On LANDSAT color composites which are formed by transmitting blue,
green and red light through LANDSAT wavebands 4, 5 and 7 respectively,
irrigated fields appear brilliant red because of the high infrared
reflectance of the vegetation. These irrigated fields can be easily
separated from non-irrigated lands because of the regular geometric
shapes and red hue. A suggested procedure follows:
Equipment
A binocular microscope, set at a magnification of 10 X to 15 X, is
the best instrument to use in assessing each block for the amount of
irrigated land. A 10 X hand lens may be substituted if a microscope is
not available.
Interpretation of LANDSAT transparencies
Examine each block (PSU) in the grid in a systematic fashion by
rows and columns and estimate the amount of irrigated lands present to
the nearest five percent. When the irrigated land estimate falls below
5 and above 95 percent, estimate to the nearest 1 percent. Record your
estimate on a form similar to that shown in Figure 3. If estimates of
other land uses are desired--such as rangeland, forestland, etc., their
percentage occupancy should be recorded, keeping in mind that the PSU
total should be 100 percent. As each block is interpreted, add the
estimated percent of irrigated land to that of the preceding block
(column four of Fig. 3). In other words, accumulate the percent of
irrigated land as the interpretation proceeds.
Selection of Sample Blocks (PSU's)
Next, one must make a selection of some predetermined number of
PSU's from the total number of PSU's examined. On the Idaho test site,
we had 9 samples from a total of 270 which represented 3.2 percent of
the area. Often one is restricted by lack of time and money from
conducting an adequate ground check. Usually, a five percent sample is
adequate.
In making the selection of PSU's, the investigator picks the
sample from a random number table. The numbers will run from zero to
the total accumulated percentage of irrigated lands (final number in
column 4, Figure 3). On the Idaho test with 270 PSU's, this total
percentage exceeded 5,000. The random numbers will determine which row
and column (PSU) is selected. All PSU's with one or more percent
irrigated land have an opportunity of being selected; see the wide range
in predicted percentages of irrigated lands for interpreters three and
four (Table 1).
Supplemental Aerial Photography
One must determine the actual area of irrigated lands (a.) within
the selected PSU's to compare it with the estimated amount from LANDSAT
interpretation. Any scale of aerial photography (1:125,000 or larger)
may be used to provide this comparison. Photos should be taken over the
randomly selected PSU's during the same time period if at all possible.
CIR transparencies are preferred because they provide better discrimination
LANDSA
TEST 4
Colum