up to about 4 m above the surroundings and is drained by numerous small 
creeks. The zoning of the gradiants (3,9 °C/hr external; 4,5 °C/hr cen- 
tral) corresponds to the ascendent morphology. 
The central part of this area is covered by only a thin swampy layer 
under which we find very fine fluvial sediments. This area seems to 
be a sand bank of the former lake. Soil samples taken nearby indicated 
45 % water contents for the upper 0,2 m. 
Automatic soil mapping 
Based on the results described above, which were achieved manually, 
automatic mapping was tried. The scanner data were rectified in refe- 
rence to an aerial photograph before data processing started corres- 
ponding to fig. 9. 
The image in fig. 1l was calculated as the ratio temperature (before 
sunrise)/heating gradient. Light greytones result from high morning 
temperatures and low gradients, dark grey tones from low morning tem- 
peratures and high gradients. 
Comparing fig. 10 and fig. 11, on one hand we find a good correspondence 
on the other hand there is no indication for soil unit C of fig. 10 
This leads to the conclusion that morning temperatures indicate units 
of similar soils, the temperature gradiants allow to subdivide these 
units. That means automatic data evaluation by the method described 
above gives small scale variations of soils. Only if the morning tem- 
perature, too, is taken into consideration a map of thermal properties 
can be derived. The kind of soil has to be verified by terrain control. 
From automatic mapping always arises the question of confidence in the 
derived borders. Calculations as described above arrive at very low 
values. Units must be distinguished on the basis of relative differences 
of the range of 0.2. That means the data processing system must be able 
to calculate temperatures, gradients and ratios at least with a relative 
accuracy of 0,1.