Geoscience Reference
In-Depth Information
the optimal facility locations and evaluate the effect on the objective function
value. Murray and Grubesic (
2012
) identified four categories of uncertainty (object
geometry, data precision, distance measurement, and proximity interpretation)
and proposed improvements of data and/or model quality along each of these
dimensions. They also argued that if one improves only one of these dimensions
but fails to address the others, the final result may still be problematic. Hence,
they concluded that model extensions are necessary to properly address the issues
of uncertainty and error in this context. Given the development of GIS based
techniques for reducing the effects of uncertainty and error, one would expect many
more such extensions to appear in the future.
19.4.5
Problem Solution
When it comes to solving location science problems, GIS may be utilized in several
ways depending on the nature of the problems in question. If the number of facilities
to be located is limited, as in the location of large infrastructures, then the solution
may be obtained by performing a straightforward suitability analysis through GIS
in order to determine the sites that meet the selection criteria.
GIS prove to be extremely useful for dealing with problems that can be directly
or indirectly solved using certain computational geometry techniques which are
standard tools in GIS. A typical example is the heuristic proposed by Suzuki and
Okabe (
1995
) for the continuous p-center problem. This heuristic relies on the
generation of a Voronoi diagram corresponding to a set of points at each step of
the algorithm. Given that the construction of Voronoi diagrams are standard GIS
functions, it seems natural to use GIS to solve this problem. In fact, Wei et al.
(
2006
) implemented this heuristic using a commercial package to locate emergency
warning sirens. In the same vein, Matisziw and Murray (
2009
) addressed the
continuous coverage problem and proved that the optimal location lies on the medial
axis of the demand area, namely the set of points having more than one closest point
to the demand area boundary. They then used GIS to implement a Voronoi-based
technique for deriving the medial axis.
When the number of feasible locations is significantly large, then a model is
required to determine the sites for the facilities to be located. Combining GIS and
some solution routines, either commercial or custom-made, in a loose coupling
sense implies a significant exchange of input and output files between the two
components and does not really exploit the capabilities of modern GIS. However,
following the rapid developments in GIS, several tight coupling possibilities
have emerged. In particular, several algorithms for solving location problems are
currently available within GIS software packages. The ArcGis Network Analyst
toolbox, the TransCad application modules for Territory Management and Site
Location Modeling and the Location Intelligence module of MapInfo are examples
of packages that offer tools for solving standard location science problems such
as the p-median or the Maximal Coverage Location Problem. Each of these
