International Archives of the Photogrammetry, Remoie Sensin, and Spatial Injormation Sciences, Vol XXXV, Part B7. Istanbul 20€ :
Figure 1. Scatter chart of population density and green
tract of land area ratio
+ Chiba
area ratio %
*Urawa
Mats
oadiodosav: ay
lama
and total
Î
|
| 10k
207 + Omiya ; .Miyamae
|
|
i
H
|
een tra:
Gr
+ Nish
asuKouto |
Dive
ely a Sora
Nakano
af i
Toshima
+ Kawasaki
DE ——
2000 4000 6000 8000 0000 12000 14000 16000 18000 20000 22000
Population density "Population/m ^
Thus, it was assumed the precondition which understood
urbanization examining which category gave the important
effect in which item the correlation was high.
4. Analysis by using Landsat TM image
It analyzed it as follows
image for city environmental understanding to verify even
where was possible by the remote sensing analysis from the
GIS item referring to the understood point.
4.1. Understanding of city properties by land coating
classification
Two kinds of land coating classifications of supervised
classification and unsupervised classification were done
from the Landsat TM image at two time of 1985 and 2000
to which geometry was corrected by using numeric map
25.000. It classified it into 7 classes, that is, wooden, non-
wooden, road] railway, bare land, forest, rice field, and
waters. In supervised classification, numeric map
25,000(map image) and the aerophotograph, the house map,
and the
land use investigation, etc. were used as grand truth. Figure
2 shows supervised classification result of the whole area
of the district for the investigation in TM 2000.
Figure 2. Land coating classification by
supervised classification (Landsat TM 2000)
This of each municipal district was cut out, the area ratio
of each class was calculated, and the pixel distribution
situation was understood. It was thought that the point that
pixel density which hit the urban area on a wooden
structure, non-wooden structure, and the road was high and
pixel density of the green tract of land in the bare land and
by the use of the Landsat TM .
the forest is low was an overcrowdedness urban area. '
classes (the road, wooden, non-wooden) were understood
as an urban area, and 3 classes (the rice field, the forest.
and the bare land) were understood as a green region. Thc
district characteristic was understood from each ratio in
each municipal district (Figure 3). Because the distribution
relation of the city was almost corresponding, a green
region and urban situation was able to be understood from
the land coating classification which used the TM image by
comparing Figure 1 and Figure 3. That is, the situation of
the overcrowdedness urban area where the green region
rate decreased extremely was able to be understood in the
place where an urban rate had increased. Moreover, the
transformation tendency to the city was understood from
the change in the number of pixels in 1985 by a similar
analysis in the TM image. Thus, the city properties as the
urban area were able to understand both the regional
characteristics and the distribution situations in the outline
by the land coating classification of the TM image.
+ Chiba
30
+ Aso
Sa
= x. Eun
S «Kashiwa * Asah
o»
& suba
5 is tUa Takatsu
© * Omiya, Matsudo" Kanne sit
ES fchikawa + Kqunan, Shibuya
c 10 ex * Narashino
$ + Malogaÿa Ninato + Nak& Vi Nokes
Oo :
; » Tsurgfe DW
e "in " ANN gyro
ayAas
25 35 45 55 65 75 85
Urban rate "X"
R=-0.98
Figure 3. Scatter chart in composition ratio of urban
area and green region (TM 2000)
4.2. Understanding of-city properties by image
operation processing
In the land coating supervised classification, it is necessary
to set the brightness value by using detailed regional
information (grand truth). On the other hand, similar city
properties to the above-mentioned were understood by
using 2 indices which were able to be calculated by the
operation processing in the satellite image, that is, NDVI
(vegetation index) and UI (urbanization index) as another
technique. Two calculated indices were made, and the
mean value was taken in each municipal district and the
scatter chart was made (Figure 4).
Vegetation index (NDVI) became a decrease tendency as
urbanization index (UI) increased. Moreover, the
distribution situation of each municipal district understood
the tendency resembled closely. Therefore, urbanization in
each municipal district and the state of vegetation were
able to be analyzed by analyzing NDVI and UI equally to
the land coating classification as well as Figure 3. It can be
said a useful technique in this analysis technique's
understanding the city properties in the outline, too.
1116
i hina
Riese Fung tins
95
AMA NUNC etae
