An example of a 3D model of the aeromagnetic AT anoma- 
lies is shown in fig 6. Block size corresponds to the 
1: 250 000 sheet Port Sudan, 1° x 1°30’; look direction is 
toward NE. The low AT values corresponding to a major SW- 
NE striking suture line, separating metavolcanics and meta- 
sediments in the north from volcanic rocks in the south, are 
clearly visible. The Wadi Amur area, shown in detail in fig. 8, 
is situated in the southwestern corner of the block. All in all, 
the geophysical data provide valuable information for small 
scale geologic studies, like for the planned 1 : 250 000 scale 
geological maps. Due to the coarse sampling grid they are 
less suited for information extraction at larger scales. 
  
  
        
       
    
           
  
   
  
              
  
    
      
S = \W> = EZ j S = = 
a zZ? m S. < 
: = DS == 
         
> 
  
  
  
Fig. 6: 3D model of aeromagnetic AT anomalies of the area 
W of Port Sudan. Block size is 1° x 1° 30°. 
3.3.5 Geochemical and mineralogical data 
Samples for geochemical analysis were taken along a 25 km 
cross-section in the Ariab-Arbaat volcano-sedimentary series, 
using a 500 m grid. Several element combinations were 
plotted against the geology, showing good correlation of 
geology and geochemistry. More data will be needed in order 
to construct a meaningful picture of element distribution 
since the interpolation between the sample points in the GIS 
leads to incorrect values . 
Thin-section petrology of rocks is also recorded in the GIS. It 
is essential, among other things, for judging abnormal values 
in the geochemistry and for selecting additional sampling 
points for ensuing field work. 
3.3.6 Additional information sources 
Texture plays an important role in visual interpretation of 
geologic features. Digital texture classification is also viable; 
however, due to inherent problems (window size versus 
boundary definition) the results are better suited to classifica- 
tion of larger areas. Small features, even with pronounced 
textures, are suppressed. Texture analysis was performed on 
a TM band 4 image by means of a 9 x 9 variance filter on 
the Terra-Mar Microlmage system. The resulting image was 
median-filtered and interpreted visually (fig. 7). The major 
lithologic units correspond to the ones seen in the geologic 
interpretation map (see fig. 8). 
Interpretation of aerial photographs is being used in conjunc- 
tion with the interpretation of satellite data. The scale of the 
photographs is in the order of 1 : 70 000. There are some 
areas close to the Red Sea coast that are blanketed by 
clouds on the TM image. Here the aerial photographs provide 
a means of extending the interpretation from the cloud-free 
parts of the TM image into the cloud-covered region. Since 
the b/w photographs contain very little spectral information, 
cloud-free Landsat MSS data were used for additional infor- 
mation and for interfacing the photographic interpretation to 
the rock units discriminated on the satellite imagery. 
Due to the central perspective of the aerial photographs, the 
interpretation data obtained from them are subject to radial 
distortion and changes in scale caused by relief. For trans- 
forming these data into a proper geocoded format, a photo- 
grammetric 3rd-order stereoplotter (ZEISS Stereotop) was 
used. Control points were taken from existing topographic 
maps at a scale of 1: 100 000. In this way, all lines were 
transformed to map projection and subsequently digitized for 
merging with the other geocoded data. 
  
  
E Very coarse Smooth 
[MM Coarse Very smooth 
Texture 
EX Mixed (granite) 
  
  
  
Fig. 7: Interpretation of a texture classification of TM band 4 
data of the Wadi Amur area; block size is 30 x 30 km?. 
The TM data were also subjected to principal component 
transformation. The resulting imagery was found to convey 
less information to the interpreter than the ratio imagery 
described above. Therefore, PCA imagery was not further 
used for interpretation. 
Not surprisingly, similar results were obtained from digital 
classification by the maximum likelihood algorithm. In the 
end, the classified imagery provided less information than 
what was obtained from visual interpretation. This, of 
course, holds true only with respect to the discrimination of 
regional geologic features, like the subdivision of different 
types of metamorphic rocks or intrusives. When it comes to 
the detection of local spectral anomalies, like gossans or 
hydrothermally altered outcrops, digital classification does 
indeed point out such anomalies in a reliable and consistent 
way. 
During the progress of the work, the results of detailed geo- 
logic field surveys carried out by Sudanese geologists prior to 
our own field work became available. This valuable informa- 
tion, complementing the data collected during the limited 
time spent in the field by the authors, will be used to 
strengthen the existing knowledge base for the final map. 
330 
  
Fit