Investigate The Potential For Improving Early Season Soybean Yield Forecasts

Improve The State Level Yield Forecasts/Estimates Through More Optimal Partitioning of The Constrained Linear Regression Estimator

Evaluate The Usefulness of Drying Green Bolls In Improving Early Season Yield Forecasts

Evaluate The Usefulness of Fiber Properties In Improving Cotton Yield Models

Investigate The Potential For Improving Early Season Soybean Yield Forecasts


This two-year cooperative research project was initiated in the summer of 1998 , and work will commence in 1999. In cooperation with faculty and students at Southere Illimois University-Carbondale, plant characteristics, as well as selected variables exogenous to the plant, of potential value as predictors of soybean weight per pod will be investigated. Sampling methods for measuring potentially useful predictor variables will be researched. Field measurements of actual weight per pod will be used to evaluate the effectiveness of potential predictor variables and associated sampling plans.

Improve The State Level Yield Forecasts/Estimates Through More Optimal Partitioning of The Constrained Linear Regression Estimator

Estimates of yield from the objective yield surveys are produced by means of a constrained linear regression model. The estimate for a region is determined first, then compatible State level estimates are produced. The State level estimates are compatible in the sense that the weighted sum of the State level yield estimates equals the regional yield estimate. The State level yields are weighted by the harvested acreage of the crop.

Because the growth of a plant is governed by natural conditions and cultivation factors which are independent of political boundaries, research is being performed to assess the benefit of partitioning an objective yield region based on the natural conditions rather than State boundaries. The precision of the compatible State level yield estimates depends on the way the region is partitioned. For example, partitioning a region into climatic zones may result in more precise State level estimates than partitioning solely on State boundaries. Under an auxiliary partition, the State level estimates would be obtained by using a linear combination of the estimates obtained from the natural partitioning.

The research will investigate the benefit of augmenting the yield models as they are currently being used with additional information already available. 

During 1998 a manuscript of "Compatible Yield Estimation Using Constrained Linear Regression" was submitted to the Journal of Agricultural, Biological and Environmental Statistics, a refereed journal of the American Statistical Association. 

Evaluate The Usefulness of Drying Green Bolls In Improving Early Season Yield Forecasts


Final weight per boll predictions from the Cotton Objective Yield (COY) survey are based on the cumulative weight per boll for cotton picked during the growing season. The COY lab weight measurements are currently delayed until the bolls open, but could conceivably be made earlier, since cotton fiber reaches maturity before the boll opens. Therefore, one approach to obtaining earlier weight measurements is to clip closed bolls that have been deemed mature and artificially dry the contents. This project is designed to evaluate the efficacy of this approach in improving early season cotton yield forecasts. 

The drying procedure must be quick, safe and effective. Commercially available microwave ovens are reasonably priced, and can dry seed cotton samples quickly. Safety and operational efficiencies can also be achieved if the drying procedure is designed carefully. Thus, research is needed prior to operational implementation to work out the logistics of the process. A primary goal of this project is to develop safe and effective drying procedures that can be used operationally.

The microwave seed cotton drying project is being conducted in Louisiana and completed its second season in 1998. It will continue in 1999, during which some of the assumptions underlying the project, and their resulting procedures, will be reassessed.

Evaluate The Usefulness of Fiber Properties In Improving Cotton Yield Models 

During the 1995 survey year, procedures were developed to lay out research units in Arkansas, from which field enumerators collected first position cotton bolls. These units were in addition to the regular, operational cotton objective yield (OY) units in the State. The cotton from the research units, along with additional cotton from one row of the regular objective yield plots, was picked, weighed, and ginned. Micronaire measurements were recorded. In 1996 the field procedures were refined to obtain all of the needed research data from the regular OY sample units. From these data and USDA cotton classing office data, values of the individual components of cotton yield were derived. A correlation analysis was performed for a preliminary look at whether any of the additional information could enhance the performance of the current, early-season yield forecast model. A research report entitled, Fiber Properties As Components Of Cotton Yield: The 1996 Arkansas Study was published in November 1997, documenting the results. 

While the 1996 data provided some insights, multiple years of data are required to make definitive conclusions about the use of fiber properties in the cotton yield models. As recommended by the Research Review Committee, the data were collected through the 1998 season. The three years of data will be analyzed in 1999 and a decision made on the usefulness of incorporating fiber property information in cotton yield models.

Top of the Page