Evaluating the Addition of Weather Data to Survey Data to Forecast Soybean Yields

The National Agricultural Statistics Service conducts surveys, during the growing season, to collect plant counts and measurements for forecasting yields of major agricultural crops such as corn, cotton, soybeans and wheat. Plot level data are aggregated to the regional (multi-state) level and used to build regression forecast models. In order to improve the accuracy of August 1 and September 1 yield forecasts, a cumulative precipitation term, indicating total precipitation from April 1 until the forecast date, was added to the models at a regional six-state level. This additional data was believed to contain information which would be helpful in crop yield forecasting. The analysis indicated that the accuracy of the regional soybean yield forecasts was not improved using this precipitation term. Additional analysis is recommended to evaluate the impact of different precipitation terms on soybean yield forecasts at the regional and state levels, and to evaluate the use of precipitation data in forecasting crop yields for corn, cotton and wheat.