between fresh biomass and reflectance in each of the bands was very 
similar to that for dry biomass. 
Plant Water: Plant water content (Figure 6), which is one 
indicator of the amount of photosynthetically active vegetation, was 
highly correlated with reflectance in each wavelength band.  Reflec- 
tance was sensitive to changes in plant water throughout the growing 
season, resulting in a much higher correlation than fresh or dry 
biomass. Another measure of plant moisture analyzed was percent 
moisture. Early in the growing season the percent plant moisture 
was high, and due to the sparse ground cover, the near infrared 
reflectance was low. As the plants grew, the plant moisture remained 
around 80 percent while the amount of green vegetation and the near 
infrared reflectance increased to a maximum. Therefore, there was no 
correlation between the spectral data and percent plant moisture early 
in the growing season. When heading occurs and the leaves begin to 
senesce, the percent plant moisture begins to decrease along with the 
near infrared reflectance. 
Prediction of Canopy Variables 
  
Estimation of canopy variables from multispectral data for use 
in crop growth and yield models is an important potential application 
of multispectral remote sensing. Understanding the relation of agro- 
nomic properties of crop canopies to reflectance in various regions 
of the spectrum leads to the development of regression models for 
estimating canopy variables from measurements in several wavelength 
bands. 
Table 2 shows results using selections of one to six wavelength 
bands for prediction of canopy variables. By computing all possible 
regressions, the best subset of each size was selected considering 
the amount of variability explained and the bias of the resulting 
regression equation. The near and middle infrared bands were found 
to be most important in explaining the variation in canopy variables. 
For leaf area index and percent soil cover, the 0.76-0.90 um wave- 
length band accounts for more of the variation than any other single 
band. The 2.08-2.35 um wavelength band is the single most important 
band in explaining the variation in fresh biomass, dry biomass, and 
plant water. The 2.08-2.35 um wavelength band is one of the two most 
important bands in explaining the variation in percent soil cover and 
one of the three most important bands explaining the variation in leaf 
area index. 
From Table 2 the difference between the number of bands entered 
that produce a near maximum R2 and the number of bands entered where 
the resulting prediction equation is unbiased can also be examined. 
An unbiased equation results when the "Cp value is equal to or less 
than the number of terms in the resulting regression equation (Mallows, 
1973). For leaf area index and percent soil cover, the near maximum