FIG 1 BLOCK DIAGRAM OF FLOW 
2.3 Topographycal Analysis 
(1) DTM data preparation Digital terrain model or 
DTM data is converted from a 1:10,000 topographic 
map . At first a color drum scanner system was used 
to read map-images and A/D conversion and input data 
to the computer. Second,image processing made 
X,Y,Z vector data from a luster image map, and the 
procedure is line-sharpness, data-normalize, 
luster/vector conversion, addition of attribute 
(example;altitude data).Third was arrangement of 
grid point data at intervals of 25m X 25m. 
(2) Subject Map Preparation Using DIM data,several 
kinds of subject maps were prepared by topographical 
analysis,as follows. 
Altitude contour map 
Summit level map 
River level map 
Difference between summit and river level map 
Erosion situation map (Remaining area) 
Erosion situation map (Lost area) 
Undulation map 
Slope direction map 
Slope gradient map 
Unevenness map 
Table-1 Training data for classification 
of wethering degree of granite 
Class A 
Perfect-Wetherd 
Class B 
A little Wetherd 
Class C 
NoWhetberd 
CH 
Item Unit 
mean 
st .dv 
meao 
st.dv 
mean 
st.dv 
1 
Erosion 
(lost) 
m 
45.2 
20.5 
55.1 
22.9 
60.8 
23.2 
2 
Erosion 
(remain) 
m 
31.4 
20.5 
34.9 
21.2 
42.2 
30.8 
3 
Undu 
lation 
m 
6.0 
3.0 
7.3 
3.5 
9.1 
4.2 
4 
Slope 
gradient 
tan 
0.300 
0.152 
0.366 
0.176 
0.446 
0.214 
5 
Uneven. 
+/convex 
m 
-0.1 
2.7 
-0.2 
3.2 
-0.3 
3.7 
6 
Direct. 
( + E,-w) 
tan 
-0.004 
0.242 
-0.002 
0.312 
-0.076 
0.318 
7 
Direct. 
(+N.-S) 
tan 
-0.006 
0.252 
0.074 
0.274 
0.114 
0.382 
8 
LANDSAT 
PC A (1st) 
PCA 
1ST 
153.8 
23.7 
156.3 
24.5 
158.7 
24.8 
Tabi 
e~2 Training 
data for 
analys 
is of degree of 1 
and slide danger 
Case 1 
Convex Area 
Case 2 
Concave Area 
Case 3 
Al 1 Area 
CH 
Item Unit 
mean 
st. dv 
mean st.dv 
mean st.dv 
1 
Erosion 
(lost) 
m 
60.6 
22.5 
42.5 
22.1 
50.0 
23.9 
2 
Erosion 
(remain) 
m 
46.6 
22.5 
65.8 
23.9 
57.0 
24.8 
3 
Undu 
lation 
m 
8.4 
3.1 
8.4 
3.2 
8.4 
3.2 
4 
Slope 
gradient 
tan 
0.412 
0.164 
0.408 
0.168 
0.410 
0.168 
5 
Uneven. 
+/convex 
m 
-3.2 
2.1 
+ 3.3 
2.0 
+ 0.5 
3.6 
6 
Direct. 
( + E,-w) 
tan 
0.118 
0.330 
0.128 
0.316 
0.124 
0.324 
7 
Direct. 
(+N.-S) 
tan 
0.172 
0.246 
0.164 
0.260 
0.168 
0.254 
8 
LANDSAT 
PCA(lst) 
PCA 
1ST 
142.6 
22.8 
140.3 
21.5 
141.7 
25.8 
2.4 Photo-Interpretation 
The land-slide areas were extracted by photo-inter 
pretation with 1:10,000 color and 1:12,500 black & 
white aero-photos. The result was plotted on 
1:10,000 topographycal map, and classfies the 
geographical type of each extracted area. The 
extracted area amounts to 519 points. And the loca 
tions of these areas were input to the computer. 
2.5 Groud Investigation 
On the spot, the land-slide areas were checked, and 
the grade of weathered granit was observed along the 
road side slope. Then the weathering grade was 
classified into three types, A,B,C. A Is a 
perfectly weathered area. B is weathered a little. 
C is not much wethered. These three types and their 
locations were input to the computer. The surveyed 
points amount to 444 areas. 
3.APPLICATION ANALYSIS 
3.1 Principal component analysis 
Topographical analysis data originated as altitude 
data.So each topographical result data has sore 
correlation to another result, and no result is 
independent. Principal component analysis (PCA) 
has seme effect of extracting new variables and 
condensing the variables. Then the results of 
topographical analysis were condensed with PCA 
operation. 
3.2 Cluster Analysis 
This analysis is unsupervised classfication , and 
classifies the PCA data and remote sensed data. The 
results produce a geographical type map and some 
kind of vegitation condition map.