695
A second advantage of the periodic coverage is the ability to monitor the
development of agricultural crops. While cloud coverage over agricultural
areas significantly influences the amount of periodic coverage available, the
coverage can still be used. The potential exists for crop yield estimation
and prediction, by comparing the crop development, as sensed from ERTS, with
a crop calendar of normal development.
TABLE 4. SIGNATURES OF TWO VEGETATION CLASSES ON TWO DAYS
JUNE
MSS-5
MSS-5
MSS-6
MSS-7
Hdw/Conif/Grass
27.32
(0.91)
17.82
(1.45)
49.62
(5.52)
27.53
(3.88)
Shrub/Swamp
27.69
(1.20)
17.25
(1.53)
47.06
(5.69)
25.75
(3.94)
MARCH
Hdw/Conif/Grass
29.96
(3.75)
28.07
(3.11)
29.93
(2.96)
15.82
(1.28)
Shrub/Swamp
23.69
(1.74)
20.25
(4.04)
22.19
(3.60)
11.38
(2.60)
Mean signature values are shown, with standard deviations in parenthesis.
Although there is not sufficient space here to describe the techniques, two
additional techniques are worthy of comment. These are proportions estima
tion techniques (Horwitz, 1974) which have major importance for accuracy of
area determination (Malila, 1973) with coarse ground resolution satellite MSS
systems and adaptive decision-directed classification techniques (Crane, 1974)
for overcoming gradual changes and variations which cause degradation in
performance.
USER APPLICATION MODEL DEVELOPMENT
One critical portion of the Earth Resources Survey System is the User
Application Model, which relates the output of the extractive processors and
ancillary data to generate information which a user can employ directly. User
models may be very simple — if the user wants a map of vegetation types, the
output of the extractive processing may serve him directly, and the
user application model is absent. But if a Department of Agriculture official
wants to know what is the projected wheat production in Kansas, the user model
may combine the total productive acreage of wheat (obtained from a remote
sensing system), with some farmer’s estimates of the yield of particular fields,
and some estimates from the weather service of future weather trends, to
calculate the total production of wheat in Kansas.
As an example of a user application model, consider Figure 3. This model
predicts the population of migratory waterfowl, given the water supply conditions,
the food supply conditions, and ancillary variables such as number of nesting
pairs, predation, and mortality. Also shown in Fig. 3 on the right are elements of