Íslenskar landbúnaðarrannsóknir - 01.03.1980, Page 19

TEST OF A HAND-HELD RADIOMETER 1 7
TOTAL DRY BIOMASS (g/m2)
TOTAL DRY BIOMASS (g/rn^)
Figure 3. Plot of data values, regression determined equations, and associated r2 values for (A) red
radiance, (B) IR radiance, (C) radiance ratio, (D) radiance difference, (E) radiance sum, and (F)
vegetation index.
computation. Regression output included
the regression derived equation, regres-
sion analysis of variance, correlation coef-
ficient (r), coefficient of determination
(r2), measures of central tendency, re-
siduals, etc. (Dixon 1974).
It should be noted that the coefficient of
determination (r2) values for the trans-
formed data using Equations 2, 3, and 4
approximate the actual statistic (s). This
results from the introduction of a small
amount of transformation error for small
and large values of total dry biomass
(Draper and Smith 1966).
Variables evaluated
In addition to evaluating the red radiance,
near infrared radiance, and the IR/red
radiance ratio, three other variables were
transgenerated from the red and infrared
radiance. These were the IR—red
radiance difference, the IR+red radiance
sum, and the vegetation index (VI) of
Rouse et al. (1974). The VI is the IR— red
radiance difference divided by the
IR+red radiance sum and has been
shown to have a high degree of utility for
vegetational biomass remote sensing
(Tucker 1979).
Equation 2 was used for the red
radiance data while Equation 4 was used
for the IR radiance, the radiance ratio,
radiance difference, and the VI. The
radiance sum could not be consistently
approximated with any of the four regres-
sion models used.
Experimental results
Inspection of the regression results indi-
cated that a substantial degree of simi-
larity existed between the first five meas-
urement trials (Table 2). Because it would
be impractical to compare all six meas-
urement trials by presenting the figures
for each of six variables for the six trials,
the complete results of the second trial
2