sity unit at least soil temperatures have to be measured. But instead 
of temperature measurements soil samples can be taken to get the water 
contents directly and for a soil profile of | m and more. These point 
measurements than are extrapolated to the corresponding map units as 
it is done with the results of remote sensing. From this point of view 
it seems not economic to calculate soil moisture from thermal scanner 
data. 
Temperature gradients of the heating or cooling phase, caused by ther- 
mal inertia, seems not to be useful for soil moisture calculations in 
which planers and engineers are interested. But temperature gradients 
and temperatures during night time offer good possibilities for auto- 
matic soil mapping. The computer derived "thermal map" must indicate night 
temperature and gradient for each mapped unit. The units have to be 
verified by terrain control. 
Temperatures and temperature gradients have to be calculated with a 
relative accuracy of at least O.l. 
The described test areas showed that a geometric resolution of 30 m x 
30 m is desirable. 
Acknowledgements 
Density slicing and further data processing were carried out at the 
Dibias System of DFVLR. In handling the system I was supported by Mr. 
Fernandez (Zentralstelle fiir Geo-Photogrammetrie und Fernerkundung, 
Munich) and Mr. Fiedler and Mr. Helbig (both DFVLR, Oberpfaffenhofen). 
Especially I have to thank Dr. Quiel and Mr. Wiesel, Inst. f. Photo- 
grammetrie und Topographie of the TU Karlsruhe, who carried out the 
rectification of the scanner data.