The emphasis of the project is on the application of Landsat 
TM data to geologic and mineral exploration in the basement 
rocks, taking advantage of their higher spectral and spatial 
resolution. Color-stretching by band decorrelation (Gillespie 
et al., 1986; Haydn, 1982) and color-coded ratio images (Gil- 
lespie et al., 1987; Sabins, 1985) provide considerably better 
image products for visual image interpretation. 
In addition, the gain in geologic information by using GIS 
technology for the integration of digital terrain models, geo- 
physical, geochemical and petrographic data in addition to 
visual interpretation and field work is being studied. The 
project is still going on and thus only some interim results 
can be presented here. One of the objectives of the study is 
mapping but only a few 1 : 250 000 sheets are planned as a 
pilot map based on remote sensing and GIS technology. 
3.1 Geologic setting 
The study area (see fig. 1) is part of the Arabian-Nubian 
shield and includes the Ariab/Oshib-Nakasib structural zone. 
It is characterized by major suture zones and related 
ophiolites (Stern et al., 1989). There is an assemblage of 
three major SW-NE-striking units, the Oshib ultramafic com- 
plex, the Ariab-Arbaat volcano-sedimentary series, and the 
Awat-Asoteriba volcanic series, all metamorphosed in the 
greenschist facies. These units are intruded by different 
generations of granitic plutons. Of special interest in this area 
are mineral deposits of massive sulfide ores and gold-bearing 
quartz-barite layers, associated with the shear zones (Cottard 
et al., 1986). 
Due to the arid climate, the rocks are well exposed through- 
out the area, if covered by desert varnish and weathering 
crusts. The relief is in the order of 1000 m or more. 
3.2 Visualization of data 
Visual interpretation of image data relies heavily on the quali- 
ty of the imagery. Therefore, visualization of information 
contained in the imagery is of paramount importance. The 
improved spectral and spatial resolution of Landsat TM data 
as compared to MSS permits the creation of significantly im- 
proved imagery, resulting in higher detail and better reliability 
of interpretation products. Especially the combination of 
different visualization techniques can provide impressive 
geologic information not available from earlier satellite data. 
3.2.1 Color contrast stretch 
The representation of the information contained in the vari- 
ous spectral bands of a TM image can be achieved in a sim- 
ple and easily interpretable way by creating a color compos- 
ite image, in which 3 bands are color-coded in red, green and 
blue, respectively. Due to the high correlation among the 
individual bands, a major percentage of the colors of such a 
three-color combination produces black resp. gray. This leads 
to a rather smudged appearance of the color image and low 
color saturation. A transformation of the red, green and blue 
colors in the so-called RGB color space into an intensity, hue, 
and saturation (IHS) space allows decorrelation of the three 
source bands and stretching of color saturation (Haydn et al., 
1982, Gillespie et al., 1986). After re-transformation into the 
RGB color space an image with much higher color saturation 
is obtained that renders more information for the interpreter. 
The usefulness of such "spectral maps" (Kaufmann & 
Schweinfurth, 1986) in geology is well established. 
328 
In the present project, spectral bands 7, 4, 1 coded R, G, B, 
respectively were selected. In this instance, this combination 
provided better results than the usual combination of bands 
7, 3, and 1. Saturation was stretched with a factor of 2 
during transformation. 
3.2.2 Combination of image data with a DEM 
In the interpretation of aerial photographs - photogeology -, 
the fact that a threedimensional image is being interpreted 
greatly improves the process of information extraction. With 
the exception of stereo SPOT imagery with tilted look direc- 
tions, height information from satellite stereoscopic imagery 
is generally unsatisfactory due to the unfavorable ratio of 
stereo base to sensor altitude. A combination of the satellite 
image with a digital elevation model (DEM) can partly offset 
this disadvantage. 
To this end, existing topographic maps at a scale of 
1 : 100 000 with 25 m contour lines were digitized, and a 
DEM with a cell size of 30 by 30 m, corresponding to the 
pixel size of the Landsat TM image, was generated under 
ARC/INFO. The georeferenced TM image, color-stretched by 
the process described above, was then draped over the DEM 
using the Terra-Mar Microlmage processing system, resulting 
in a rather graphic representation of the landscape and the 
spectral reflectance properties of the individual lithologic 
units, as shown in fig. 4 on the color plate. 
3.3.3 Color-ratio imagery 
As mentioned, the Ariab district contains mineral deposits of 
polymetallic sulfide ores, as well as several gold occurrences 
in quartz-barite layers, some of which are economically inter- 
esting (e.g. the Hassai gold mine). 
For a subset of Landsat TM scene 172-047, acquisition date 
January 1, 1989, a color ratio-image was created for en- 
hancement of gossans associated with the sulfide bodies. In 
a modification of a ratio combination found useful by Sultan 
et al. (1987) in similar geologic settings, band ratios 5/7, 
5/1, and the product of 5/4 x 5/3 were used and coded with 
red, green and blue. Subsequently, the three bands were 
transformed into the IHS color space where saturation was 
stretched by a factor of 2. For intensity, the original band 7 
was substituted and the resulting image transformed back 
into RGB space. In this combination, gossans are clearly 
visible, showing a distinctive red color (fig. 5, on the color 
plate). 
Comparing the resulting map to occurrences known from 
field studies executed by the French BRGM, it turned out 
that all gossans identified in the field were also highlighted 
on the ratio image. In addition to the already known occur- 
rences, a number of new ones was detected on the image 
and subsequently verified in the field. 
3.3.4 Visualization of geophysical data 
Geophysical data of the study area are available in the form 
of small-scale maps. Even if these data are not detailed 
enough to significantly contribute to the interpretation of the 
geology at larger scales as shown in figs. 4 and 7, they are 
very valuable for the 1 : 250 000 scale geological maps the 
production of which is planned. 
The existing geophysical data are Bouguer gravity maps and 
aeromagnetic maps at a scale of 1 : 500 000. Geocoded 3D 
surfaces were created from these maps by digitizing the 
isolines and generating 3D models.