680 
Figure 2. An outline of geology 
4 DEPTH ESTIMATION OF THE SEDIMENTARY ROCKS 
4.1 Hypothesis for estimation 
Basement rocks mainly are affected by mechanical ero 
sion agency under arid and semi arid climate. This 
hypothesis was presended by Bildel (1957) as the con 
ception of "double surface of leveling" (see figure 4), 
According to his conception the level of the sediment 
surface is changed by flood and climatic condition, 
(a) Wash plain of seasonal flooding is as much as 100 
m above the weathering front. Pediments fringe the 
wash plain, (b) Wash plain is lowered by rejuvenation 
or climatic change. Inselberg and marginal pediments 
are exhumed or regraded to the lowered wash plain. 
This geomorphological process indicate that the wash 
plain being more extensive area or longer distance 
among inselbergs, have relatively deeper sedimentary 
deposits. Using this hypothesis, depth estimation 
was done with Landsat imagery data. 
rocks. 
sity 
depth 
30° S 
Lower surface: basal surface of deep weathering 
(a) 
Recessional or rim inselberg 
Figure 3* Process of evaluation 
3 METHOD 
3.1 Data 
Landsat MSS (CCT) path 115 row 8l 
DATE 6th February 1973 
3.2 Process of evaluation 
After several trial, depth estimation of sedimentary 
rocks, land coverage, ratio images are employed for 
the evaluation. Process of study is shown on the 
figure 3- 
Figure 4. Schematic explanation of hypothesis of 
"double surface leveling" 
4.2 Depth estimation map 
The map is prepared with the following processes. 
(1) Basement rocks distribution 
Because of the heterogeneous nature of the study area, 
the unsupervised classification method is used. The 
analytical process involved the clustering method 
using an algorithum to divide the four-channel Land 
sat data into groups of points having homogeneous 
spectral characteristics. 
Coverage type was classified into sixteen items by 
the clustering. Then, three of them are interpreted 
that basement rocks are comming on or near the ground 
surface, by means of the ground survey and aerial 
photo-analysis. This pixel map is called "Basement 
rock distribution map". 
(2) Density of basement rocks 
Number of pixel classified to basement rock were 
counted in each mesh unit, 1 Km x lKm. The maximum 
number of it is 100, and minimum is 0. The original 
distribution map is shown on Fig. 5.a. The result 
was, however, too noisy to understand the trend of 
its distribution. Then, smoothing technique was em 
ployed to solve this problem. The original data was 
replaced by the average number of basement rock pixels 
of the surrounding mesh units. The smoothing window 
size was examined by 5 Km x 5Km, 7 Km x 7Km, and 9 Km 
x 9 Km. In this case window size, 9 Km x 9 Km was the 
most reliable for the purpouse (Fig. 5.b). 
(b) 
B 
B 
0 
(3) Depth estimation 
Basically trend of basement rock density must be 
Figure