Early prediction of spring wheat yields in Saskatchewan from current and historical weather data using the CERES-Wheat model

Wheat yields were simulated at three locations in Saskatchewan (Swift Current, Saskatoon and Melfort) using the CERES (crop estimation through resource and environment synthesis-wheat)-Wheat model. The simulations were made using climatic data of selected (base) years from the start of the growing season up to the prediction date; data for the remainder of the season (runout data) were taken from 1960-1990 climate records. Years corresponding to the first, second, and third quartile yield values were selected for the three study sites. Predicted yields were calculated as the means of the yields for the runout years, and were made at five growth stages: (i) emergence; (ii) terminal spikelet initiation; (iii) end of vegetative growth; (iv) start of ear growth; and (v) start of grain filling. Predictions using this method were found to agree well with measured data, suggesting that simulations made using a combination of historical climate records and current weather data as inputs provide good indications of yield. Use of these predictions would allow producers and those in agribusiness to choose alternative actions for minimizing losses when yield prospects are poor. (C) 1997 Elsevier Science B.V.