A MODEL OF SPRING WHEAT (Triticum aestivum) FOR LARGE AREA YIELD ESTIMATIONS ON THE CANADIAN PRAIRIES

Abstract

A physically-based model is described which calculates hard red spring wheat yields for the Canadian prairies given daily weather parameters. The model uses a simple resistance analog to calculate plant water stress and a nonlinear response factor for temperature effects. Yield data on a large area (crop district) basis were used both for calculating coefficients in the model with a nonlinear least squares algorithm and for verification purposes. Weather effects were separated from technological trends using multiplicative and additive non-linear time functions. The model with the additive time trend produced smaller standard errors of estimate (SEE) on all soil types; SEEs between actual and calculated yield averaged 157, 225 and 253 kg ha−1 for the Brown, Dark Brown, and Black soil zones respectively using the additive time trend. Use of the model for predictive purposes is discussed.Key words: Time trends, nonlinear fitting, physical model, crop district scale