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 particular interest to crop growers is the possibility of applying crop models for real-time yield forecasting. This investigation evaluated the utility of the SUCROS model for site-specific real-time crop biomass and grain yield forecasting. A stochastic forecasting approach was used combining generated weather data with observed data for model updating. The forecast procedure 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 is able to forecast the final biomass and grain yield with <10% bias. There was no significant difference between observed and forecasted biomass and grain yield for forecasts made at anthesis or milky grain stage although earlier forecasts did show significant differences. The ranking of the observed and forecast biomass and grain yield were also highly correlated for the later forecasts. (C) 1999 Elsevier Science B.V. All rights reserved.