| Abstract: | Seasonal evapotranspiration from dryland and irrigated wheat plots at Goodwell, Oklahoma during a three year period (1964-1967) could not be adequately described, without the inclusion of a drainage model, from data available. Short term evapotranspiration calculated during periods without irrigation or with only small amounts of precipitation approximated measured values. A multiple linear regression equation based on precipitation received during nine physiological growth stages and nitrogen fertilizer interactions was used to predict dryland wheat grain yields. The yield prediction model allowed for predicting yields based on the planting date, soil moisture at planting and expected rainfall at selected probability levels. Yield predictions could be updated at anytime during the growing season by replacing expected precipitation values with values of precipitation actually received. The resulting regression equation was adequate for predicting yields at the same location (Goodwell, Oklahoma) for 1968 and 1969 based on average precipitation or the minimum precipitation expected 50% of the time. Predicted yields made after planting, based on a combination of measured and expected precipitation, showed considerable variation from the actual yields. This was explained by unrealistic signs and magnitudes of some of the regression coefficients that cancelled out when the whole equation was considered. Additional years of data should provide an equation with coefficients that more nearly describe the long term effects of precipitation received during each of the nine growth stages. |
|---|