Predicting Dry Bean Yields in Phenological Time

Abstract

A multiple regression model was developed to predict dry bean yields for the dry bean-growing region of western Nebraska. Within the context of the dry bean phenological growth stages, the model assesses the significance and magnitude of weather, climatic and irrigation disasters, and technology. Yield data was taken from four western Nebraska counties (Box Butte, Morrill, Scotts Bluff, and Sheridan) for 1940 to the present. Weather data used to predict yield were daily maximum and minimum temperatures and precipitation from a single weather station in each respective county. Farmers and industry personnel in each county provided historical recollections of years when county-wide disaster events occurred, such as hail and early frost. Four 21-day growth stages were defined as: emergence and stand establishment, rapid vegetative growth, flowering and pod development, and pod fill and maturation. The model predicts current season yields at the end of each growth stage as the season progresses. In 1995, the model predicted a yield of 1731 lb/A—3% below the final USDA estimate for Scott Bluff county. The 1996 predicted value is for 2162 lbs/A—the fifth largest in history. Providing accurate real-time yield predictions assess which weather-related factors are significant, and ranks the relative impacts of weather effects on dry bean yields. Technological progress in yield can also be measured. This information aids farmers in the selection of varieties and management practices that reduce yield losses, predicts regional crop production for agribusiness planning, and provides plant breeders the guidelines for variety development.