Calibration and validation of CERES-wheat model for wheat in middle Gujarat region

Abstract

CERES (Crop Environment Resource Synthesis)-wheat model (DSSAT v 3.5-Decision Support System for Agrotechnology Transfer) was calibrated and validated for wheat cv. GW-496 at Anand using experimental data collected under different management practices (Date of sowing x Irrigation) during 1995-2007. Results showed that optimum sowing date (D2–15th Nov.) validation was found better as compared to early (D1–1st Nov.) and late (D3–30th Nov.) sowings. The validation of model for different irrigation regimes showed that the performance of model was poor in treatment having less irrigation (I1 and I2). The model performance was found good and satisfactory in treatments having 6-7 irrigations (I3 and I4). On an average, the performance of model for I4 treatment was found good. This showed that model worked better under optimum sowing with optimum irrigation. The various test criteria for evaluation of model showed that highest correlation was observed in D2I3 treatment. The lowest MAE was observed in D3I3 treatment. Similarly, lowest MBE (12.77), lowest RMSE (46.04) and highest index of agreement (1.0) were observed in D1I4 treatment. The error per cent by CERES-wheat model showed that in majority of the cases the models had underestimated wheat yield. Per cent error ranged between -0.020 to -56.02. The average per cent error was found lowest in D1I4, D2I4 and D3I3 irrigation treatments as compared to other treatments. This showed that the model worked good in all model test criteria. In a nutshell, the validation results showed that the model worked better under optimum sowing with optimum irrigation as compared to early/late sowing and moisture stress conditions.