rier
ition
sics,
58.
essung
eri
ltech/
p. Image
Death
rain
nnual
and
pp.
23.
»cul-
ind.
Jniv.
-61.
mo 437,
E
H
|
+3,No.8.
\ufnahmen",
r Remote
USA,
Image
ition of
n Image
al., USA.
, God-
PP.
tral
_L2(2).
raphic
on
Cana-
ALTERNATIVES IN CONDITIONAL TECHNOLOGIES
Knut Bulow
Remote Sensing Research Center
U.S. Department of Agriculture
Houston, USA
I. Introduction
Experience gained by analyzing several thousand scenes through the
Large Area Crop Inventory Experiment (LACIE) (Ref 1) has taught the
United States Department of Agriculture (USDA) analysts that there
is no one optimum approach to processing all agricultural scenes.
There exists an infinite number of ways to structure decisions in
image interpretation, e.g. by spectral information, by procedures
based upon variations in crop development, by relationship of
spectral data to soil, climate, etc. It has been discovered that
it is possible to "control" the variability among scenes by delin-
eating areas that are relatively similar in soils, climate, and
topography, (Ref 2), and by using alternate processing techniques
based on scene characteristics.
In remote sensing of agricultural activities there exists a wide
variety of characteristics which dictates those candidate techniques
which can be used in order to arrive at the best and most efficient
results. Each of the different processing techniques, that have
been developed in LACIE and other interpretation experiments has
its own inherent limitations. Because of this, each technique should
only be applied under certain conditions. When analyzing remotely
sensed data in agriculture it is necessary to provide analysts with
alternatives in conditional technologies versus those technologies
which have no constraints and demand no alternatives.
The constraints imposed upon a processing technique utilizing multi-
spectral data are related to manpower, timeliness, data availability
and characteristics, scientific limitations, accuracy and economic
considerations. Each user can control several of these constraints
through his requirements and resources. The two that he has little
control over are scientific limitations and data availability and its
characteristics. Since users have no control over scientific limi-
tations one is forced to develop techniques for processing remotely
sensed data around data availability and its characteristics.
Space Administration (NASA), and the National Oceanographic and
Atmospheric Administration (NOAA) to develop methods to estimate
wheat production over designated areas. The method which was
developed used meteorological and historical yield data to estimate
wheat yields and Landsat data to estimate the areal extent of the
wheat crop.