Development of a Multi-criteria Spatial Planning Support System
Transcript of Development of a Multi-criteria Spatial Planning Support System
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 1/16
See discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/283155799
Development of a multi-criteria spatialplanning support system for growth potential
modelling in the Western Cape, South Africa
ARTICLE in LAND USE POLICY · JANUAR Y 2016
Impact Factor: 3.13 · DOI: 10.1016/j.landusepol.2015.09.014
READS
21
7 AUTHORS, INCLUDING:
Adriaan Van Niekerk
Stellenbosch University
48 PUBLICATIONS 262 CITATIONS
SEE PROFILE
Danie Du Plessis
Stellenbosch University
20 PUBLICATIONS 34 CITATIONS
SEE PROFILE
Sanette Lacea, Aletta Ferreira
Stellenbosch University
27 PUBLICATIONS 191 CITATIONS
SEE PROFILE
Ronnie Donaldson
Stellenbosch University
63 PUBLICATIONS 166 CITATIONS
SEE PROFILE
All in-text references underlined in blue are linked to publications on ResearchGate,
letting you access and read them immediately.
Available from: Adriaan Van Niekerk
Retrieved on: 02 December 2015
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 2/16
Land UsePolicy 50 (2016) 179–193
Contents lists available at ScienceDirect
Land Use Policy
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / l a n d u s e p o l
Development of a multi-criteria spatial planning support system forgrowth potential modelling in the Western Cape, South Africa
Adriaan van Niekerk a,e,∗, Danie du Plessis b, Ilze Boonzaaiera, Manfred Spocter c,Sanette Ferreira c, Lieb Loots d, Ronnie Donaldsonc
a Centre for Geographical Analysis, Department of Geography and Environmental Studies, Stellenbosch University, Private Bag X1,Matieland 7602, South
Africab Centre for Regional and Urban Innovation and Statistical Exploration, Department of Geography and Environmental Studies, Stellenbosch University,
Private BagX1,Matieland7602, South Africac Department of Geographyand Environmental Studies, StellenboschUniversity, Private BagX1,Matieland7602, South Africad Department of Economics, University of theWestern Cape, Private BagX17, Bellville 7535, South Africae
Schoolof Plant Biology, TheUniversity ofWestern Australia, 35 Stirling Hwy, Crawley WA6009 Perth, Australia
a r t i c l e i n f o
Article history:
Received 16 May 2015
Received in revised form 8 September2015
Accepted 18 September 2015
Keywords:
South Africa
Growth potential
Settlements
Spatial decision support systems
Growth modelling
a b s t r a c t
Growth potential modelling is useful as it provides insight into which settlements in a region are likely
to experience growth and which areas are likely to decline. However, growth potential modelling is an
ill-structured problem as there is no universally-agreed set of criteria (parameters) that can be combined
in a particular way (rules) to provide a definitive growth potential measure (solution). In this paper we
address the ill-structured problem of growth potential modelling by combining multi-criteria decision
making (MCDM), geographical information systems (GIS) and planning support systems (PPS) to gen-
erate a number of growth scenarios for settlements in Western Cape province of South Africa. A new
framework and methodology for selecting, structuring and analysing multiple growth potential criteria
is proposed. The framework, based on the principles of innovation potential and growth preconditions,
was applied to demonstrate how it can be used to identify a series of candidate criteria relating to the
growth potential of settlements. The criteria were subjected to a MCDM process involving criteria selec-tion, weighting and normalisation. Two criteria sets, weighting schemes and normalisation methods were
considered. Two different classification techniques were also evaluated. A total of 16 scenarios were gen-
erated using a newly-developed growth potential PPS (GPPSS). The paper shows how the GPPSS can be
used to quantitatively and qualitatively assess the various scenarios and to select the most appropriate
solution.© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The importance of space andplace in effectivedevelopment pol-
icy is reinforced by the renewed focus on regional development
(Ascani et al., 2012). Continued systematic research on therole and
function of urban settlements within the developmental context
of a region is required to provide a sound foundation to support
well-founded strategic decisions (Pike et al., 2010). Of particular
interest, especially in developing countries, is the identification of
regions or settlements that are most likely to experience sustained
∗ Corresponding author at: Centre for Geographical Analysis, Department of
Geography and Environmental Studies, Stellenbosch University, Private Bag X1,
Matieland 7602, South Africa. Fax: +27 218083109.
