A DESIGN FOR AN INTEGRATED GEOGRAPHIC
INFORMATION SYSTEM
Mark Gahegan
Department of GIS, Curtin University,
PO BOX U 1987, Perth 6001, Western Australia.
Tel: +619 351 3309, Fax: +619 351 2819
email: mark@cs.curtin.edu.au
ABSTRACT
This paper is the first part of two, describing in an informal way some of the issues related to the
integration of GIS and remote sensing activities into a single logical framework. The advantage of this
approach is that the lines of communication between remote sensing and geographic systems can be
improved drastically, leading to better control of image analysis and better understanding of the
meaning and applicability of the geographic features derived from remotely-sensed imagery. A design
is presented for an integrated system that presents a number of distinct views onto the underlying data,
of varying degrees of abstraction, to support the needs of an expanded user base. Multiple spatial
representations of individual geographic features are essential for an integrated system; their origins
are discussed and they are specifically included in the design.
1. INTRODUCTION
The first step towards full integration of geographic information systems (GIS) and remote sensing is to provide an
architecture whereby the different types of data and different data structures used in each not only co-exist, but can
also interact freely. This in turn requires that particular care be given to the process of data abstraction and feature
extraction, since within a single application, data may differ by spatial type, data structure and implied meaning
(semantic content).
Ehlers et al (1991) note the importance of an integrated approach to the design of the spatial data structures used,
and it is precisely this approach that is taken here. It is worthwhile reflecting on why this should be important. Only
if the relationships between the source and derived data can be specified within the semantics of a single data model
does it become possible to describe the process of data integration formally. A unified data model is presented that
integrates continuous (field), image, overlay and feature data, and by extension, others too. The data model is built
upon on a single generic object-type that may be projected to different levels of abstraction. The user is presented
with logical (abstracted) concepts that are based around the operations to be performed, rather than the data types to
be used. Mappings from these concepts to appropriate data are provided seamlessly. Thus the user has a model that
naturally reflects the different levels of meaning inherent within their data in a manner that is completely
independent of issues such as data structure and type.
1.1 THE NECESSITY OF AN INTEGRATED APPROACH
The geo-objects (here termed features) used in GIS are often derived from lower forms of data, such as remotely-
sensed imagery, by applying a process of extraction which may be manual or automated. Current systems do not
have an integrated model to describe the transformations used in extraction; the outcome of each step is treated as *
* Presented at the International Society for Photogrammetry and Remote Sensing “Workshop on New Developments in
Geographic Information Systems”, Milan, Italy, 6-8 March 1996.