levels. Plant traits such as LAImax, grain yield, biomass, grain N content, and total plant N content were evaluated for model calibration and validation. The results showed that LAImax simulations performed well under calibration and validation conditions (R2 = 0.92 − 0.63, RMSEn = 19.2–23.8%, RMSE = 0.92 − 0.7), and a relatively acceptable level of accuracy was achieved for biomass in calibration and validation conditions (R2 = 0.9 − 0.76, RMSEn = 6.2–12.8%, RMSE = 831–1499 kg ha− 1). In calibration and validation, R2 values (0.9 − 0.83), RMSE (464 − 225 kg/ha), and RMSEn (7-3.4%) were obtained in simulations of grain yield. The model was evaluated based on R2 (0.55–0.83), RMSE (7.5–13.4 kg ha− 1), and RMSEn (6.2–14.3%) to determine its optimal accuracy in estimating the total nitrogen content of grains and plants. The results indicate that the CERES-Rice model was reliable enough for simulating grain yield and plant nitrogen content.