USING THE SOUTHERN OSCILLATION TO FORECAST TEXAS WINTER-WHEAT AND SORGHUM CROP YIELDS

Time series models are used to examine the impact of Southern Oscillat ion (SO) extreme events in estimating and forecasting Texas sorghum an d winter wheat yields. It is shown that a significant correlation betw een the SO events and yield does not necessarily translate into better forecasts. Including SO events in forecasting yields decreases the fo recasts mean squared error for winter wheat, but has no significant im pact on sorghum forecasts. The most important factor impacting yields over the dataset length (1876-1993 for wheat and 1924-92 for sorghum) is technological change. This change is more important for the period after World War II. Examining the dataset by periods, before rapid tec hnology change and after, indicates the SO impacts crop yields differe ntly between periods.