A stochastic modelling approach for real-time forecasting of winter wheat yield

Crop simulation models are receiving increasing use in agriculture and are recommended as multipurpose tools in research and farm management. Of one p articular interest to crop growers is the possibility of applying crop mode ls for real-time yield forecasting. This investigation evaluated the utilit y of the SUCROS model for site-specific real-time crop biomass and grain yi eld forecasting. A stochastic forecasting approach was used combining gener ated weather data with observed data for model updating. The forecast proce dure was tested with field data collected at four sites in the UK over two growing seasons. The results showed that across all site-years, the model i s able to forecast the final biomass and grain yield with <10% bias. There was no significant difference between observed and forecasted biomass and g rain yield for forecasts made at anthesis or milky grain stage although ear lier forecasts did show significant differences. The ranking of the observe d and forecast biomass and grain yield were also highly correlated for the later forecasts. (C) 1999 Elsevier Science B.V. All rights reserved.