Adaptation of the CERES-wheat model for large area yield estimation in Argentina

To estimate yield over large areas in the southern part of the Buenos Aires province, Argentina, the functional model CERES-wheat was calibrated with experimental plot data on the effect of water stress on a spring- wheat cultivar and then validated with several field data sets. Adjustment of genetic coefficients and some source code modifications to the growth and soil water balance submodels were necessary to improve the simulation of crop growth and yield. The revised version of CERES-wheat successfully reproduced wheat yields obtained from the different water stress treatments but was not adequate for simulating the physiological maturity or for the simulation of canopy development under stress. With independent data sets, the revised version adequately predicted observed yields and yield components with a root mean square error for grain yield of 536 kg ha(-1). The results obtained suggest the importance of having detailed initial inputs and the need to include an effect of water stress on phenology and an effect of frost damage on yield in the model.