546 -
rests of a basaltic lava flow within a sand plain. For the in
terpretation and recognition of an unknown lithological unit we
need much, larger areas where grey tone and texture can be de
termined correctly.
For instance, the sandstone complex between Bardai
and Aozou (Northern Tibesti) could be subdivided into four
different lithological units for the first time exclusively by
interpretation of aerial photographs (ROLAND 1973), but the
same differentiation could not be made on the ERTS-1 images.
The stratigraphic classification, on the other hand, had to re
ly on fossil discoveries in the field (ROLAND 1971).
Grey Tone, Multispectral Characteristics, Texture
Since the satellite image shows very large areas un
der uniform "real time" conditions in contrast to an aerial
photographic mosaic of the same region, the observation and
measurement of the grey tone is much more important than in the
interpretation of aerial photographs.
For the interpretation of arid regions, grey tone
seems to be the most important parameter in the evaluation of
satellite images. But similar to geological field work, where
rock color alone is not sufficient for the identification of a
rock, the grey tone in the satellite image alone does not allow
us to draw unequivocal conclusions on the kind of rock in ques
tion. Visual interpretation is also rendered more difficult by
the fact that areas of equal grey tone (corresponding to a cer
tain rock unit) are often small, and differences in grey tone
minute.
Here differences in the spectral remissivity of rock
surfaces are of help, especially where soft rocks are concerned.
Fig. 1 shows a sand plain in the eastern Tibesti, where some
sands show up much darker on band 7 (0.8-1.1 pm) than for in
stance on band 5 (0.7-0.8 pm). In other cases, the opposite
effect can be observed. What reasons beside different humidity
contents are responsible for this effect will have to be
cleared up by laboratory investigations.
Since we cannot get a stereoscopic view of most satel
lite images as yet, texture becomes a very important factor in
the interpretation of orbital images. Texture depends on scale
(CARTER & STONE 1974); many details that can be observed as
separate "signals" in an aerial photograph, are squeezed toge
ther into an often characteristic texture in the small scale
satellite image. Quantitative analysis of texture patterns will
probably be of tremendous help in the identification of rocks
from satellite images. Somewhat simplified it could be said
that detectability depends mostly on differences in grey tone
(contrast !), and recognizability of differences in texture.
Fig. 2 shows some typical examples of textures of
different rock units in the Tibesti.