E-mail address: [email protected] (A. van Niekerk).
growth and where investment and interventions such as land use
changes and infrastructure projects will have the greatest socio-
economic impact (Henderson and Wang, 2007). Although many
settlements have solid developmental bases and are experiencing
dynamic growth, some are experiencing reduced economic activi-
ties, poor service delivery and deteriorating infrastructure (Bowns,
2013). Decreasing social and economic service levels within settle-
ments invariably impacts negatively on the quality of urban and
rural life as the surrounding hinterland is usually also affected.
Strategic decisions to promote particular types of development in
specific areas require accurate and timely information.
Empirical analyses, such as the application of growth equations
andregression modelling, are often used for estimating the growth
potential of regions (Arbia et al., 2010; Barro, 1991; Battisti and
Vaio, 2008). However, appropriate data for such models is often
notavailable at theappropriate spatial or temporal resolutions.For
http://dx.doi.org/10.1016/j.landusepol.2015.09.014
0264-8377/© 2015 Elsevier Ltd. All rightsreserved.
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 3/16
180 A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193
instance, economic measures such as growth value added (GVA)
are often only available at regional (e.g. provincial or district) level
and are as such not applicable at settlement level. While popula-
tion figures are usually available at the suitable spatial scale (e.g.
ward level), they are in many countries only updated once every
ten years. Trends based on data collected at such long intervals are
often unreliable, particularly in developing countries experienc-
ing high levels of population growth and urbanization. Empirical
models based on the physical growth of settlements are also not
pertinent as many settlements have policies in place that restrict
urban expansion and encourage densification (Musakwa and Van
Niekerk, 2013).
An alternative approach to empirical modelling is to make use
of a range of growth-related factors andto analyse them in a deter-
ministic manner. For instance, Zietsman et al. (2006) showed how
a range of spatial indicators and indexes can be structured into a
framework to model the growth potential of settlements and to
guide spatial development policy. However, identifying suitable
indicators of growth potential is challenging due to the many fac-
tors that may affect a region or settlement’s capacity to develop.
Another problem relating to a deterministic approach to growth
potential modelling is that indicator selection is often subjective
or guided by data availability rather than its true suitability. This
often leads to the selection of indicators that are strongly corre-lated, which canlead to compensabilityproblems. Inaddition, some
factors contributing to growth potential may be difficult to map
or quantify. For instance, entrepreneurial innovation often stimu-
lates economic activities and social development but is extremely
difficult to predict.
The use of empirical and deterministic modelling of growth
potential is conceptually flawed as growth potential modelling
is not a well-structured problem (Saaty, 1978). Well-structured
problems (e.g. mathematics-related problems) have single, cor-
rect and convergent answers, while ill-structured problems do
not have a finite number of concepts, rules, and solutions (Hong,
1998). Ill-structured problems, also referred to as unstructured or
semi-structured (Ascough et al., 2001; Densham, 1991; Goodchild
and Densham, 1990), cannot be solved with an algorithm ora predefined sequence of operations. For instance, there is no
universally-agreed set of criteria (parameters) that can be com-
bined in a particular way (rules) to provide a definitive growth
potential measure (solution). Ill-structured problems may have
multiple solutions, solution paths, and criteria (Kitchener and King,
1981). According to Voss and Post (1988), ill-structured problems
can be solved by (a) representing the problem, (b) stating the
solution and (c) evaluating the results. The representation of the
problem consists of stating the nature of the problem and col-
lecting all appropriate information. During the solution stage of
the problem-solving process, various solutions or scenarios are
generated and selected for evaluation. Evaluation often involves a
process of assessing the solution construction and finding consen-
sual agreement among a community about the most viable, mostdefensible and preferred solution.
