115 
indicate the results of this test and confirm the visual validation. 
(3) 
point at 
e corre- 
on the 
that for 
topog- 
to the 
resam- 
process. 
results 
tows the 
ted areas 
Number of 
checkpoints 
RMS for i-direction 
[pixels] 
RMS for j-direction 
[pixels] 
Scene Zug 
(parametric approach) 
186 
0.07 
0.19 
Scene Rigi 
(parametric approach) 
309 
0.12 
0.09 
Scene Rigi 
(non-parametric approach) 
249 
3.57 
1.37 
Table - 1: Result of the quantitative effort for the comparison between the geocoded images and the forest 
map with the RMS for the east-west direction (i) and north-south direction (j) of the digital eleva 
tion model. The check points used for the non-parametric approach for scene Rigi are a selection 
from the entire number of points used for the parametric approach. 
No systematic deviation was found. To double-check the performance of the new parametric approach scene Rigi 
was geocoded using the improved, non-parametric rubber-sheet approach (Itten and Meyer, 1993). Table 1 (line 
3 ) presents the result for the improved rubber-sheet approach and shows the better performance of the parametric 
solution. 
The whole procedure was implemented using the IDL (Interactive Data Language, a proprietary programming 
language, Research System Inc., 1993). 
geo- 
3 - ATMOSPHERIC CORRECTION 
To utilize the data for analysis of materials on the surface the AVIRIS data have been corrected to apparent surface 
reflectance using algorithms based on the MODTRAN (Berk, 1989) radiative transfer code. This correction 
begins with the signal measured by AVIRIS. An example of the AVIRIS sensor measured signal versus channel 
number for a vegetated site is shown in Figure 5. 
Wavelength (mu) 
Figure - 5: AVIRIS Digitized Number from Swiss 1991 Data. 
Through AVIRIS calibration, these DN are calibrated to upwelling radiance per spectral channel as shown in 
isting 
ways 
real- 
1987) 
dis- 
:rage 
1