ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
The algorithm for determining changes in the feature space 
region between two time points is as follows: 
[Step 1] 
By setting T as system time and performing a feature space 
search with P as the reference point for feature space D, 
obtain a vector (V1) constituting the feature space region and 
a connector (C1) representing the feature space region. Set 
Q = C1. Set N = 1. 
[Step 2] 
By setting T' as system time and performing a feature space 
search with Q as the reference point for feature space D, 
obtain a vector (V2) constituting the feature space region and 
a connector (C2) representing the feature space region. 
[Step 3] 
When V1 = V2 (all points are equal), Q'N = Q, R'N = 1, and 
the procedure ends. Otherwise, proceed to Step 4. 
[Step 4] 
Compare V2 with V1, and partition V1 into a region included 
in both V1 and V2 (Vc, which may be a set of multiple 
polygons), and the part of feature space outside that region 
(part included in V1 but not V2) (Va1 - Vak: a set of polygons). 
Add Va1 - Vak to the list of Va. 
[Step 5] 
If Vc exists, set Q'N = Q and seek R'N, and house each point. 
Set N = N+1. 
[Step 6] 
If Va exists, select one polygon Vat from the list of Va. 
Generate a point inside Vat, set that point as Q, and return to 
Step 2. If Va does not exist, the procedure ends. 
6. Data utilized and strategy for data management^] 
6.1. Data utilized and strategy for data management 
Census data consists of base unit number and shape 
information for base unit area, and base unit number is used as 
a key to relate this data to total information (population, number 
of households, etc.). The main shape information consists of 
base unit area boundaries and the main attribute information 
consists of base unit numbers and total information. We make 
some proposals regarding the method of management using 
DyMSIS/KIWI+ format. 
(1) Base unit area boundary data 
This data is managed as shape data using vectors. Base unit 
area boundaries are revised after the previous census date, 
and all revisions are completed prior to the current census date. 
Accordingly, temporal factors for the base unit boundary 
vectors are set as follows (Figure 6a): 
GS: Date of data revision (in the case of initial data, date of 
data creation) 
GE: First census date after data revision date 
ES: Date of data revision (for information that is not subject to 
extinction, MAX value) 
EE: First census date after data revision (for information that is 
not subject to extinction, MAX value) 
Further, feature spaces are defined as follows: 
(1) Base unit area feature space 
Area feature space defined using base unit area data (vector) 
and base unit number data (connector). 
(2) Total data feature space 
Area feature space defined using base unit area data (vector) 
and each item of total information data (connector). Defined in 
accordance with total data type. For example, a population 
statistic feature space is defined using population statistic 
connectors. 
Not Exist 
► 
1995.10.1 Revision Revision 2000.10.1 
Date A Date B 
Census Date 
(a) T Factors for Base Unite Boundary 
1995. 
Exist 
Total Result 
Data in 1995 
Not Exist 
Exist 
Total Result 
Data in 2000 
0.1 
Not Exist 
2000.10.1 
Census Date 
Census Date 
(b) T Factors for Total Result 
Figure 6. T Factors for Census Data 
6.2. Revision of base unit area data 
Base unit areas are revised with the census date. In the 
following, revision of base unit areas is considered from the 
viewpoint of data management (Figure 7). 
(2) Base unit number data 
This data is managed as representative points of areas using 
connectors. Because base unit numbers, which are connector 
elements, change with revision of base unit boundaries, the 
method of setting temporal factors is the same as for base unit 
boundary vectors. 
(3) Total result data 
This data is managed as representative points of areas using 
connectors. Because total information changes every time a 
census is conducted, the following temporal factors are set for 
total census information (Figure 6b): 
GS=GE: Census date 
ES=EE: Day before next census date 
(1) Merging of base unit areas 
Two base unit areas are merged to form a single area. In data 
management, the common boundary between two regions is 
eliminated. In this case, the base unit numbers for the two base 
unit areas are eliminated and a new base unit number is 
generated. 
(2) Division of a base unit area 
One base unit area is divided into two areas. In data 
management, a boundary line dividing two regions is 
generated. In this case, the base units number for the original 
base unit area is eliminated and new numbers are generated 
for the two new unit areas.