Corn Yield Prediction Using Climatology

Abstract

A method is developed to predict corn yield during the growing season using a plant process model (CERES-Maize), current weather data and climatological data. The procedure is to place the current year's daily weather (temperature and precipitation) into the model up to the time the yield prediction is to be made and sequences of historical data (one sequence per year) after that time until the end of the growing season to produce yield estimates. The mean of the distribution of yield estimates is taken as the prediction. The variance associated with a prediction is relatively constant until the time of tassel initiation and then decreases toward zero as the season progresses. As a consequence, perfect weather forecasts reach their peak value between the beginning of car growth and the beginning of grain fill. The change in the predicted yield in response to weather as the growing season progresses is discussed for 1983 and 1976 at Peoria, Illinois. Results are given of an attempt to incorporate 30-day Climate Analytic Center outlooks into the predictive scheme.