(morning or afternoon), the ratio is nearly linearly related to plant cover.
Essentially, the opposite obtains with the ND, shown in figure 28. The values for
the sunlit soil conditions, representative of solar noon, are nearly linear with
plant cover. For shaded conditions, the values increase rapidly for low values of
plant cover and become insensitive to plant cover changes as the fraction becomes
large and approaches 1. These calculations show the relative merits of the two
indices with respect to sensitivity to plant cover.
Figure 28.--Calculated normalized difference using a red (MSS5) and an infrared
(MSS7) band as a function of plant cover for the conditions of sunlit plants
and sunlit and shaded, wet and dry soil.
Calculations for the reflectance PVI are shown in figure 29. They show that
the PVI increases linearly with plant cover. Essentially, no difference can be
seen between wet and dry soil, showing the ability for the PVI to remove the soil
The lines for shaded soil have a negative intercept.
This is an
indication that the soil line, extrapolated from sunlit conditions (figs. 23 and
24), is not a complete representation of the total situation and needs additional
RADIOMETER RESPONSE FUNCTIONS
We have used the term “band” to signify a wavelength interval and have
identified these bands with names (i.e., red, IR) and given numbers to specify
the bounds (example, red band of Mark II, 0.63 to 0.69
m). This implies that all
of the radiation (from 0.63 to 0.69
m) striking the radiometer detectors is
In practice, filters do not cut off radiation at precisely a given
wavelength. Some radiation less than 0.63
m (for our example) is detected, and
not all of the radiation greater than 0.63 is detected. The value of 0.63 is a
nominal value. A plot of the fraction of the radiation received versus the
wavelength is known as a response function. In figure 22 of the
Figure 29.--Calculated Perpendicular Vegetation Index (PVI) values using MSS5 and
MSS7 of the Exotech as a function of plant cover for the scene conditions,
sunlit plants, sunlit wet and dry soil, and shaded wet and dry soil.
last section, spectra for soil and alfalfa were shown for the wavelength region of
0.4 to 1.0
m. In that figure, dashed lines were used to delineate the nominal
band boundaries. Using response functions at each wavelength, the relative re-
sponse can be multiplied by the spectrum for a particular target (e.g., alfalfa
and soils) and summed to yield a value proportional to the actual response of a
radiometer if used over the same target for which the spectrum was measured. Thus,
with sets of spectra and with response functions for several instruments, differ-
ent bands can be compared as to their sensitivity to vegetation, and vegetation
indices can be calculated and compared among instruments.
Relative response functions for four radiometers
: Relative response functions
for the PMT 2-band and the Mark II 3-band are shown in figures 30 and 31. These
data were provided by C. J. Tucker.
Figure 32 shows the response functions for
the Exotech 4-band instrument (data provided in the instruction manual), and fig-
ure 33 presents data for the LANDSAT-l MSS (taken from Slater 1979). The four
figures have identical values for the ordinate and the abscissa to facilitate
comparisons. The symbols shown in figures 32 and 33 are for the purpose of identi-
fication because of the overlapping of the bands. They are not intended to imply
A comparison the PMT and the Mark II instruments shows that the red bands are
nearly identical in width, whereas the IR band of the Mark II is almost twice as
wide as is the IR band of the PMT. The Mark II has a band (called the water ab-
sorption band) at 1.55 to 1.75
m that the other three devices do not have.
Figure 30.--Relative response functions for the PMT 2-band radiometer.
The Exotech and LANDSAT bands are much wider than any of the visible or IR
bands for the PMT or the Mark II. The MSS5 band (identified with crosses in figs.
32 and 33) includes most of the area covered by the red bands of the PMT and Mark
II, but is wider toward the lower wavelengths. The IR bands of the PMT and Mark
II are partially included in MSS6 and are almost completely included in MSS7.
A comparison of the Exotech and the LANDSAT response functions shows that
MSS4, MSS5, and MSS6 are reasonably similar, but MSS7 differs in that the
Exotech appears to have about a 0.05-
m shift towards the shorter wavelengths.
The significance (or nonsignificance) of the different band widths becomes evident
when relative response to spectra is calculated.
Field spectrometer data
: In early April 1979, a team from NASA/GSFC
a field spectrometer to Phoenix to gather spectra over wheat plots at the U.S.
Water Conservation Laboratory. Measurements were made over three plots, each plot
containing four subplots. The three plots had been planted at different time in-
tervals (Nov., Dec., Feb.), and the subplots received different irrigation treat-
ments. Spectra for 2 of the 12 subplots are shown in figures 34 and 35. In figure
34, the data are for a well-watered plot planted in
E. Chappelle headed the team and provided the spectrometer data.
Figure 3l.--Relative response functions for the Mark II 3-band radiometer.
December 1978. The plants had just begun heading and covered about 95 percent of
the soil. The plants were mostly green with only a very few brown leaves showing.
Figure 35 shows spectra for a plot planted in February 1979. At the time of mea-
surement, the plants had not yet headed and covered about 40 percent of the soil.
In addition to late planting, this plot received fewer than the optimum number of
Digitized spectrometer data were interpolated to yield values at every
nanometer (1/1000 of a micrometer). Response functions were digitized at each
nanometer, and the product of the response function and the spectra at each
nanometer within a band width was formed and summed to yield a spectral response
value for each waveband on each of the four radiometers for spectra from the 12
wheat subplots. The absolute value of the summation is not of interest, but the
relative magnitudes among bands and instruments allow a comparison to be made of
the various bands.
The digital count range for MSS4, MSS5, and MSS6 on LANDSAT is 0 to 127. For
MSS7, the range is 0 to 63. To make our results somewhat comparable to LANDSAT,
we divided the summed values of response times spectra by two. We will not con-
sider atmospheric effects on radiative transmission to satellite altitudes in
this discussion. Atmospheric effects have been treated by Turner et al. (1971),
Turner and Spencer (1972), and Richardson et al. (1980).
