Tadashi Ooishi1, Takashi Nanaumi1, Yoshiyuki Takei1 and Akihito Yamauchi2
Abstract
Japan has been applying Densely Inhabited Districts (DIDs) as an urbanization index considering peoples' living environment. DIDs serve the criteria for the Government's supporting-budget allocation to the local authorities. However, until past few years, the census was a manual statistical calculation which was time consuming and cumbersome. The necessity was realized by the Management and Coordination Agency of Japan to develop a system for the automatic compilation of census statistics for DIDs. PASCO developed an automated system based on ArcInfo to perform iterative calculations for population density rankings.
The Statistics Bureau and Statistics Center of the Management and Coordination Agency of Japan developed an Automated Census Mapping System (CMS) by applying the Geographic Information System technology. The geographic data of Basic Unit Block boundaries introduced in the 1990 population census and the attribute data of population census are registered in the database by this system, which made it possible to understand statistical maps and distribution maps using Graphic user interface.
The population census adopts Basic Unit Blocks as the most basic small-data units, and thereby allows the compilation of statistical and distribution maps (which show various information such as population, number of households, population density, population growth/decrease rate, percentage of employed persons by industry) for each prefecture or municipality by processing data based on each Basic Unit Block. It is also possible to demarcate Densely Inhabited Districts (DIDs) and process data for small areas such as subdivisions by cho and aza, school districts, and divisions by health center jurisdiction or local residents association. A wide range of applications include the tabulation of the population and number of households in a commercial quarter within a radius of 500 m from the station, and the calculation of population and number of riverside households, which are prone to damages by failure of embankment, in disaster prevention efforts.
The purpose of introducing the notion of Densely Inhabited District (DID) was to reveal the urban population based on the results of population census. In the traditional statistical presentation, the terms "all shi (i.e. cities)" and "all gun (i.e. machi (towns) and mura (villages)" were customarily used to represent urban and rural areas, respectively. However, many towns and villages were merged under the Towns and Villages Consolidation Promotion Law of 1953 and the New Towns and Villages Creation Promotion Law of 1956. This resulted in the involvement of sparsely inhabited agricultural areas within the boundaries of "all shi" that were originally meant to represent urban areas. The population density of the "all shi" was consequently lowered, and the term "all shi" lost its original sense of urbanity. It was then felt that the statistical data of DIDs were becoming increasingly inappropriate.
Under such circumstances, the Statistics Bureau introduced a new definition of boundaries that would clearly represent urban areas in the 1960 population census. As a result, statistical data that were the accurate representation of the actual state of urban population were provided. These data are now widely used as a criterion for calculating local allocation taxes, and as a valuable source for establishing such administrative policies as town planning, regional development programs, urban redevelopment programs, industrial location planning, transport planning, public health policies and crime and disaster prevention policies, as well as being applied in diverse academic and private marketing fields.
At the operational level, however, the demarcation of DIDs took more than seven months by manual operation. To overcome this problem, the procedures for demarcating DIDs were redesigned by grouping a number of districts evaluated as contiguous on the basis of population and population density (as well as household density) of a certain level.
To computerize the new demarcation procedures, experimental studies on system development started in June 1992 based on ArcInfo, and the development of a prototype system started in February 1993 utilizing the CMS data collected in 1990. This was followed by a series of verification studies that started in September 1993, which led to the modification of the prototype system in September 1994. Furthermore, AML was implemented to suit the new conditions in May 1995, for the purpose of demarcating DIDs in the population census of the same year. Today, the results of all these efforts are realized in the Automated Census System for Densely Inhabited District at the Statistics Bureau of the Management and Coordination Agency of Japan.
The following hardware and software were used for the development of the Automated Census System for Densely Inhabited District.
@@@Configurations of major hardware and basic software
Hardware
RS-6000(IBM) 48H
one set
RS-6000(IBM) 25H
two sets
X2020iNScalcompj
Basic software
ArcInfo
rev 6.1.2
bikan-font
The flow chart for DID demarcation is given in Figure 1.
Figure 1. Flow chart for DID demarcation
Flags were added to each attribute of enumeration districts for DID demarcation. Table 1 shows the data characteristics of enumeration districts and the respective flags.
Table 1 Condition for database construction
Data characteristics
Flag Enumeration districts with a population density of X X (inhabitants/ km2) or more
` Enumeration districts with a population density of X X (inhabitants/ km2) or more in the area excluding the surface area of urban public facilities
a Enumeration districts defined as the DIDs in the 1990 census which have a population density of Y Y (inhabitants/ km2) or more
b Enumeration districts in which the DIDs defined in the 1990 census account for Z Z% or more and which have a population density of X X (inhabitants/ km2) or more in the area excluding the surface area of urban public facilities
c Any enumeration districts in which the surface area of urban public facilities account for Z Z% or more
d Other enumeration districts including those surrounded by the districts defined as above (see the data processing procedures shown in Fig. 3), and those involving isolated lands
e * The following thresholds have been set here as an example.
X X as 4000, Y Y as 2000 and Z Z as 50
The flow charts given in Figure 2 and 3 were adopted to create the data shown in Table 1. Figure 2 shows the database construction procedures that are adopted for cases where the DIDs of 1990 and the enumeration districts of 1995 are different. For example, if the boundaries of a district do not match, as shown in the figure, the district is demarcated with a threshold of 50%.
