. 
'I'able 2. Effectiveness of single band imagery in extracting semi arid geomcrphology. 
SCORE USED IN ASSESSING EFFECTIVE EXTRACTION OF SUBJECT:- 
Good 12, Moderate 8, Poor 4, Not Visible 0. 
IMAGE PROCESSING TASKS Single bands and FCC (linear stretch) 
GEOMOFPHIC PHENOMENA 
Geological structure 
Lithol. discrimination 
Solid-drift boundary 
Fan entrenchment 
Anastomosing fans 
Occurence of croutes 
Dep. from playa 
Fan-playa boundary 
Variation within the playa 
Alluvium 
River channels 
Eio-geomorphic 
1 
2 
3 
4 
5 
7 
123 
345 
457 
8 
8 
12 
12 
12 
12 
12 
12 
12 
0 
4 
8 
12 
12 
12 
4 
12 
12 
0 
12 
12 
12 
8 
8 
4 
12 
12 
4 
4 
8 
8 
12 
12 
4 
12 
12 
4 
8 
8 
4 
12 
12 
8 
8 
8 
8 
8 
8 
4 
4 
8 
8 
4 
12 
4 
4 
8 
4 
8 
12 
12 
8 
8 
4 
8 
8 
12 
8 
12 
8 
8 
12 
4 
8 
8 
12 
12 
12 
4 
12 
12 
4 
12 
12 
8 
4 
4 
8 
8 
8 
8 
8 
8 
8 
12 
12 
8 
12 
8 
12 
12 
2 
4 
8 
8 
8 
12 
12 
It is apparent from the table that some TM bands 
contain more information for specific geomorpholog- 
ical phenomenathan others, especially by bands 2 anu 3 
picking up alluvium while the infra-red bands are 
better for lithological discrimination. The striking 
point to note is that some bands score very similar 
for the same feature, a fact mirrored by the FCCs. 
This suggests that the amount of geomorphological 
information held in TM bands is very similar for 
certain features and little extra information is 
contained in FFC for the same phenomena. Visual ex 
amination of the images suggests that the bands show 
a marked degreeintercorrelation and this is shown in 
Table 3. Strong correlations^ are displayed between 
all bands with even stronger relationships among the 
visible bands, the middle infra-red bands and band 4 
and the other bands. It is well known that the TM 
was designed primarily for vegetation discrimination 
with bands selected to take advantage of the spectral 
response of vegetation (Salomonson er al. 1980). The 
implication from this study is that for investigations 
of bare, arid (ie non-vegetated) surfaces, after 
studying an infra-red or an FCC, additional single TM 
bands are of little value unless very specific ranges 
of pixel values are utilised. 
Table 3. Correlation matrix derived for six reflect 
ive TM bands for the El Guettar test area. 
and as a result it is not possible to show an example 
in this paper. A basic but powerful image processing 
technique is the ratioing of two spectral bands. 
This reduces topographic noise and enhances subtle 
spectral differences of surface features. However, to 
effectively use such methods, the relationships bet 
ween the surface material and its spectral response 
must be understood. Hunt & Salisbury (1976) and 
Kahle (1982) have shown that up to a wavelength of 
2.5um, there is an increase in absorption of reflect 
ed energy by clay particles. Thus a ratio of TM band 
5 (1.55-1.75um) and band 7 (2.08-2.35um) will display 
the occurrence of clay particles as lighter tones. 
In the 5/7 ratio (Fig. 7a), the light area in the cen 
tre of the chott corresponds with the unvegetated part 
of the playa where clay particles are expected. It 
is interesting to note that the high concentration of 
clays on the southern alluvial fans and on the south 
western part of the image which is subjected to aeol- 
ian deposition from the playa. 
1 
2 
3 
4 
1 
1.000 
2 
.963 
1 .OOO 
3 
.948 
.983 
1.000 
4 
.891 
.937 
.953 
1.000 
5 
.807 
.880 
.911 
.918 
7 
.812 
.885 
.910 
.905 
1.000 
.974 1.000 
An effective processing technique for single band 
imagery is edge enhancement or convolved images. 
This allows the user to highlight a boundary or edge 
in the image by smoothing out other detail. Edge 
enhancement operates by passing a digital filter or 
kernal, in the form of a matrix, over the data and 
has been very successful in many studies of geologic 
al lineaments (Bailey et al. 1978). Figure 6 shows 
the result of passing two different filters, a north 
(Fig. 6a) and an east (Fig. 6b), over the image. 
Note how the north filter picks out the mountain-fan- 
playa boundaries which is not as evident in the other 
filtered image. 
The simplest multiband processing technique is the 
production of false-colour composites using single 
band imagery and as stated previously, they contain 
very little additional information. This can be im 
proved by an Intensity, Hue, Saturation transform 
which is simply a rotation in colour space. The eff 
ectiveness of this processing is best shown in colour