155
h histogram
W rÛL
icy histogram
R-B not on LFC
h histogram
rHftufliïT
icy histogram
MJ
h
"C not on SIR
tents:
present on LFC
ent on SIR-B.
tr to it overre-
Sudan, the sun
îun elevation of
in both cases
>nt: 50 E from N
■B and 40 E from
illumination is
for sunlight on
he radar energy
E lineaments on
d, volume scat-
1 be less than
parameters four
;re made:
ited from SIR-B
:rom SIR-B image
ireted from LFC
from LFC photo
jroups (fig. 9)
rations : a broad
row peak around
c directions are
classes).
Table 3. Results of the x 2 test for four groups of
directional classes.
I
II
III
IV
I
-
25.5*
96.3
124.2
II
25.5*
-
107.8
110.6
III
96.3
107.8
-
30.6*
IV
124.4
110.6
30.6*
-
df : 17 (degrees of freedom)
* Ho: Distribution similar, differences due to
random errors.
To determine if significant differences occur
between the groups of directional classes, a x 2 test
is used (table 3).
As can be expected, comparison of groups I and II
(both from radar interpretation) gives a small
variance of 25.5.
We can accept the hypothesis (Ho) that no signi
ficant differences occur between the samples and
that the differences are due to random errors. The
same is true for Groups III and IV (both from LFC
interpretation). Comparisons of Groups I and III, I
and IV, and II and III give rather high values, so
the null hypothesis has to be rejected and the
samples have to be considered as different.
Considering the contribution of the different
directional classes to the total variances it can
be seen (table 4) that midclass directions 310° and
320° and midclass directions 50° and 60° cause the
principal differences. For comparison of Group I
and III (all radar lineaments and all LFC lin
eaments) , the directional classes 310°-320° of the
observed frequencies for radar interpretation are
considerably higher than the expected frequencies
according to the null hypothesis, whereas for the
LFC photo-interpretation they are lower. For direc
tional classes 50°-60°, the reverse is true; here
the observed frequencies on radar are much lower
than the expected frequencies according to null
hypothesis.
Apparently the enhancement of lineaments oriented
perpendicular to the radar look direction causes
the overrepresentation. The direction perpendicular
to the sun azimuth for the LFC photo apparently is
of no influence, because this direction is under
represented. Radar look direction (50°) is of nega
tive influence in lineament detectabilty as was
expected, but sun azimuth direction for LFC seems to
have little influence.
Comparing Groups II and III (only radar lineaments
not present on LFC and all LFC lineaments), the con
tribution to the variance is still even larger for
the directional classes parallel and perpendicular
to the radar look direction. In directional class
50° (radar look direction) , not a single dike lin
eament has been observed on radar which was not also
visible on the LFC against 30 lineaments on the LFC
image not interpreted from radar.
Considering the LFC lineaments not present on
radar in comparison with the distribution of all
radar lineaments (Groups IV and I), we again see the
negative influence of radar look direction on lin
eament detectibility. In direction 310°-320° and
340°, an overrepresentation of lineaments is found
on the radar image; 310°-320° is partly due to the
radar azimuth orientation, but the 340° direction
may be a result of penetration through superficial
sand cover.
CONCLUSIONS
In the Iranian case, the differences in length re
cognition of the strike ridges interpreted from
radar and from hand-held camera images are largest.
Nearly double the length (28 km) can be seen on
Table 4. Contribution to variance for different
directional classes. Midclass direction:
Group I/Group III Group II/Group III
w
270
.12
.11
W
1.80
.66
280
.00
.00
.36
.13
290
1.27
1.15
3.20
1.22
300
2.72
2.47
.25
.09
310
6.34+
5.75-
18.23+
6.73-
320
8.62+
1
M
00
r-
21.24+
7.84-
330
.74
.67
5.07
1.87
340
1.13
1.03
.00
.00
350
1.71
1.55
1.33
.49
N
0
1.19
1.08
N
.30
.11
10
.75
.68
1.58
.58
20
.34
.31
.07
.03
30
.64
.58
2.70
.99
40
2.55
2.31
4.62
1.71
50
13.05-
11.84+
12.94-
4.78+
60
7.56-
6.86+
4.34
1.60
70
1.76
1.60
.31
. 11
80
.00
.00
.30
.11
E
E
Total
= 96.3
Total =
107.8
Group II/Group IV
Group I/Group IV
W
1.27
1.04
W
.09
.18
.02
.02
.09
.19
.72
.59
.08
.16
.14
.11
.99
2.01
12.13+
9.96-
4.10+
8.28-
9.55
7.85
3.05+
6.17-
2.13
1.75
.17
.34
2.29
1.88
3.45
6.96-
.01
.01
.01
.02
N
1.69
1.39
N
.03
.07
.36
.30
.03
.05
.03
.03
.00
.00
1.35
1.11
.02
.05
4.03
3.31
1.81
3.66
13.53-
11.11
11.93-
24.11+
11.25-
9.24+
14.82-
29.93+
.11
.09
.02
.03
.09
.08
.45
.91
E
E
Total
= 110.6
Total
= 124.2
+ Observed frequency considerably higher than
expected frequency according to null hypothesis.
- Observed frequency considerably lower than
expected frequency according to null hypothesis.
radar before the strike ridges seem to disappear
under the superficial sand cover. The ground resolu
tion of the SIR-A image is much better than that of
the hand-held shuttle photography. Unfortunately, no
more detailed images or field observations have been
obtained over this area, so that no conclusive de
duction could be made regarding possible penetration
of microwaves through superficial sand cover.
In the arid area of the eastern desert, Egypt/Red
Sea hills, Sudan, the study of length and frequency
of dike swarms in sandy areas did not conclusively
reveal penetration capability of microwaves through
aeolian sands to reveal lightly covered dike struc
tures. In most cases, frequency and cumulative
length of interpreted dikes from the SIR-B images
was smaller than that interpreted from LFC photo
graphy. Both types of images have approximately the
same order of spatial resolution (approximately 20
m). The tonal range of the LFC photos is larger and,
moreover, they can be studied in stereo, giving them
a slight advantage for interpretation. Only in area
3 was a higher frequency of radar lineaments found.
Close visual comparison of the two interpretations
gives the impression that penetration of microwaves
through superficial sand cover has resulted in a
larger number of detectable dikes on the radar
image.