Computer systems areoftenused to support the process of find-
ing solutions for ill-structured problems. Decision support systems
(DSS), for example, are computer systems that were specifically
designed to solve such problems. Planning support systems (PSS)
can be regarded as a subset of DSS aimed at bringing together the
functionalities of geographical information systems (GIS), models,
and visualization (DeMers, 2009). The purpose of these assem-
blages is normally to gather, structure, analyse, and communicate
information in planning. PSS are oftenloosely coupled assemblages
of techniques to assist planners, technicians and other role players
involved in the planning process (Tanguay et al., 2010). Although
there are some overlap between GIS, spatial decision support sys-
tems (SDSS) and PSS, the latter can be differentiated based on its
aim that is purely focussed on planning support. Although SDSS
and GIS technologies can normally also be used for planning sup-
port if required,theyarenot solelydedicatedto that use(Vonk etal.,
2007). Theconceptof PSShaveevolved substantially since theearly
urban models of the1960s and1970s that failedto meet theexpec-
tations of users (Batty, 1979) and the introduction of arguments in
the late 80’s for thinking beyondGIS. PSS nowinclude a wide range
of approaches suchas rule-basedaccounting(e.g. Whatif?), cellular
automata (e.g. SLEUTH),and microsimulation (e.g. UrbanSim) mod-
els (Kaza, 2010). One of the six important information-handling
functions of a PSS is information analysis aimed at generating new
information from existing data (e.g. the use of multicriteria analy-
sis systems) which is of particular relevance to this research (Vonk
et al., 2007).
PSS normallyincorporate predictive analysis to present decision
makers with different scenarios to explore the possible effects of
their decisions. This type of interactive exploration enables a deci-
sion maker to develop a better understanding of an ill-structured
problem. PSS normally consist of a database management system,
analytical modelling capabilities, analysis procedures, and a user
interface with display and report generators. GIS are often used in
combination with PSS to find solutions for geographical or spatial
problems (Agrell et al., 2004). With the capabilities of GIS to store,
manipulate, analyse and present spatial data, Spatial PSS are pow-erful tools for supporting complex spatial decisions (Ascough et al.,
2001).
Although numerous methods exist whereby GIS and PSS can
be used to analyse multiple factors and to combine them into a
model (Chang, 2006), the multi-criteria decision making (MCDM)
approach is one of the most popular due to its ability to divide
complex problems into smaller understandable parts that are then
evaluated independently. The results of the individual evaluations
are integrated to provide an overall solution to the original prob-
lem (Malczewski, 1999). By using MCDM, solutions can be found
to decision making problems with multiple alternatives, evaluated
by decision criteria ( Jankowiski and Nyerges, 2001).
This paper adopts the approach suggested by Voss and Post
(1988) f or solving ill-structured problems by combining MCDM,GIS and PSS to generate a number of growth scenarios for settle-
ments in the Western Cape province of South Africa. A framework
and methodology for selecting, structuring and analysing multi-
ple growth potential criteria is proposed. The framework, based on
the principles of innovation potential and growth preconditions, is
applied to demonstrate how it can be used to identify a series of
candidate criteria relating to the growth potential of settlements.
Scenarios are generated using a newly-developed growth poten-
tial PSS (GPPSS). The various scenarios are then quantitatively and
qualitatively evaluated to select the most appropriate solution.
The next section provides an overview of the study area and the
methods that were used to model growth potential at settlement
level. Although the focus of the paper is mainly on methodologi-
cal considerations, a short discussion of the analysis results and itsvalue for regional and local decision support is also provided. The
paper concludes with comments on remaining challenges and how
the modelling methodology can be improved.
