| om 
photo- 
inven- 
riment 
by 
WW a 
on an 
ip the 
ange and 
has a 
iph. 
> the 
ised on 
jequent- 
ker 
-ailed 
le 
ow that 
ant 
network 
mation. 
'erlays 
nfor- 
he same 
eatures 
| have 
of 
irst 
gh- 
ured 
by 
not 
g 
by 
and,if 
es were 
JR 
oe 
Sac i phos 
m 
not detectable as parallel black and white lines. 
The interpretability of pale and dark lines, in relation to 
their nature, proved to be poor the only distinction possible is a very 
general one, based on the continuity of the lines on the image and terrain 
roughness 
1. continuous lines 
a) black lines — —» smooth surface 
U 
b) white lines ——» rough surface ’ 
f 
highways 
ship canals 
abandoned railways 
motorway complexes 
Ü 
2. short white lines perpendicularand seawards on the 
coastline .~ breakwaters 
3. wide black lines in a cross pattern 
= airport 
Superposition of the image overlays and mapoverlays shows 
that it is impossible to do more detailed visual interpretation without 
further enhancement. 
Furthermore, about 50 $ of the lines detected on the Seasat 
image, as well pale as dark ones can not be correlated to any kind of ways 
or rows of trees. The explanation for this phenomenon has to take in conside- 
ration other types of reflectors as pin wires, and narrow ditches between 
the fields. 
In order to check whether the appearance of linear features 
on the image can be correlated to the relative direction of the features 
on the terrain, according to the flightline an analysis was made of frequency 
of lines for each orientation. Frequency was chosen rather than the length 
because : 
1. most lines appearing on the photograph have a lenght within 
a range of 0,5 to 2 cn. 
2. only very few features in reality appeared on the image as 
interrupted lines, with the same orientation 
The range direction of the satellite was chosen as the O?-180? 
axis. All other directions were divided in classes with a 10? interval. The 
result of this analysis shows a rather unequal division. However, the 
quantities detected per direction on the image must be interpreted in corre- 
lation with the division of the directions of linear features on the terrain. 
This division of directions was tested in all combinations of pale, dark and 
both types of lines on the image, in relation to all kind of linear features 
in reality. Each of the figures showed a remarkable peak for the azimuthal 
direction and the range direction of the satellite. 
However the relative difference between reality and image was 
not more than 65 $ significant. 
We therefore repeated the whole procedure for another Seasat 
image of the same area, but from adifferent date (19 aug.1978). This photo- 
251 
I. es > DE EEE