You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

The 3rd ISPRS Workshop on Dynamic and Multi-Dimensional GIS & the 10th Annual Conference of CPGIS on Geoinformatics
Chen, Jun

ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001
Michinori HATAYAMA 1 Shigeru KAKUMOTO 2 Hiroyuki KAMEDA 3
1 Disaster Prevention Research Laboratory, Kyoto University / Japan Digital Road Map Association
Gokasho, Uji, Kyoto, 611-0011, Japan
Phone: 81 774 38 4037, Fax: 81 774 38 4044,
E-mail: hatayama@imdr.dpri.kyoto-u.ac.jp
2 Frontier Collaborative Research Center, Tokyo Institute of Technology / Central Research Laboratory, Hitachi, Ltd.,
E-mail: kakumoto@crl.hitachi.co.jp
3 Graduate School of Informatics, Kyoto University / Disaster Prevention Research Institute, Kyoto University /
Earthquake Disaster Mitigation Research Center, RIKEN
E-mail: kameda@imdr.dpri.kyoto-u.ac.jp
KEY WORDS: Spatial Temporal GIS, Data Structure, Census Data
The temporal data is one of the most important factors for making an analysis of the geographic data, because the entities that compose
of geographic data change with temporal factor. So, we have developed the Dynamic Management Spatial-temporal Information System,
called DyMSIS. In this paper, we firstly show how to manage the temporal information. And then, we explain the algorithm to pick up the
difference of census data between one time and another time. Finally, we demonstrate the example of the temporal analysis of census
1. Introduction
Geographic information changes rapidly with time. So temporal
information may be used to manage geographic data effectively.
In this paper, a geographic information system with a geographic
data management function employing temporal information
(spatial temporal geographic information system) is considered.
First, the requirements imposed on the proposed system are
examined. Then, a data base schema to meet those
requirements is proposed, and a spatial temporal geographic
information system, DiMSIS, which can handle such a database,
is described. Finally, we propose a cumulative data management
method for census data.
2. Requirements for a spatial temporal geographic
information system
[Requirements from the viewpoint of data management]
(1) Data is in compact form.
(2) The data structure is simple and easy-to-understand.
(3) The system can handle temporal information having
uncertain values.
[Requirements from the viewpoint of data usage and analysis]
(1) The system can handle maps for specified times.
(2) Maps at two time points can be visually compared.
(3) The system can handle regions of change between two time
3. KIWI+ {A Spatial Temporal & Simple Topology - Open
Database Schema: ST2-ODS)
The KIWI+ format combines the KIWI format proposed in
ISO/TC204/WG3.2 as a data format for car navigation systems
and the DiMSIS data structure proposed in [1]. In this format,
spatial temporal objects can be handled on the basis of temporal
management using the Space-Time Approach model [2], and by
describing objects using the implicit description and calculation
type of data model [3], a compact, easy-to-understand structure
is obtained. In the following sections, the characteristics of this
system are described.
3.1. Database Structure
The KIWI+ describes all geographical information in the forms of
two elements: vector for shaping graphic data, and connector for
relating attribute data. Each element has the spatial and
temporal factors. The temporal factors can be resolved into four
factors: generation start time (GS), generation end time (GE),
extinction start time (ES), and extinction end time (EE) (Figure 1).
If the dates of generation and extinction cannot be identified, it is
also possible to represent temporal error using these factors.
3.2. The concept of feature space
In order to utilize the data structure explained in 3.1 efficiently
and to calculate the topological structure, the concept of feature
space is introduced in the KIWI+ format [3].
(1) Definition of feature space
A set of objects with the same type ID is called a class group.
In accordance with the objects they comprise, class groups can
be classified into vector class groups and connector class
groups. A set of correlated multiple vector class groups and
multiple connector class groups is defined as a feature space.
(2) Spatial temporal analysis using feature space
A feature space has a physical significance as a factor in terms
of spatial temporal analysis. As such, there are four types of
feature space: point, line, area, and solid feature space (Figure
2). Defined as factors in a topological relationship, these
correspond to points, lines, areas, and solids. Spatial temporal
analysis is performed in feature space by relating vectors,
which comprise geometric information, and connectors, in
which attribute information is linked. Because these
relationships are calculated in real time when a processing
command has occurred, it is possible to dynamically
supplement objects corresponding to topological