International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004
models evaluation because we don't know about any factor
maps interactions such as Environment Parameters and natural
factors.
2.3.4 Genetic Model: Before describing Genetic execution
routine, it is necessary to introduce some corresponsive
concepts as below.
Choromosom : It displays one possible solution for a required
problem. In site selection problem each set of selected spatial
units are primary solutions for industrial estates location and
indicates one Choromosom. Each unite in mentioned set is
called Gene (Gen, 1997).
Population : The possible amounts of Choromosoms in each
generation, correspond to the population. On the other hand,
population is a set of solutions (Gen, 1997). In industrial
estates site selection, all selected spatial units of corresponsive
layer (primary solutions) are called initial population.
Fitness Degree : With comparing population of new and
previous generation, fitness degree is determined using fitness
function. This parameter identifies surviving feasibility in the
next generation for all of choromosoms (Gen, 1997). For
Genetic model execution in industrial estates site selection
chromosomes spatial units values in the new generation are
compared with corresponding spatial units values in previous
generation. Then, fitness function is Algebraic comparing
statement that identifies strong genes for industrial estate
construction by cutting poor genes out that aren’t accordant to
function.
Control Parameters : They are specific conditions such as
number of generations, number of choromosoms, length of
choromosoms and model terminating condition that caused
upper efficiency on model execution routine (Gen, 1997).
In Genetic model first, initial population are selected. Then,
factor maps are integrated based on definite integration
operators (such as index overlay) and second generation is
generated. After that, with using fitness degree, poor genes are
appointed and eliminated. Finally, new generation is selected
as initial population and above procedure is repeated until
reaching constant state. We can decrease repetitions by
entering control parameters in the execution routine.
Therefore, this model is selected for our evaluation because site
selection is a complex and composite problem with different
criteria and conditions and genetic algorithm is a good model
for solving these kinds of problems.
3 EXECUTION AND EVALUATION OF MODELS
In this section, we describe regular steps that identify the
optimum model for industrial estates site selections as shown in
following flowchart.
Study Area
v
Determination of Efficient Criteria and Factors
Y
Data Preparation as Factor Maps
v
Data Weighting
v
Models Execution
Y
Models Evaluation
v
Finding Optimum Model and Location
Figure 1. Models execution and evaluation steps
3.1 Study Area
Study area characteristics affect on industrial estates factors and
criteria. More features diversity and spatial data density
increase factors and criteria. For example, political factors are
important on boundaries of cach country. In the case of
political boundary of study area, we usually consider political
conditions. Therefore, its complete recognition causes better
selection of effective factors and criteria.
This area is selected based on some parameters such as
economic factors, politic strategies, data limitations and land
physical attributes (Demers, 1992). According to experts' ideas
of Iranian industrial estates company and necessary factors,
first, we selected Azarbaijan province at the north-west of Iran.
Then, a zone that covers a region of 1959.59 (km^2) area and
located around Tabrize (capital of the east Azarbaijan province)
was used as our study area. According to Figure 2, there are
good density of data in our selected study area.
= Cay
Raver
Wazeresourte
»
Pu-fhceisymsoli
Casstagen
4 Powerdine
Rxfettro-station
+ Forcesation
» Medical<erterirymdel;
Eœutationat<entetsymboli
1/2 Bingieratezy
nr
^
Ji Man-read
447 tiger
Gravetrere
ZZ Frene ay
ANNE
Arphot-reasz
Figure 2. Selected study area
This area consist of different GIS ready layers with scale of
1:25000 produced by NCC (National Cartographic Center of
Iran) and edited to be GIS ready data by K.N.Toosi University
of Technology.
By
inc
an