Location and GIS - Location Science - page 493

be known prior to conversion. This should determine the cell size of the output
raster map during conversion. If the cell size is large, then data may be unnecessarily
generalized. On the other hand, if the cell size is too small, an excessive number of
cells may be created resulting in a huge amount of data and slower processing times.
Each of these two spatial data models is characterized by certain advantages and
disadvantages. In the vector model, data can be represented at its original resolution
without generalization. Moreover, since most geographic data is in vector form, no
data conversion is required. On the contrary, the location of each vertex needs to be
stored explicitly. In addition, continuous data such as sea-depth or elevation cannot
be represented easily in vector form. As far as the raster model is concerned, the
location of each cell in the raster is implied by its position in the grid which implies
that no geographic coordinates need to be stored, other than one reference point, e.g.,
the top left corner of the grid. On the other hand, it is not easy to represent in a raster
model linear features or network structures. For more details on the advantages and
disadvantages of these models, see Church and Murray ( 2009 ).
Image data may also be used to store remotely sensed imagery, such as satellite
scenes or aerial photos. Image data is typically stored in a variety of formats (e.g.,
.TIFF, .PNG, .JIF, etc.). Most GIS software packages allow the input and display of
such formats typically, through conversion into a raster format (and perhaps vector)
to be used analytically with the GIS.
Finally, attribute data is typically represented through relational database models
where data is organized in tables containing rows and columns. Each row corre-
sponds to a record and each column stores the values of a specific attribute. Most
GIS packages offer an internal relational data model as well as support for external
relational databases thus enabling the use of large existing datasets.
Most of the early GIS implementations gave greater emphasis on spatial data and
tended to ignore the time dimension in data representation. However, the existence
of a huge volume of spatial-temporal data and the ever advancing technology have
necessitated the extension of traditional models to cater for the temporal dimension
as well. The inclusion of time often results in complex, large, and highly varied
datasets. At the moment there does not seem to be a standard database model or
analytical approach to handle these complex datasets. As reported by de Smith
et al. ( 2013 ), specialized techniques have been developed for specific cases. Typical
examples include the approach employed to capture land-use change (see IDRISI's
Land Change Modeler package (IDRISI ( 2013 ))), the modeling of coastline advance
and retreat (see Ahmad 2011 ) and the extension of spatial scan statistical procedures
to spatio-temporal point data for crime analysis (see Cheng and Adepeju 2013 ).

19.2.1 GIS Functionality

Since the mid 1990s a large variety of GIS tools has been developed that have
been employed for academic as well as commercial purposes. Some of them are
generic GIS packages that may be used in different applications whereas others were