United States
(
Department
af
Agriculture
Statistical
(Reporting
-(
Service
~.
Statistical
Research
Division
June 1983
(
(
Comparison of the
CEAS and Thompson-
type Models for
Soybean Yields
in
Iowa, Illinois, and
Indiana
Jeanne L. Sebaugh
James J. Cotter
COMPARISON OF THE CEAS AND THOMPSON-TYPE MODELS FOR SOYBEAN YIELDS IN IOWA,
ILLINOIS AND INDIANA. By Jeanne L. Sebaugh and James J. Cotter; Statistical
Research Division, Statistical Reporting Service, U. S. Department of Agri-
culture, Columbia, Missouri 65201; June 1983. SRS Staff Report No. AGES830613.
ABSTRACT
The CEAS and Thompson-type regression models compared in this report use the
basic input variables of year and monthly average temperature and total pre-
cipitation to predict soybean yields in Iowa, Illinois, and Indiana.
Individual
CEAS models are developed for each CRD and State using stepwise methods, often
including stress variables based on estimates of PET and ET. The Thompson-type
models pool CRD observations within each State and pool State-level observations
within the region.
The fixed weather variables consist of 12 linear and
quadratic terms expressed as deviations from normal weather.
There is little
difference in the models based on the comparisons according to eight model
characteristics.
The accuracy of the CEAS model's predictions is somewhat
higher in Iowa and Indiana. HoWever, both models tend to underestimate in
above-average yielding years and overestimate in below-average yielding years.
Key Words:
Model comparison, crop yield modeling, regression models, soybean
yield models, pooled models.
*********************************************************
*
*
: This paper was prepared for limited distribution
:
t to the research community outside the U. S. Depart-
t
t ment of Agriculture.
The views expressed herein
:
t are not necessarily those of SRS or USDA.
t
*
*
*********************************************************
The authors wish to thank Wendell Wilson for his comments and assistance,
Jean Sparks for typing this report, and J