APPLICA nON OF REMOTE SENSING FOR RICE:
UNITED STATES EXPERIENCE
Galen F. Hart,
Paul W. Cook,
and William H. Wigton
United States Department
of Agriculture, Washington,
This paper describes procedures that integrate Landsat technology into the
area sampling frame methodology
to improve estimates of crop area. The
methodology described is currently being used in four states in the United
States, and it is planned that by 1985, ten states will be included. The orienta-
tion is statistical rather than "remote sensing" because estimating crops
inference, and other principles of statistics. Land area
probability sampling obtained data are combined with full frame digital Land-
sat data through an application of statistical regression. Results are presented
for rice, soybeans and cotton area estimates in Arkansas. Although
cedures have been documented
in other papers and reports, results of the
Arkansas study have not been presented.
AREA SAMPLING FRAME
An area sampling
frame is used by the Statistical Reporting Service (SRS) to
at both state and national
these crop acreage estimates.
some time is
area sampling and its use.
The concepts of the area sampling
frame are very simple:
the total area
to be surveyed
into N small
land areas (sampling
any overlaps or omissions.
2. Select a random sample of n sampling units (segments).
G.F. HART ET AL.
3. Obtain the desired data for reporting units of the population that are in
4. Estimate population
totals by multiplying the sample totals by N/n
is used for crop acreage, livestock, and other agriculture
parameter estimation, and is a dependable method. The use of random sam-
pling in selecting units from the population accomplishes two things:
1. It gives a basis for making inference about the po