Wednesday, November 4, 2009
Convention Center, Exhibit Hall BC, Second Floor
Crop growth simulation models are usually calibrated and validated for a few variables, e.g. yield and it components. The use of variables related to nitrogen dynamics to calibrate and validate CERES-Maize is rarely reported. The objectives of this study were to i) calibrate CERES-Maize crop growth model for variables related to plant nitrogen dynamics, and ii) compare the predictions with observed, independent data from field experiments. Two years (2003 and 2004) of maize fertilization experiments were used to calibrate and two years (2006 and 2007) of similar experiment to validate the model. The experiments included four nitrogen rates at sowing (0, 70, 140, 210 kg N ha-1) and in-season applications at different growth stages, and conducted at Parana, Argentina (31.5ºS 60.3ºW). Model parameters as mineralization factor (SLNF), nitrogen content in above ground biomass at emergence (TANCE) and radiation use efficiency (RUE) were optimized during calibration process. Model performance was assessed through regression analysis, and the root mean square error (RMSE) using IRENE software. The model predicts adequately the studied variables. The simulation of grain yield closely matched the observed data (r 2 = 0.96, RMSE = 1011 kg ha-1), whereas biomass was simulated with less accuracy (r2 = 0.64, RMSE = 4541 kg ha-1). Simulated grain size (r2 = 0.94, RMSE = 0.016 mg grain-1) was more accurate predicted than grain number (r2 = 0.74, RMSE = 567 grain m-2). Tops nitrogen at maturity was adequately simulated (r2 = 0.76, RMSE = 56 kg ha-1), but less accurately than grain nitrogen at maturity (r2 = 0.86, RMSE = 51 kg ha-1). CERES-Maize showed an adequate prediction of crop yield and its components adding to an acceptable simulation of variables related to nitrogen plant dynamics.