partir
2 pos-
denti-
uvent
ution,
ossi-
t
RESEARCH ADVANCES IN SATELLITE-AIDED CROP FORECASTING*
by
J. ERICKSON, J. DRAGG, R. BIZZELL, and M. TRICHEL
National Aeronautics and Space Administration
. Lyndon B. Johnson Space Center, Code SH
Houston, Texas 77058, United States of America
ABSTRACT
Newly developed automated information extraction procedures for
non-U.S. crop analysis of multitemporal Landsat data do not require
ground-based observations. Experimental results indicate a greatly improved
capability for practical and affordable use of the more accurate and efficient
techniques, on a global basis. This resesarch is carried out as part of the
Inventory Technology Development Project in the joint program for Agriculture
and Resources Inventory Surveys Through Aerospace Remote Sensing (AgRISTARS).
1. INTRODUCTION
Research continues to improve current procedures in satellite-aided
crop forecasting of global production and to make advances toward practical,
viable, systems capabilities. These capabilities, when added to current
agricultural information systems, are expected to provide more timely and
accurate non-U.S. crop information than is now available.
While the overall objective of this research! is to develop
technology for extracting agricultural information of various kinds, the focus
is on improved production forecast technology for assessing non-U.S. areas.
The new technology will be evaluated by the U.S. Department of Agriculture
(USDA) for possible integration into its information systems. More
specifically, our objective is to develop procedures for using aerospace
remote sensing and related technology at several times during the growing
season to provide more objective and reliable crop area forecasts, with
improvéd preharvest production estimates for selected countries and crops.
Requirements specifications on future satellite and sensor systems for
non-U.S. Agriculture information uses is also a part of this objective.
We have adopted four criteria that reflect key characteristics of
the technology and guide us in the research These are timeliness, affordabil-
ity, general applicability, and accuracy. Timeliness emphasizes the quick
extraction of information. Timeliness is associated with early season
estimates, as well as with estimates made throughout the season. Affordabil-
ity reflects efficiency and inexpensiveness. General applicability includes
having suitable techniques in non-U.S. crop regions, as well as having
objective and improvable procedures. Accuracy reflects the degree of bias and
variance over time and responsiveness to factors affecting departures from
average.
II. TECHNICAL APPROACH
We are developing technology that provides satellite-based, objec-
tive estimates of area, yield, and production as a set of inputs into a
comprehensive, multidata source information system. The conceptual framework
involves estimating crop area and yield for specified regions and multiplying
the two to obtain production at the regional level. Automatic data processing
approaches are considered necessary to provide objective, timely, and reliable
SE SD
*Presented at the International Society for Photogrammetry and Remote Sensing,
Commission VII International Symposium held in Toulouse, France,
13-17 September 1982,
367
> " —
p 3 s x TERN Ee d RR PER t ede MR RE =
MEM se 8 US CA PU LA a SC E atc s