Figure 3 shows procedures for judging whether an area surrounded by DIDs should be included in the DID. For example, if its surface area is less than 0.5 km2, it is included in the DID, but left as a blank area if its surface area is more than 0.5 km2.
@
For detailed statistical analysis of a city, region segmentation is performed on the basis of population, number of households, and the contiguity of enumeration districts.
1) Definition of the contiguity of enumeration districts
The contiguity of enumeration districts is defined as follows:
P
Those adjacent to each other at a point or on a side are regarded as contiguous
(top left in Fig. 4)
Q
Those separated by a distance of 8 m or less are regarded as contiguous
(bottom left in Fig. 4)
R
Those surrounded by other enumeration districts that have a total surface area of less than 0.5 km2 are regarded as contiguous
(top right in Fig. 4)
S
Where all the districts apart from water enumeration districts of a municipality are regarded as contiguous, the whole area of the municipality including the water enumeration districts is regarded as a single DID
(bottom right in Fig. 4)
Figure 4. Definition of contiguity
2) Evaluation criteria for contiguous districts
Searching the districts defined as contiguous in the way described in the table above, the contiguous districts are evaluated based on the statistics. Three thresholds are set: a population of Level A or more, Level B or more and less than Level A, and Level C or more and less than Level B.
* The following thresholds have been set as an example here.
@@@Level A as 5000, Level B as 3000, Level C as 1000
The contiguous districts are marked with DID numbers. In this way, several districts with a dense population within a city can be modeled as shown below.
The comparison between the old procedures and the new system is given in Table 2, which demonstrates the efficiency of the latter.
Table 2. DID numbers and their conditions
* See "2) Evaluation criteria for contiguous districts" for the details of Level A, B and C.
Naming
chc No.#
Conditions
chc
P`TO
The total population of the contiguous districts is Level A or more (Numbers are given, starting from 1, in the order of decreasing population).
Quasi-chc
TP`VO
The total population of the contiguous districts is Level B or more and less than Level A (Numbers are given, starting from 51, in the order of decreasing population).
Planed districts
VP`XX
The total population of the contiguous districts is Level C or more and less than Level B (Numbers are given, starting from 71, in the order of decreasing population).
Figure 5 shows contiguous block patterns regarded as DIDs
Figure 5 Contiguous block model with DID numbers
Below is an output example of DID demarcation of a city carried out in 1995. The area bounded by blue lines is defined as a DID which is the final target area of demarcation. Red lines represent the DID boundaries defined in the 1990 census, which demonstrates well the changes during the five years. The tint black lines represent enumeration districts. This boundary map of DID shows well that the city is being divided into small districts.
Table 3. Comparison between the new system and old procedures
Old manual procedures
Automated Census System for Densely Inhabited District
Data entry
Enumeration districts maps: A-2 size each
By means of the CMS (for each municipality)
Area calculation
Reading of the planimeter twice
By means of the basic functions of ArcInfo
Calculation of population density
By means of the host computer (ACOS)
By means of EWS (RS-6000)
Boundary demarcation
Transferring onto the topographical maps drawn on a scale of 1: 25000 produced by the Geographical Survey Institute, the Ministry of Construction.(Certain errors are produced because the enumeration districts maps are on the plane rectangular coordinates while the topographical maps are on the Universal Transverse Mercator system.)
Demarcation data are saved as attributes.(The data are archived on the plane rectangular coordinates.)
PASCO developed an Automated Census System for Densely Inhabited District based on GIS to enhance the efficiency of procedures which was time consuming demarcation and cumbersome manual process. The highlighted results include: (i) demarcation period was reduced from 7 months to 3 months (less than a half); (ii) surface area accuracy of data was improved from the conventional accuracy of around 1:25000 by compiling on more accurate scales including 1: 2500; (iii) it became possible to avoid human errors by utilizing computers; and (iv) certain demarcations that had hitherto been regarded difficult to be performed by other GIS systems were made possible by employing various overlay functions of ArcInfo.Future objectives include the establishment of DID calculation for cases where districts are divided by man-made facilities such as railways or by natural boundaries such as rivers, as well as the further improvement and application of the GIS technology.
In developing this system a great deal of assistance was received from the Population Census Division of the Statistical Bureau of Management and Coordination Agency.
(i) Population Census of Japan (Densely Inhabited Districts), 1990
(ii) Population Census of Japan Analytical Series No.3 (Densely Inhabited Districts), 1995
Name
Tadashi Ooishi1, Takashi Nanaumi1, Yoshiyuki Takei1 and Akihito Yamauchi2
Title
Automated Census System for Densely Inhabited Districts
Address
1PASCO CORPORATION 1-1-2, Higashiyama, Meguro-ku, Tokyo 153, Japan TEL:+81-3-3715-1530, FAX:+81-3-3715-1470 E-mail:tadashi_ooishi@sed.pasco.co.jp
2 Management and Coordination Agency Government of Japan 19-1, Wakamatsu-cho, Shinjuku-ku, Tokyo 162, Japan TEL:+81-3-5273-1134, FAX:+81-3-3203-8358
