Scale Change Distortion Projection 7 tation/R : 
Ponaval Change ranslat on/ otation 
Coordinate 
Fig.5 Fundamentals of Geometric Manipulation 
cently the transfer on university from 
center of Tokyo have brought the new land 
development and resulted the environmental 
change for aiming new city planning. Here 
can reach in one hour drive by car forward 
western direction from center of Tokyo. 
4.1 Change Detection 
Two temporal Landsat TM data were 
prepared for the change detection analy- 
sis. One was observed on Dec.4, 1984 and 
other was Dec.5, 1990. As the preprocess- 
ing, geometric correction was carried out 
for both of TM data in order to overlap on 
the map coordinate. Each corrected imagery 
consists of 512 pixels and 400 lines data 
with 4 bands of band 1,2,3,4 and is stored 
in one floppy disk. 
In this area, typical environmental 
change pattern is supposed that forest 
has been changed to bare soil by land 
development. 
Therefore using two temporal NVI 
data, which concerns the status of vegeta- 
tion cover, land cover change detection 
that is from vegetation to non-vegetation 
(i.e. land development ) is carried out. 
NVI is calculated by following formula; 
  
(“Band à - Band 3 } 
NVI = *Gain + Offset 
(€ Band 4 +“ Band: 3°) 
Gain = 100, Offset = 50 
Here, this change detection means where 
and how many area of land cover change has 
been occurred in six years from '84 to 
"90. 
Actual detection procedure is as 
follows; (1) Calculation of Normalized 
Vegetation Index (2) Histogram Measure- 
ment for NVI data (3) Define threshold 
value for separating land cover type into 
vegetation and non-vegetation classes. 
(4). Classification to. two classes by 
Density slicing. These processes from (1) 
to (4) must to be executed for each data. 
Finally change detection is carried out 
for two results of (4) by logical opera- 
tion. 
4.2 Regional Characteristics Investigation 
In this practice, every trainee can 
understand how GIS will be used with 
remote sensing data. 
At the first practice, town condition 
evaluation by population density is exe- 
cuted as an analysis using polygon and its 
attribute GIS data. This analysis proce- 
dure is as follows; (1) Calculation of 
population density using town population 
data as the attributes for each town (2) 
Grade for each town by the above calculat- 
ed data (3) Coloring for each grade using 
town boundary raster typed data (polygon 
type) (4) Computation the statistics for 
the result of grading. 
At the second practice, the relation- 
ship between population density land cover 
change is examined by combining the result 
of grading of population density and 
result of change detection. This process 
is the one of the summations of results, 
obtained by remote sensing. 
5. Conclusion 
In this developed system mentioned in 
this paper, there are two main training 
categories such as fundamentals and 
applications. 
In the fundamental training, the 
image data structure, format and its own 
characteristics as some of basic image 
processing techniques could be understood 
through histogram measurement and density 
slicing practices. As fundamentals for 
GIS, GIS data structure and geometric 
manipulation could be learnt through GIS 
data generation by digitizer and vector- 
raster data type conversion. 
In the application training, through 
the analysis of land cover change detec- 
tion as an application of remote sensing, 
what kind of informations extract from 
remote sensing data could be learnt. And 
how combine with the result of an analysis 
of remote sensing data and GIS data such 
as town boundary polygon data, town popu- 
lation as its attribute data. could be 
understood through regional characteris- 
tics investigation. 
This system has been developed in a 
continuing form as a series of educational 
tool for remote sensing and GIS. There are 
still some revising points in order to 
teach and learn easily. The authors hope 
to have your suggestions on this matter.