613 
For effective analysis, a feature model 
requires more than positional information. 
A full feature definition includes: 
• Positional information locating the 
feature and (where this is not 
implicit or easily deduced) 
information relating the feature to 
other features or surfaces. 
• Cartographic information 
depicting where the feature should 
be displayed on a map. This is 
not always the same as a feature’s 
actual position, since it may not 
be possible to display the feature 
clearly at its actual position. 
Features are often labelled 
requiring separate positioning 
information. 
• Graphic attributes designating 
how the feature and its label 
should be displayed on a map; 
e.g., special symbols designating 
different types of roads or 
boundaries. 
• Attributes which provide 
non-graphic information about the 
feature; e.g., class, grade and 
maintenance data for roads. 
A special hybrid of a surfaces and features 
model is the polygonal cover. Here, the 
area of interest is divided into polygons 
that combine the spatial components of the 
surface model with the attribute definitions 
of the features model. Common GIS 
polygonal covers include ownership, 
forestry stands and soils map information. 
software. In-depth analysis of these 
models is the topic of many books on GIS 
and is well beyond the scope of this 
presentation. However, a brief overview 
follows to further discussion 
Surfaces tend to fall into two types of 
models: tessellation or irregular 
triangulated networks (TINs). The 
tessellation models subdivide the area in a 
regular manner, and provide information 
for each regularly shaped entity. That 
information may be accompanied by an 
implicit or explicit generalization 
technique. The most natural example of 
this is remotely sensed imagery. Here, the 
input provides a single value for each 
picture element or pixel representing the 
reflectancies read by the sensing device. 
This technique has long been used to 
model elevation. The most common 
method uses square pixels and provides the 
elevation at the centre of each square. 
Many software packages provide a 
generalization technique such as a bilinear 
fit to provide information for positions 
other than pixel centres. This method 
provides a matrix of values and is most 
commonly known as the raster technique. 
A second surface model is provided by a 
TIN. A TIN divides the area into an 
irregular network of triangles and provides 
information about the vertices of these 
triangles. Most systems generalize to 
positions within the triangles by fitting a 
plane to the vertices, although some use 
more complex surfaces. 
GIS Models 
For each construct, there are several 
models used in commercially available