e a . + , , WAVELENGTH (NN) 
vos sm. o 78. d 
  
DRTE: 788783 TIME: 12.27 - 12.28 RRER: 15 
Fig. 3. Calculated reflectance from data in fig. 2. * | standard 
deviation is plotted at each measurement point. 
method used is the same as mentioned above. The instrument factor k for 
each wavelength interval is derived from 
v Od r 
OR NE 
where p is the bidirectional reflectance, Jr the directional reflectance 
from the known reference panel, L, the instrument value of the radiance 
from the panel and E; the instrument value of the irradiance on the diffus- 
ing disc. 
Rejection of spurious instrument values 
In a set of measurements some of the values can be anomalous, as seen in 
fig. 2. If these outlying values are wrong and accepted or right and rejected, 
the computed reflectance value will be biased. Hence, it is desirable to find 
an objective criterium for acceptance or rejection of values. Several sta- 
tistical methods have been proposed by e.g. Anscombe (1960), Bliss et al 
(1956) and Grubbs (1950). However, very few values are sampled at each wave- 
length interval during a set of radiance measurements. This makes it 
questionable to use probability tests. As the energy distribution of the 
spectrum is relatively constant but the effect varies, the ratio have been