1. Vacant areas: A new layer of vacant areas is produced 
from the layer of development by classifying, general- 
izing and finally performing a selective search on the 
result. 
e development-classes = Local(classification) of 
development 
e vacant-developed = 
development-classes 
Local(generalization) of 
e vacant — Local(search) of vacant-developed 
. Dry lands: A new layer of dry lands is produced from 
the layer of moisture by classifying, reducing detail and 
performing a selective search on the result. 
e moisture-classes — Local(classification) of mois- 
ture 
e dry-wet — Local(generalization) of moisture- 
classes 
e dry — Local(search) of dry-wet 
. Level sites: A new layer of level and smooth sites is 
produced from the layer of hypsography by comput- 
ing, classifying, generalizing and finally performing a 
selective search on the result. 
e slope — Focal(surfacial) of hypsography 
e slope-classes = Local(classification) of slope 
e level-steep — 
classes 
Local(generalization) of slope- 
e level — Local(search) of level-steep 
. Accessible areas: A new layer of accessible sites by 
the existing road network is produced implicitly from 
the layer of development by highlighting the road net- 
work, computing, classifying, generalizing and finally 
performing a selective search on the proximities. 
e roads — Local(search) of development 
e road-proximity — Focal(Proximity) of roads 
e road-proximity-classes — Local(classification) of 
road-proximity 
e accessible-inaccessible — Local(generalization) of 
road-proximity-classes 
e accessible = accessible- 
inaccessible 
Local(search) of 
. South-facing areas: A new layer of south-facing areas 
is produced from the layer of hypsography by comput- 
ing, classifying, generalizing and finally performing a 
selective search on the aspects. 
e aspect — Focal(surfacial) of hypsography 
e aspect-classes — Local(classification) of aspect 
e south-north — Local(generalization) of aspect- 
classes 
e south — Local(search) of south-north 
. Good-sites: A new layer of sites that satisfy the set of 
constraints and opportunities is produced by the suc- 
cessive overlay of layers produced in the previous steps. 
Finally, good sites are highlighted by performing a se- 
lective search on the result. 
e vacant-dry — Local(overlay) of vacant and dry 
e vacant-dry-level — Local(overlay) of level and 
vacant-dry 
e vacant-dry-level-accessible = Local(overlay) of 
accessible and vacant-dry-level 
e vacant-dry-level-accessible-south = Lo- 
cal(overlay) of south and  vacant-dry-level- 
accessible 
e good-sites — Local(search) of vacant-dry-level- 
accessible-south 
7. Candidate sites: A new layer of sites that satisfy the set 
of constraints and opportunities and have adequate size 
is produced from the layer of good sites by measuring 
the sizes of zones and highlighting those that are within 
the predefined size interval. 
e good-sites-size — Zonal(measurement) of good- 
sites 
e candidate-sites — Local(search) of good-sites- 
size 
4 SPATIAL DECISION MAKING AND BOOLEAN 
LOGIC 
The potential of operations available in current GIS packages 
is heavily restrained from the standard logical foundation in- 
corporated into them. It is argued [Leung et al., 1993] that 
the employment of boolean logic (the all-or-nothing system) 
in GIS design causes the following problems: a) it imposes ar- 
tificial precision on intrinsically imprecise information, graded 
spatial phenomena and processes, b) it fails to determine and 
communicate to users the extent of imprecision and error, 
C) it is inappropriate to model human cognition, perception 
and thought processes, which are generally embedded with 
imprecision, d) it is inadequate to model natural languages, 
which are imprecise in nature. 
Several of these impediments are originated from the standard 
logical foundation incorporated into data-interpreting opera- 
tions available in GIS packages. Following the classification of 
Section 2, GIS operations fall into three categories according 
to their intent: 
e Computational operations: they compute and assign 
new attribute values to individual locations based on a 
mathematical model (i.e., overlay, proximity, connec- 
tivity and measurement operations). 
e Retrieval operations: they perform a selective search 
on analyzed data (i.e., search operations). 
e Auxiliary operations: they process further analyzed 
data in order to facilitate the retrieval of desired in- 
formation (i.e., classification and generalization oper- 
ations). 
The logical foundation adopted in the design of a GIS pack- 
age is tightly interwoven with the last two categories of oper- 
ations (i.e., retrieval and auxiliary operations). Currently, the 
linkages between the spatial entities (i.e., individual locations 
on a layer) and their non-spatial attributes are based on the 
membership concept of classical set theory, that is, an entity 
either has an attribute entirely or does not have it at all. No 
third situation is allowed. Hence, the selective search is in- 
tended to provide as a result the set of individual locations 
831 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996