Evaluation of CERES-Wheat and CropSyst models for water–nitrogen interactions in wheat crop

Abstract

Crop simulation models can provide an alternative, less time-consuming and inexpensive means of determining the optimum crop N and irrigation requirements under varied soil and climatic conditions. In this context, two dynamic mechanistic models (CERES (Crop Environment REsource Synthesis)-Wheat and CropSyst (Cropping Systems Simulation Model)) were validated for predicting growth and yield of wheat ( Triticum aestivum L) under different nitrogen and water management conditions. Their potential as N and water management tool was evaluated for New Delhi representing semi-arid irrigated ecosystems in the Indo-Gangetic Plains. The field experiment was carried out on a silty clay loam soil at the Research Farm of the Indian Agricultural Research Institute, New Delhi, India during 2000–2001 to collect the input data for the calibration and validation of both the models on wheat crop (variety HD 2687). The models were evaluated for three water regimes [I4 (4 irrigations within the growing season), I3 (3 irrigations within the growing season) and I2 (2 irrigations within the growing season)] and five N treatments (N 0, N 60, N 90, N 120 and N 150). Both the models were calibrated using data obtained from the treatments receiving maximum nitrogen and irrigations, i.e., N 150 and I4 treatments. The models were then validated against other water and nitrogen treatments. For performance evaluation, in addition to coefficient of determination ( R 2), root mean square error (RMSE), mean absolute error (MAE) and Wilmot's index of agreement (IoA) were estimated. Both CERES-Wheat and CropSyst provided very satisfactory estimates for the emergence, flowering and physiological maturity dates. For CERES-Wheat overall prediction (pooled result of the three water regimes) of grain yield was satisfactory with significant R 2 values (0.88). The model, however, under estimated the biomass under all water regimes and N levels except for N 0 level, under which biomass was overpredicted. CropSyst predicted yield and biomass of wheat more closely than CERES-Wheat. The combined RMSE for the three water regimes between predicted and observed grain yield was 0.36 Mg ha −1 for CropSyst as compared to 0.63 Mg ha −1 for CERES-Wheat. Similarly, RMSE between observed and predicted biomass by CropSyst was 1.27 Mg ha −1 as compared to 1.94 Mg ha −1 between observed and predicted biomass by CERES-Wheat. Wilmot's index of agreement (IoA) also indicated that CropSyst model is more appropriate than CERES-Wheat in predicting growth and yield of wheat under different N and irrigation application situations in this study.