4.1 Evolution stage 1: Independent sub-systems
A GIS can be composed from a set of sub-systems as shown
in Figure 1. Although this seems to be a rather easy (but
expensive) approach to construct a GIS, it can only be
regarded as a low level of integration with a low degree of
unification. Each sub-system offers only a subset of all the
functions of a GIS to carry out some specific tasks along the
geoinformation production line. With respect to the above
defined criteria, this kind of system has the following
characteristics.
1) The system is composed of several sub-systems either in
form of hardware or software. Therefore, it is not compact.
2) Different sub-systems may need different hardware and
OSs, e.g., VMS, Unix, MsDOS, MacOS.
3) The system cannot provide a central control panel.
Functions of a sub-system are only reachable from the
respective local control panel.
4) The data cannot be accessed from a single entry point.
The data transfer from one sub-system to the other may need
to be done manually, e.g., using floppy disks, if the sub-
systems used for consecutive operations are not connected
on-line. Data conversion is likely to be required because
typically each sub-system will used its own data structure.
5) Components of information are usually stored separately
in the local database of each sub-system. For example, data
representing man-made objects may be stored in the CAD
sub-system, data of terrain relief in the DTM sub-system, data
of other terrain objects in the 2D geoinformation sub-system.
This implies that metric computation must be used to
integrate data from different sub-systems before topological
relationships can be created.
6) The system does not provide a common user-interface.
The user-interface is locally provided and dependent on each
sub-system. This implies elaborate user training and
operation liable to mistakes.
7) The investment costs are high because each sub-system
has to be purchased separately. A number of sub-systems is
required to achieve the required functionality.
8) Maintenance is difficult and also expensive. Different
vendors may be responsible for different sub-systems.
9) Data redundancy is most likely very high because of
separate and independent storage.
10) Dealing with uncertainty is necessary in operations that
involve datasets from different sub-systems.
11) Users have to cope with many problems, so the
productivity is not likely to be very high.
12) Difficult to tum the production line into automation due
to several limitations mentioned earlier.
13) This approach requires various supporting personnel, e.g.
OS specialist, application specialist, application programmer,
etc., to ensure operation.
14) The size of user organisation is quite large in terms of
number of personnel and space required for placing the sub-
systems.
296
(Cice iste r&y ncs Organizational shell .;1. niossoolo gy
fei cordons jyom noüsinesnd sf vd fo)
|
| T es 3 E Fe ETE are
|. [| 2D. IS (7) IE Cartography
pr —P— pa a
I" Koh
y cons sonde
|| / Data | de RS/Image |
Lk CAD I = Conversion Md Processing
t |
rosas
ed Rn
DIM [| IE Photogrammetry |
DM me d
Ax rS da
Figure 1 System composition by independent sub-systems
Figure 1 is a graphical illustration of this approach. The data
conversion plays a central role to integrate components of
the spatial model which is stored separately and
independently as databases in various sub-systems.
4.2 Evolution stage 2: Functional integration
A system based on this architecture combines all necessary
functions into one software package. Figure 2 illustrates this
approach. The system has the followings characteristics.
1) The system is compact because all sub-systems are shrunk
down into functions or software modules that are
implemented within the system.
2) The system is based on one OS and hardware platform.
3) The system provides a central control panel.
4) The data can be accessed from a single entry point. The
data transfer between software modules can be done as
background process.
5) Each module may have its own data structure to store
data. For example, coverage data and TIN data in Arc/Info
are stored in separate datasets with different data structures.
Topological relationships among data elements across
different datasets do not exist.
6) The system can provide common user-interface.
7) This approach is less expensive than the independent sub-
systems and the client/server (see section 4.3) because it is
based on only one software package.
8) Maintenance is easy, because of fewer pieces hardware
and software are to be maintained and only one vendor to
be dealt with.
9) Data redundancy still exists among different datasets.
10) Problems in handling of uncertainty are similar to
evolution stage 1.
11) The productivity is likely better than the client/server
approach because all processes are locally performed under
one system shell, thus, requires less time for data transfer and
message translation.
12) Many operations can be automated which makes it more
feasible to automate the whole production line. The user has
a possibility to optimize and streamline the operation after
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
Ali
att
the
ca
us:
to
Th
int
to
Th
1
the
of:
2)
dif
Col
pla
wit
3)
fur
fun
req
rec
req
