P. Agouris (eds), Au-
'om Aerial and Space
ic road extraction by
chives of Photogram-
.3/1, pp. 324-332.
and Kutka, R., 1994.
rial images. In: Inter-
Remote Sensing, Vol.
., 1995. A hierarchi-
from aerial imagery.
an (eds), Integrating
nalysis and Machine
ification of deriche’s
tional Conference on
pace events in image
| of Photogrammetry
, 1988. Cooperative
In: Computer Vision
ınd Kübler, O., 1995.
n: A. Gruen, O. Kue-
ion of Man-Made Ob-
user Verlag, pp. 105-
1 of street objects in
nal Archives of Pho-
[, Part 4, pp. 311—317.
994. Road network
ven system. In: Inter-
Remote Sensing, Vol.
ctures: A differential
ves of Photogramme-
r, H. and Radig, B.,
nages. In: A. Gruen,
c Extraction of Man-
s, Birkhàuser Verlag,
ad tracing by profile
uen, O. Kuebler and
: Man-Made Objects
Verlag, pp. 265-274.
A HIERARCHICAL SPATIAL CANONICAL DATA MODEL —-
TOWARDS FEDERATING HETEROGENEOUS GISs
Yaser A. Bishr, M.Sc.
Ph.D. Candidate
Department of Geoinformatics
International Institute for Aerospace Survey And Earth Sciences, ITC
P.O.Box 6, 7500 AA Enschede The Netherlands
email: Yaser@itc.nl
Inter-Commission Working Group III/IV
KEY WORDS: Database heterogeneity, federated GISs, semantic data sharing, spatial canonical data mode!
ABSTRACT
Decision making process is usually multi-level, e.g., local, regional and national. Taking a decision at any level requires the
consideration of the other ones. Decision support systems which may support each level of decision making contain data,
information, and knowledge pertaining to the problem domain. In the framework of developing a multi-level spatial decision
support system for watershed management, it is required to share data, information, and knowledge between the involved
disciplines. This requires the federation of the GISs which support such decision support system. Providing the co-operation
between these autonomous heterogeneous GISs while at the same time maintain their autonomy has been an area of great
interest in the past few years. This situation is called interoperability and the system which manages the interoperability is
called federated database system. Most of the current research in the federated databases is tackling the problem of syntactic,
schematic, and to a lesser extend, semantic heterogeneity of federated databases. This is because most of the publications are
from non spatial database perspective, federating heterogeneous GISs will prose some complexities for the canonical data
model.
A canonical data model, also called a unified data model, is a wrapper around the heterogeneous databases gives non-local
users the illusion of an integrated view. It is a uniform interface to the underlying databases. The is an integral part of the
client server architecture. it is a mediator between heterogeneous GISs which allows the exchange of data and services.
The objective of this paper is to represent a model for sharing geographic information. The concept is distinguished from
other concepts by its high semantic contents. It presents a spatial canonical model and show that in order to provide reliable
and full-fledged interoperability, semantic relationship and similarity between objects should be considered. The ideas which
will be presented here are considered as an extension to GIS theory. Originally, the theory is focused on defining object
hierarchies in a single database. Here the theory is extended in order to accommodate several hierarchies within different
databases in order to provide co-operation between multiple heterogeneous GISs.
1. WHAT SHOULD BE SHARED
The need for shared information is a direct result of the multi-
level nature of watershed management. There are many
different views of a given data set by the three decision levels,
local, regional, and national. Difference in views can also arise
when management plans, designed at one level, are examined
at a higher one. This is due to the diversity in the objectives of
each management level. When various perspectives overlap,
they necessitating the for sharing of information if management
plans are to proceed concurrently and cooperatively. For more
information to be shared, there must be a commonly
understood representation and vocabulary [McGuire G.J., et al.
1993].
While computers are used extensively in product development,
existing approaches do little to facilitate information sharing
and coordination. The approaches are only focused on
providing front end interface across the Internet. The interface
allows users to browse a metadata directory and to locate data
sources. Data are then transferred from the source either in a
standard data format or users might be exposed to a set of
available data standards from which they can select [Alaam M.,
1994] and [Otoo, J.E., et al., 1994]. The main disadvantage in
59
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996
this approach is that users spend a substantial amount of time
restructuring their data in order to comply to their data model.
Interoperability is defined as the ability of GIS users and
developers to transparently access geographic data sets and
processes available on the net from heterogeneous systems,
provided that the autonomy of the members of the federation is
maintained [Bordie M.L., 1993]. Accessibility of data,
knowledge, or functions is guaranteed under full consistency,
integrity constraints and concurrency specifications, i.e., full
interoperability. Members of the federation are called
component SDSSs, and their GISs are called component GISs.
Understanding what exactly is needed to be shared is a
precursor for providing full interoperability. Spatial objects are
traditionally described in geographic databases by their
geometric and thematic attributes. This is known as the
syntactic description of geographic objects. Object
identification based on their syntax in a distributed
heterogeneous database renders users to search for objects by
their geometric and/or thematic attributes. The shared objects
might have different meanings in both the data source and the
receiver. For example, a user might query the federation for
height information of a particular area which has aspect equal
to 7, measured on a discrete scale from 1 to 9. On the other
hand the data source might have aspects stored in degrees
