nter-
GIS
s are
users
ft of
that
such
logy
it 1s
of all
recht
t can
ation.
ready
ly no
or to
Wise,
tions,
) per-
rious
4sion,
X the
repre-
ms' 1S
ation,
Tr op-
>1-0r1-
ough,
1ld be
e ran-
above
rmany
onfer-
NFO?
51 [ze] Ee
Search: | 7 E ^ a
Interpolation Thematic Search Spatial Search (Re-)classification
Location Analysis: € a. 34 SOS
Buffer Corridor Overlay Thiessen/Voronoi
get Cu e
Terrain Analysis: m RN as E B
Slope/Aspect Catchment/Basins Drainage/Network Viewshed Analysis
prob of A e M Persei
Re : A, ZZ s
eis Iber oc Cost/Diffusion/Spread Proximity Nearest Neighbor
i i d. iL | * Aa 1 Nl
Spatial Analysis: d: aaa Pa
Multivariate Analysis
Measurements: perm]
Measurements
Pattern/Dispersion
Centrality/Connectedness Shape
Figure 1. The 20 universal analytical GIS operations
A number of operations are well-known and there is no dis-
cussion about how to categorize them (e.g. the group of terrain
analytic operations). The omission of Network functions is ar-
guable; however, a GIS user interface needs to be adjusted to
the field of application, and in a utility (network) application it
makes a lot of sense indeed, to have network functionality as a
separate group heading. In like manner, the naming of the
‘cost’ operation could equally well be called diffusion’ or
‘spread’ - it is a matter of domain, which term is more appro-
priate; algorithmically they are all the same.
In the following, six groups of analytical GIS operations are
differentiated (group headers will always be capitalized,
whereas the individual operations begin with a small letter).
Search operations can be partitioned into thematic searches
and search under geometrical constraints. In most cases, the
search operation is succeeded by a ‘select’ and subsequent
analysis of the selected object(s). As such it is not a typical
analytic operation itself. It is included here, because especially
the 'search-by-region' operation can not expected to be found
in other but GIS packages. The search-by-region operator uses
a user-defined search (rectangular) window, an arbitrarily
shaped mask, or a filter that has some spatial properties. '(Re-)
classification' is basically a database operation. In most cases,
however, the filter that is used for a reclassification has a spa-
tial determinant. As a matter of fact, the whole concept of Map
Algebra' can be regarded as a form of reclassification. The unit
of measurement does not need to be metric. Distances could be
expressed just as well in time or in relative spaces, such as
number of nodes in a network.
The 'Locational Analysis' group is comprised of four opera-
tions that are among the best-known and most often used GIS
operations at all. 'Buffer' and 'corridor' are quite similar, and it
could be argued that a ‘corridor’ is a ‘buffer’ operation that
takes two distances (the inner and the outer boundary) and can
only be applied to a group of 2- or higher dimensional features.
81
But that is a technical view of the operation and does not meet
the requirement of user-orientation set forth as a prerequisite
of this work.
The probably best-known analytical GIS operation is ‘overlay’.
It is comprised of many other operations such as ‘clip’, ‘erase’,
split’, ‘identity’, union’ and ‘intersect’, and can be applied to
any combination of spatial features. The degree of sophistica-
tion of a GIS application depends on the knowledge that a user
possesses about the resultant of any of these operations. The
last item in this group is the "Thiessen/Voronoi' operation. This
one is sometimes also categorized as a "Neighborhood" opera-
tion. However, from a task-oriented perspective, it fits well
with the three above. One operation that became completely
subsumed under 'Voronoi/Thiessen' is the flow-between-
regions. By assigning weights to the Voronoi nodes, it is pos-
sible to simulate the flow between the Thiessen polygons.
These four operations satisfy most needs in the large set of lo-
cation/allocation problems. There are overlaps with function-
alities in the 'Veighborhood' and "Terrain Analysis’ groups, but
for the sake of a clear categorization with not all-encompassing
groups, other operations suitable for location/allocation prob-
lems were kept with their more prototypical group headings.
The operations of the next functional group deal with explicitly
or implicitly 3-dimensional data. They are all well-known and
therefore need no explanation beyond the description of pa-
rameters used in the algebraic specification. Slope/aspect' re-
quires an input file that contains height values. In the excep-
tional case of the input file being a TIN, the result of this op-
eration is already implicitly recorded. Although not being a
truly analytical operation, hill-shading can be accomplished
within the same operation, if the sun azimuth and the viewer's
elevation are provided as additional parameters. The 'catch-
ment/basins' operation takes either a 'height' or a 'slope' file and
calculates the extent of a single basin (if an additional selec-
tion point is provided) or the entire set of basins. Similarly, the
‘drainage/network’ operation computes either the flow from a
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996
