A comparison of the classification results with ground truth and air photo 
interpretation shows that there is a good agreement in the distribution of 
lakes, forests, regeneration areas and agricultural and barren land. A more 
detailed study of classification accuracy has revealed that regeneration 
areas with a dominance of young deciduous shrubs have not been separated 
from areas dominated by young coniferous stands. The reason is that we did 
not have any type object with deciduous shrubs in the experiment. The appor- 
tionment between regeneration areas with and without predominance of coni- 
ferous plants has also been disturbed by the occurance of deciduous shrubs. 
By comparing the classification results from the Landsat registration from 
1972 and 1973 one can identify visually the new clear cut areas. A more 
detailed discussion of this forest inventory project is published by Orhaug, 
Wastenson, Akersten (1976 and 1977 .). 
Land use mapping from Landsat-2 CCT^s 
A Landsat-2 scene from July 3, 1975 at Üstergótland has been utilized for 
land use mapping. Severe disturbances were found in the digital data, espe- 
cially in band 6; this channel has consequently been omitted in the data 
analysis. Table 2 shows the used classification system. 
Table 2. Classification system in land use mapping from Landsat data at 
Linkóping, Ustergotland, Sweden. 
  
. Forest Land 
31. Coniferous forest 
311. Pine 
312. Spruce 
. Mixed forest 
. Deciduous forest 
34. Regeneration area 
. Hydrography 
41. Lakes 
42. Streams 
5. Wet land 
. Rejects 
. Urban and Built up Land 
11. Town, city, commercial 
services 
12. Residential 
121. Apartment houses 
122. Villa quarter 
13. Industrial, Institutional 
and Communicational Land 
131. Industrial 
132. Communication 
133. Institutional 
. Agricultural Land 
21. Crop land 
22. Pasture and parks 
  
An area of 2 500 km“ - 7 % of a Landsat scene - has been classified using 
29 different test objects. This a encompasses 0.5 M pixels and the total 
CPU-time in IBM 360/75 was i es 
Figure 4 depicts a subregion recorded with the ink jet technique and the 
colors are representing 11 different land use classes, The left image 
is printed on the approximate scale 1:200 000. In the right image a sub- 
region is shown in a larger scale compared with ground truth. No geometric 
correction is performed in the classification examples in Figure 4, but 
in a later phase of the project this has been done and the classified land 
sification results 
T data from 1972-094 
lear cut area. 
ire. 7 = bedrock.