UTILITY OF CERES-BARLEY UNDER ARGENTINE CONDITIONS

The model CERES-barley was calibrated for five genotypes sown on five dates between June and September under optimal growing conditions. The n, the model was validated with data sets (n = 83) obtained under a wi de range of environmental conditions covering different sowing dates, years, cultivars, fertility, soils and climates. Results of calibratio n showed differences between cultivars in their response to photoperio d, vernalization and in their dry-matter partitioning. In validating t he model, the root mean square error (RMSE) obtained for flowering-dat e estimates was 5.6 days with a greater bias for the extreme sowing da tes (early June and September). Grain yields corresponding to normal s owing dates (mid June-mid August) were adequately simulated; the bias was 15 kg ha(-1), its mean absolute error was 288 kg ha(-1) and the ro ot mean square error 397 kg ha(-1), for observed values ranging betwee n 2000 and 5500 kg ha(-1). We can conclude that under normal sowing da tes CERES-barley is able to simulate grain yield reliably for Argentin e conditions. (C) 1998 Elsevier Science B.V. All rights reserved.