Table 4. Urban Dynamics information in Study Area
Mean of burned-
out areas
rate of resi-
Ward name)
denntial areas
population
density
2
summary # corpolation
density
2
[NAGATA
2 5 4 1
Bul od 043800 0. 220408 45. Em 18.6780
M9| — 2.546550 0. 228408 8.5672
2 5
0
HIGAST- 2 2 4
NADA 3 1.0480 0.034470 0. 8567 59. 118
2 2
22| — 1,043830 0.034470 0.8567 59. 1419
Table 5. Rate of Urban Land Use in Study Area
NADA | HIGASI
NADA
NAGATA
Ward name HYOGO
ildi
Wooden-framed
Non-wooden-fr amed
buildings
areas 25
36
6. ANALYSIS OF REMOTE SENSING
Much time and labor was spent on the methodology of
stereographical analysis and the mapping system for
urban information. Therefore, in order to reduce
this time and labor, this study used an additional
method, that of remote sensing analysis, for the next
stage. Remote sensing was done to identify those zones
that would be in danger of the spread of fire after
an earthquake. This was done according to data
provided by TM Landsat in September1984, when this
remote sensing analysis was carried out.
6)-1. Determination of the spectrum of classification
for ground truth data (Figure 4.):
This study extended the use of the decipherment map
of building types, already drawn up for ground truth
data, into the process of analysis of land use
classification.
According to the examination for classification of
land use, there were five classes characteristic of
the spectrum:
370
1. Wooden-framed houses in densely built-up areas
2. Non-wooden-framed buildings in densely built-y
areas
3. Open space areas (green land )
4. Open space areas (agricultural land and grasslanj)
5. Not classified
Wooden-framed houses
.Non-wooden-framed buildings
250
$ 200
2 180
2
]- 1" 1- id
m
I 3
Openspace areas
an Openspace areas
(agrioultural land and grassiand)
iz
IMMER vey
1- T-TFrLrri
Figure 4. Spectrum Data of Classification in each band
By the simulation of visual image quality improvement
using ground truth data, it was confirmed that the
useful bands were # 2, 4, and 7.
Although in this study, the land use type was
determined with an analytic efficiency of the target
ground surface of 30m square in Nagata district, it
is still useful to receive conceptional recognition
of urban features where replacement of reflectors is
very difficult.
The red zone on next page shows the outline of the
danger area as estimated by remote sensing. (Fig.7)
It was absolutely essential to determine the danger
area before an earthquake or other urban disaster, $0
as not to have a repetition of the same burned-out
disaster.
7. RESULT OF KOBE'S DISASTER
The result of this study in Kobe are as follows :
1) Research overlaps for the danger zone from over
layer analysis compared with mapping of building
types (Fig. 5.) and mapping of remote sensing (Fig.7./
can be seen.
2) Also, there can be seen the need for further
research in view of the nearly doubled area of fire
damage derived from over layer analysis compared wit
building type mapping and fire spread mapping. (Fig.6)
3) Simulation and estimation of danger areas in té
residential zone will be desirable for the citizen
in the future renewal planning by the urban depar tment
of the government.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996
