Evaluating CropSyst simulations of wheat management in a wheat-fallow region of the US Pacific Northwest

The need for tools to evaluate cropping systems and their impact on production and the environment is increasing. Field experiments, which are usually site specific, require considerable resources and time to test hypotheses or management options. Computer models for cropping systems simulation are an alternative, but they require calibration and validation before applications can be developed for a given region. The ability of CropSyst, a cropping systems simulation model, to simulate spring and winter wheat yields and evapotranspiration in a wheat-fallow region of eastern Washington for different tillage and residue management practices was evaluated. Validation of the model was accomplished by (a) providing initial conditions prior to planting for each cropping season of a 6-year wheat-fallow rotation and (b) providing initial conditions at the beginning of the rotation without reinitializing the model in later years. The statistical structure (mean and SD) of simulations and field data were similar. The root mean square errors fluctuated from 7 to 14% of the observed mean for grain yield and from 5 to 9% for evapotranspiration, depending on the type of evaluation, and was 17% for soil moisture profile in the case of continuous 6-year rotation simulation without reinitialization. The Wilmott index of agreement fluctuated from 0.92 to 0.97 (an index of 1.0 indicating perfect agreement), with all values but one equal or better than 0.94. Similar water production functions were obtained from simulation and field data. Based on this validation, CropSyst appears promising as a tool to analyze the effectiveness of management practices for water conservation in a wheat-fallow cropping region. (C) 1998 Elsevier Science Ltd. All rights reserved.