— 661 — 
The main pattern of oblique "upglacier" crevasses has been separated in photo 8 
by way of elimination of the other direction, as can be seen from the spectrum 
of fig. 7b. It now appears that to the right of the snow patch and apart from the 
eliminated vertical crevasses, a system of crevasses also occurs running from 
top right to the lower left. This very localized feature is associated with the rock 
step, the existence of which is thus more clearly brought to light. 
On photo 9 the dominant, oblique, crevasse direction has been eliminated as well 
as possible. This could not be effected completely with the filters at hand and 
some oblique crevasses remain vaguely visible in the image. Nevertheless it is 
obvious that the influence of the left-right stress, as visualized by the vertical 
crevasses, reaches much further to the right than would be expected from the 
non-filtered crevasse pattern of photo 6. Particularly in the central part of photo 
9 numerous vertical crevasse elements can be seen. The dominance of the oblique 
crevasses in the non-filtered image is apparently such that the less prominant 
vertical pattern is largely suppressed. The Fourier spectrum corresponding to 
photo 9 is given in photo 7c. 
It will be evident that the filtered images obtained with this technique can be 
analyzed in turn using density slicing techniques, etc. A more simple approach 
to quantification has also been attempted by the author in this case. Both the 
number of fractures per cm-image and the total length of fractures per sqcm.- 
image have been determined for the entire width of the image and are presented 
graphically in fig. 2. The snow covered zone was left out of consideration for 
obvious reasons. It is obvious from the graph that the oblique fractures are 
strongly dominant except in the extreme left whereas vertical crevasses are ab 
sent in the right but gain rapidly in importance towards the left. 
CONCLUSIONS 
1. Image analysis techniques are equally important for the study of terrain as 
image recording techniques. Particularly interesting is the quantitative infor 
mation that can be obtained and statistically evaluated. 
2. Quantitative relief analysis is particularly valuable since it is directly linked 
with terrain form. Stereoscopic study of aerial photographs, stereo-SLAR, and 
topographic profiling with orthophotography (dropped lines), radar and laser 
altimeters, etc. merit full attention in the study of the terrain. 
3. Quantitative density analysis is generally less reliable since the measurements 
are related to cover type (vegetation, crops, etc.) and thus are only indirectly 
linked to terr a in form. Also, in most cases it is not possible to use the image