quires all
ation nor-
the above
y encoun-
le data is
the user.
> required
user’s ap-
y percent
on, is de-
etting up
ivided be-
out 20%)
0%). The
vare tools
vare.
on refor-
1e data is
adily uti-
eady has
driven or-
opment of
f the well
ent of DI-
One still
the devel-
ally justi-
the prob-
recognize
tC/INFO,
ün spatial
> and then
on with a
relational
stc.). The
al data in
or perfor-
atial data
ial tables,
ion within
, database
'e reasons,
t to store
from the
ndor spe-
n internal
o the GIS
s that the
he GISs is
really that of data exchange, or interoperability
between database management systems. This al-
ternative view of data interchange problem in GIS
is the motivating factor behind the development
of a Federated Spatial Information Management
System code named Delta-X.
The current scenario in a GIS application de-
velopment is that the required data sets are avail-
able from agencies and institutions that, for op-
erational reasons and the nature of their organi-
zational structures, are geographically dispersed.
Development of value added spatial data sets, by
different organizations, sometimes are referenced
to different base maps or to base maps that are
out of date. The data purchased by users from
data producing agencies follow a zero-one func-
tion. For example, in making a request for some
selected features of a specific map sheet, the user
must purchase all the information compiled for
that map sheet and extract the feature relevant to
his applications. The tendency then is for users to
avoid the cost of repeated purchases of the same
digital information. A digitized map sheet is pur-
chased once and reused over a long period even
when the data is out of date. The different rates
in which different organizations compile, and re-
vise their data sometimes force organization that
are very dynamic to collect the spatial data they
require themselves instead of waiting for a copy
from the authoritative source that may be very
slow paced in its operations. The result is the
proliferation of inconsistent replicas of the same
digital information at different precision and ac-
curacy.
Granting that GIS software is highly dependent
on the underlying databases, it appears then that
some of these problems can be easily resolved by
providing an integrated but selective access using
a distributed database solution. Unfortunately,
the scope of the problem is still beyond that ad-
dressed by a distributed database management
system [3, 4, 13]. Rather, a multi-database sys-
tem [3, 13, 17], has the potential of resolving a
greater proportion of the problems. Some prob-
lems not easily addressed include:
Issues of Legacy Data: It should be recog-
nized that the compilation of most GIS data
predates the development of database tech-
nology. As a result, large volumes of data-
sets are currently not maintained with any
DBMS. They are maintained simply as large
sequential files on reels of tapes.
Underlying Data Models of DBMS:
Even if all the data can be managed us-
ing database technology the databases have
independent custodians and as such, have
underlying database management systems
(DBMS) that are based on different data
models: relational, network, and object-
oriented.
Cost of Adopting a DBMS: The volume
of these data-sets requires a major financial
commitment and undertaking to restructure
them using DBMSs. Only a small percent-
age of the data can often be feasibly made
available on-line. In some cases the rate of
data capture is so high that the responsible
agencies have no chance of ever restructuring
the already existing data.
Data in the Past: Most database manage-
ment systems operate with very little sup-
port for the time dimension. Unlike the
corporate databases, e.g., personnel records,
GIS users sometimes still require access to
both past and current data to model and
predict situations in the future. Efficient ac-
cess to temporal databases requires support
from the underlying DBMS. Most commer-
cial DBMSs do not adequately support tem-
poral queries.
The issues raised above suggest then that a so-
lution to most of these problems requires the de-
velopment of a spatial multi-database that pro-
vides efficient access to both legacy and temporal
databases [21]. Strategy for interoperability in a
network of database management systems, that
use the same underlying relational data model,
has only recently been addressed [8, 10, 5]. The
question of heterogeneous database interoperabil-
ity is still an active area of research [1, 13, 16]. In
the mean time, there is a large number of opera-
tional GISs being used in different circumstances
and in different applications. How can the con-
tents of these multi-databases, be accessible from
different Geographic Information Systems?
The Federated Spatial Information Manage-
ment System (FSIM) attempts to address this
question in a consistent and coherent manner.
Basically, it is a multi-database system with
a common integrated global conceptual schema
503
