AUXILIARY USE OF LANDSAT DATA
IN ESTIMATING CROP YIELDS
F. s. S:fgma n
G. A. Lar~en
Econom:fcs and Statistics
U. S. Dept. of Agriculture
AU~ILIARY USE OF LANDSAT DATA IN ESTIMATING CROP YIFLDS.
By R. S. Sigwan
and G. A. Lar8en; Statistical Research Divi~ion, Econowlcs and Statistics
Service, V. S. Dept. of Agriculture, Decem~er 19~O.
In this report a regression-like
eFtimator i~ inve~tigated ~s a ~ethod
to use LANDSAT riata to improve E5S ob)ective yield (OY) esti~ates for corn
The estimator's pri~ary variahle, which is required to be
known only for samplec fields, is estimate~ field-level yield compute~
from o~served plot rata.
auxiliary variahles are
field-level means of MSS radiometric v~lues and/or various MSS vegetative
By definition, ·auxiliarv varia~les must ~e kn0WTI over the ent~re
population, which in tbis c~se is all land planted to the crop of interest
within S0we target area.
Since a pixeJ's population me~hership
known for pixels exterior to June Enumerative Survey (JES) segments, the
set of all pixels classified to the crop of interest is used to define a
for the estimator.
creates an esti~ator bias which
is estimated from la~eled LA~TSAT data coinciding with JES segments.
of the developed estimator with 197e unitempora1
producen ~ixed results in su~-state analysis areas.
In some areas of
Iowa, no yield estimation improvements fro~ LANDSAT were indicated.
other parts of Iowa, yield estimation improvements were moderate for soy-
heans and marginal for corn.
Haze correction was used to develop
improvements were modest
for hoth corn and soybeans.
However, if LANDSAT data are available from
the acreage estimation program at no additional cost then some savings can
be realized In the OY program through reduced sample size.
Yield Data Coll