Evaluation of the CERES models in different production regions of the Czech Republic

The main goal of this work was to calibrate and evaluate the CERES-Barley and CERES-Wheat crop models. The experimental fields used for the model evaluation are situated in three different production regions (maize, sugar beet and potato main growing regions, respectively) with altitudes of 179, 204 and 560 meters above the sea level. Grain yield and date of anthesis together with maturity dates served as reference for the model evaluation. Two evaluation approaches were tested in this study. The first one uses historical data series and it is based on long-term field experiments with capability to reflect interannual weather variability. The second approach uses results of one-year multiple treatment experiment. The model evaluation is then based on a set of treatments differing e.g. in sowing date or an amount of used nitrogen fertilizer. Grain yields simulated by both models are acceptable when compared with experimental results: the coefficient of determination for historical series varied from 0.69 to 0.86 for evaluation of CERES-Barley at the three examined sites and reached values of 0.60 and 0.86 for the CERES-Wheat model at two experimental sites. The lower coefficient of determination of the wheat model was recorded at the locality with the highest altitude and coldest winter conditions. There, also the worst reliability of simulated phenological development was noted. At the second locality where the CERES-Wheat model was tested and at all three localities where CERES-Barley was applied, the simulated duration of vegetation period and anthesis dates were relatively accurate and yielded strong statistical correlation. The one-year multiple treatment experiment proved to be useful to determine the models sensitivity to differences in crop management. The combination of both approaches seems to be the best solution for evaluation of similar crop models if the detail long term experimental data are not available.