Figure 32.--Relative response functions for the Exotech 4-band radiometer. The
symbols are for band identification purposes and do not imply data points.
Comparison of bands among instruments
: We chose the Mark II instrument to
compare with the other three radiometers in the following figures. A high corre-
lation coefficient indicates that one instrument has no advantage over the other
for obtaining information about vegetation conditions. A low coefficient indi-
cates that one band may contain information not shown by the other.
Figure 36 compares the red band of the Mark II with the red band of the PMT
2-band instrument. As one would expect from the close alinenent of the response
functions (figs. 30 and 31), the correlation between the two instruments is very
good, with an r
= 0.999. The red band of Mark II is compared with the MSS4 and
MSS5 of the Exotech in figure 37.
A relatively high correlation exists between
the bands, especially with MSS5, which includes the red region. Similar good cor-
relations exist between the Mark II and the LANDSAT MSS4 and MSS5 as shown in
Figures 39, 40, and 41 compare the Mark It IR band with the PMT 2-band IR
and the Exotech and LANDSAT MSS6 and MSS7 bands. Although some difference in band
widths was noted for the PMT and Mark II, the IR bands are correlated with a co-
efficient of 0.996, indicating that the band width is not too critical if it is
at a longer wavelength than about 0.75
m (fig. 34). This statement gains
Figure 33.--Relative response functions for the LANDSAT-1 multispectral scanner.
The symbols are for band identification purposes and do not imply data points.
additional support from data in figures 40 and 41 where the Exotech and LANDSAT
MSS7 bands are related to the Mark II IR band with a correlation coefficient of
0.995 and 0.991, respectively. MSS6, which encompasses the abrupt shift from low
to high reflectance over vegetation, shows much less correlation, having coeffi-
cients of 0.768 and 0.753.
We conclude, from the above discussion, that the red bands of the Mark II and
PMT 2-band and the MSS5 of the Exotech and LANDSAT will yield equally good results
over a wheat crop. Also, the Mark II and PMT IR bands and MSS7 of the Exotech and
LANDSAT will give equally good results.
The water absorption band of the Mark II instrument has no comparable bands
on the other three instruments.
Figure 42 shows that it is not correlated with
the IR band, but, as shown in figure 43, it is reasonably well correlated with the
red band. Correlations (r
) made between all pairs of the 13 bands on the four
instruments are presented in table 3. The data show that the water absorption band
is reasonably well correlated with the MSS4 bands on the Exotech and LANDSAT
(0.935 and 0.934), and slightly less well correlated with the MSS5 bands (0.92).
The red bands on the PMT and the Mark II have coefficients of 0.896 and
Figure 34.--Spectrum obtained over a well-watered wheat plot planted in December
1978 at Phoenix (data furnished by E. Chappelle, NASA/GSFC). Plant cover was
about 95 percent.
These correlations raise the question:
information is contained in the water absorption band that is not in the visible
green and red bands?
We pose this only as a question since we are working with a
limited data set.
Extensive field use of the Mark II should show the value of
Comparison of vegetation indices among instruments:
In addition to com-
paring individual bands, it is of interest to compare vegetation indices as would
be obtained over the same target with different instruments. The IR/red ratios
for the Mark II and the MSS7/MSS5 ratios (also IR/red but wider band widths) for
the Exotech were calculated and plotted in figure 44.
Linear regression analyses
indicate the two ratios are linearly related with a correlation coefficient of
0.996. We conclude that over a range of vegetation densities, from about 40 per-
cent to 100 percent cover, the ratio data from the two instruments could be
readily compared using a linear transformation; however, for sparse vegetation
and bare soils the data may not fit the linear function given in figure 44.
Clarification of the relation for sparsely covered soils awaits more experimental
Figure 35.--Spectrum obtained over a water stressed wheat plot planted in Febru-
ary 1979 at Phoenix (data furnished by E. Chappelle, NASA/CSFC). Plant cover
was about 40 percent.
The ND’s for the two instruments are shown in figure 45. Over the range of
plant densities shown here, the relationship is linear with an r
value of 0.999;
however, the nonzero intercept indicates that the relation may not be linear over
the entire range from 0 percent to 100 percent plant cover.
The water absorption band
The water absorption band is sensitive to water
in plants and exhibits the greatest contrast between green vegetation and bare
soil (Learner et al. 1978). This sensitivity to water could greatly improve our
ability to detect the presence of water stress and other factors that inhibit
water uptake by plants. With a number of radiometers in this band, data should
soon be available to evaluate its potential.
Some questions to be answered are:
Should this band be ratioed with another? If so, which one? What form of vegeta-
tion indices can enhance information in this band? Must we use reflectances or
can radiances be readily corrected for sun angle?
Answers to the above and other questions await data from field experiments.
The information gained by hand-held radiometers should prove to be a valuable
guide to the interpretation of Thematic Mapper data that should be available
after the launch of LANDSAT D.
Figure 36.--A comparison of the red bands on the Mark II and the
2-band. Data are
for 12 wheat subplots.
Figure 37.--A comparison of the red bands on the Mark II with MSS4 and MSS5 of
the Exotech. Data are for 12 wheat subplots.
CALCULATION OF APPROXIMATE LOCAL STANDARD TIME FOR LANDSAT OVERPASSES
LANDSAT satellites were launched in sun synchronous orbits inclined 99° from
the Equator, causing the satellites to cross the United States in a south-
southwestwardly direction, crossing the Equator nominally at 0930 local civil
time in descending mode. Precise knowledge of the local standard time that the
Documents you may be interested
Documents you may be interested