ES ms = * 
mM E is 
graph was taken with another flightline orientation. In spite of another 
Seasonal condition, most of the antropogenic linear features have simular 
reflection characteristics, so that comparison with the same groundtruth 
information was possible. 
Here again the interpretability of linear features has the same 
level as the other image and once again, a striking difference can be seen 
between the number of lines detected on the image for each orientation-class 
and the division of linear features for each orientation in reality. 
  
  
   
  
  
  
10% 
Fig. 
AZIMUTH 
40 % 5% 5% 10 9, 
fig. 1 : relative division of orientation of black and white lines on the 
image according to the occurence of linear features in reality. 
Significance of the difference is not more than 65 % for bo 
Seasat SAR 8 oct. 1978. 
feature 
Occurei 
The azimuth and range peak levels are so high that a 99 $ significant diffe- SAR imac 
rence between image and terrain was found. lines. 
highly c 
Furthermore 50 $ of the linear features could not be correlated it is in 
to real features other than pin wires and ditches. 
Superposition of overlays of bothimages shows a clear difference nal sele 
in the pattern of reflected lines. The overlapping lines are less than 5 $ 
of the total in number. Exception must be made for highways and shipcanals 
which are easily detectable on both images. However the low percentage of related 
overlapping shows clearly the low efficiency of visual interpretation of 
the pher 
linear features without further enhancement. 
accordir 
Another assessment concerning the tonality of the lines on the 
image is that some diffuse reflectors show different reflection characteristics 
252