To study impacts of climate variations on crop production, the growth models are used to simulate yields in present vs. changed climate conditions. Met&Roll is a four-variate (precipitation amount, solar radiation, minimum and maximum temperatures) stochastic weather generator used to supply synthetic daily weather series for the crop growth model CERES-Maize. Three groups of experiments were conducted in this study: (1) Validation of Met&Roll reveals some discrepancies in the statistical structure of synthetic weather series, e.g., (i) the frequency of occurrence of long dry spells, extreme values of daily precipitation amount and variability of monthly means are underestimated by the generator; (ii) correlations and lag-1 correlations among weather characteristics exhibit a significant annual cycle not assumed by the model. On the whole, the best fit of the observed and synthetic weather series is experienced in summer months. (2) The Wilcoxon test was employed to compare distributions of maize yields simulated with use of observed vs. synthetic weather series. As no statistically significant differences were detected, it is assumed that the generator imperfections in reproducing the statistical structure of weather series negligibly affect the model yields. (3) The sensitivity of model yields to selected characteristics of the daily weather series was examined. Emphasis was placed on the characteristics not addressed by typical GCM-based climate change scenarios: daily amplitude of temperature, persistence of the weather series, shape of the distribution of daily precipitation amount, and frequency of occurrence of wet days. The results indicate that some of these characteristics may significantly affect crop yields and should therefore be considered in the development of climate change scenarios.
