Figure 2. The PELCOM base map (only a black-and-white 
copy without legend is presented here). 
  
  
  
  
  
Figure 3. Example of a NOAA-AVHRR NDVI composite: 
maximum value of July/August 1997 (only a black-and- 
white copy without legend is presented here). 
The DISCover data base offers an overview of the 
distribution of the major land cover classes at the global 
scale. However, when zooming in on Russia it must be 
concluded that the classification results are not very 
accurate. This was also concluded when comparing the 
DISCover data base with CORINE results for other regions 
in Europe (Veldkamp et al., 1998). 
Sull, a multitemporal classification of NOAA-AVHRR 
images may yield the land cover classes water, built-up 
areas, forest, grassland and arable land. Mücher et al. 
(1994) showed that such results can be quite satisfactory in 
a statistical sense, but also when looking at the spatial 
distribution of the land cover classes. Figure 3 shows an 
example of a NOAA-NDVI image of Europe, indicating the 
spatial variation of the information present in such an 
image. 
5.2 Land Cover Classification at Level 3 (Regional) 
At the regional level (level 3) some land cover 
classification experiments were performed. During the first 
experiment (Clevers et al., 1996) two images covering the 
test site in the Kursk region were used: 
- multi-band SPOT image of 26.07.91, and 
- multi-band MSU-E image of 25.09.91. 
Additional information was used during the image 
processing, including: 
- cartographic materials for the test territory; 
: ground truth on crops identification obtained in 
the period of the ground survey in 1991; 
- some a priori information; for example, crop 
calendars for the test site. 
Combination of these two images was used for a land cover 
classification, showing that MSU-E data are 
complementary to SPOT-XS data (Clevers et al., 1998). It 
was found that the use of multitemporal images in the 
classification increases the separability of classes in 
comparison with the classification using only an image of 
one date. Figure 4 presents the final classification result. 
  
  
  
  
  
Figure 4 Result of the supervised classification for the 
Kursk test site, using information from both the SPOT and 
MSU-E images (only a black-and-white copy without 
legend is presented here). 
100 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
out 
Da 
pro 
Scc 
of 
sup 
exp 
wit 
are 
eve 
sea 
the 
im] 
Th 
inf( 
the 
SP! 
TK 
' per 
im: 
TK 
of 
mu 
wit 
Tal 
umc 
cla 
fall 
WII 
(wl 
Wir 
har 
Wir 
(in 
buc 
bar 
spr 
sur 
oat 
nat 
gra 
WO 
tot: