RE A 
field 
1 the 
Lon 
rning 
Espe- 
pesol- 
max- 
here. 
  
f edges 
+ white) 
  
6 and 
  
that parcelling pattern and different forest types may be analysed. The ratio 
between bands 6 and 7 (photo 8) brings out lakes and büilt-ap areas precisely, 
and especially areas of grassland in bright tones. 
Field parcelling in North-Germany is important, because it reflects natural con- 
ditions, to which man has to adapt (MÜHLFELD, 1976). Two main types of parcel- 
ling are encountered in the Neumünster area: 
1) Large size, irregular shaped and irregulary arranged forms, which can be ex- 
plained by the hummocky young Pleistocene relief and rapid changes in soil 
quality, which prevented the arrangement of field boundaries in a planned manner. 
Soil types found there are generally good, because they originate from nutrient- 
rich Pleistocene marl or loam. 
2) Medium sized, rectangular shaped and regularily arranged forms which are ex- 
plained by level to undulating glacial outwash with more gradual changes in 
soil quality and the possibility to arrange field boundaries in a planned 
manner. Soil types are generally moderate to poor because of coarse parent 
material, but field management is notably easier than for 1). 
6. PROCESSING FOR SPECTRAL PATTERN CLASSIFICATION 
To further enhance the spectral pattern in the images, supervised classifica- 
  
tion according to the maximum-likelihood method was applied. The results of 
classification are useful for an improved interpretation of landuse classes and 
combinations, which themselves are good indicators of soil condition at LANDSAT 
scale. 
Spectral analysis supposes, that land use types 
can be separated well enough by theirspectral 
signature at the date of scene acquisition. With 
training areas selected according to known land 
use and homogeneous spectral response, computer 
calculations will result in a cluster in the four- 
dimensional feature space. In the following clas- 
sification of the whole image, the membership of 
each pixel to a training class is being calcula- 
ted according to the maximum likelihood. 
A two dimensional picture of these clusters is 
useful to check the separabilty of the original 
training set, in which all possible pair combi- 
nations (six) of four channels are being compared. 
It can be shown, that as a rule, channels 5 and T 
(K2 and Kh on the diagram) show best separability 
(photo 9). 
In interpreting the classification results for 
soil survey, the following general assumptions can 
  
be taken as valid for the area Neumünster. Loamy Photo 9: Cluster diagrams, 
soils of the young pleistocene moraines are indi- classification 28. 8. T5, 
cated by the presence of broadleaf forest, the (color) 
presence of rape seed, cereals like wheat, and 
by the presence of root crops like sugar beets, all of them crops grown on 
soils of higher nutrient and water availability. Soils formed on young loamy 
glacial drift under North-German climatic conditions are mainly Pseudogleys, 
Lessivés and rich Brown Earths, soils which have high remission in all spectral 
bands, when encountered bare. 
Soils formed on glacial outwash on the other hand are poor in nutrient and 
water capacity and form Podsols and poor Brown Earths, and when bare, show 
frequently low spectral response due to organic matter of higher content and 
lower quality. Crops grown on them are mainly cereals like barley, rye, and 
corn. In addition, various types of coniferous forest are found, together with 
large gravel and sand pits, which take advantage of the sorting in glacial out- 
401 
a ÓÀ—— M — o D D IER LERRA RE EE RR RR E