National Wheat Yield Prediction of France as Affected by the Prediction Level.

Abstract

The effect of aggregation of simulation results on the prediction of national total wheat yield of France was examined using five prediction models. Three of these models were combinations of trend functions and averages. The other two, an additive and a multiplicative model applied crop growth simulation results in combination with a trend function. The simulation results were aggregated to sub-regional, regional and national level and subsequently introduced, in combination with the estimated planted area, in the prediction models. The prediction results werecompared with official yield statistics. The study demonstrates that aggregation of the simulation results affects the prediction results. Better results are obtained when the predictions are executed at a regional or sub-regionallevel. The multiplicative model performed best.