Testing winter wheat simulation models' predictions against observed UK grain yields

Wheat models such as CERES-wheat, AFRCWHEAT2 and SIRIUS predict grain yield and have been widely used, in particular to assess possible effects of climate change. Here, observed yields from well-managed and documented UK agricultural experiments were used for a large-scale study of these models' grain yield predictions. None of the models accurately predicted historical grain yields between 1976 and 1993. Substantial disagreement was found between the models' predictions of both yield and yield loss due to water limitation. A regression of observed yields on monthly climatic variables indicated that indirect climatic effects play a considerable role in UK well-managed yields. The study shows that more work is needed before such yield predictions can be used with confidence in decision support or climate change assessment in the UK. (C) 1998 Elsevier Science B.V.