2. Methods
2.1. Study area
Donaldson et al. (2012b) evaluated the development potential
of 24 non-metropolitan local municipalities in the Western Cape,
South Africa by employing a range of spatial indicators collected
at municipal level. The resulting indexes and classifications were
analysed andinterpreted to formulate a set of generic interventions
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 4/16
A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193 181
Fig. 1. Western Cape Province of South Africa.
for stimulating development in the non-metropolitan settlementsof the Western Cape (Donaldson et al., 2012a). Therefore, building
on that study’s findings the Western Cape province of South Africa
was chosen for demonstrating how the GPPSS can be used to sup-
port growth potential modelling at settlement level. The Western
Cape is South Africa’s fourth largest province, covering 11% of the
country’s land area (see Fig. 1). In 2011 the province accommo-
dated approximately 5.8 million people, 11.2% of the national total
(Statistics South Africa, 2012). At 129 462km2 it is about the same
size as England or Bangladesh.
The Western Cape is well known for its environmental and bio-
logical diversity. It comprises most of the Cape Floristic Region
(CFR), theonly floralkingdom located entirelywithinthe geograph-
ical confines of one country. The CFR is recognized globally as a
biodiversityhotspotwhichcoversonly 0.05% of theearth’sland sur-face, but as for biodiversity it contains three per cent of the world’s
plant species (Broennimann et al., 2006).
Most (85%) of the province’s economic activities are geograph-
ically concentrated in the Cape Metropolitan Area (CMA) and
the adjacent Cape Winelands District (CWD). According to the
Western Cape Government (2013), finance, insurance, real estate
and business services make up the core of the province’s economy.
Manufacturing is the second largest economic sector with a con-
tribution of 17% of the gross domestic product per region (GDPR).
More than 90% of this activity is concentrated in the CMA, CWD
and Eden districts. Retail, wholesale, catering and accommodation
accounts for 15% of the GDPR, and the government sector con-
tributes 10%.Although the agricultural sector is relativelysmall (4%
of GDPR), it is thebackbone of thenon-metropolitandistricts where
most of thelandis used forcultivation and grazing. It is noteworthythat about 11 million hectares (84%) of the province’s land surface
is currently producing more than 55% of South Africa’s total agri-
cultural exports, of which the principal products are fruit (27%),
winter grain (21%), white meat (18%), wine (18%) and vegetables
(16%) (CNDV Africa, 2005). Currently the Western Cape is gener-
ating more than 20% of South Africa’s gross farming income while
employing one quarter of all farm workers.
The province is experiencing an alarmingly high population
growth rate of 2.86% which is the second highest of the nine
South African provinces (Statistics South Africa, 2012). Popula-
tion growth, together with an urbanization level of 90% (Kok and
Collinson, 2006), are causing increasing needs for housing, employ-
ment and food which place immense pressures on the province’s
natural and human resources. According to Jacobs and Du Plessis(2015) the net interprovincial migration to the Western Cape
between 1991 and 1996 was 133 419 and between 1996 and
2001 a total of 69 321. This figure increased to 192000 between
2001 and2011. In-migrationis mainlydriven by productionism (in
search of employment, education and better services) with most
migrants being unmarried, young (25–29 years) and unemployed
(or not economically active), with low incomes. A large proportion
(31.3%) end up living in informal dwellings in backyards or infor-
mal settlements largely concentrated in the CMA. A smaller but
prominent sub-stream of in-migrants consists of affluent, highly
skilled, mostly married individuals from other metropolitan cities,
especially Gauteng. These migrants are driven by environmental-
ism, and favour the CMA and adjacent municipalities, as well as
the intermediate sized settlements along the south coast. These
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 5/16
182 A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193
migration patternsand population dynamicshold important devel-
opment implications at both provincial and municipal levels.
It is clear that the province’s natural, human and infrastruc-
tural resourcescannotbe managed sustainably without performing
sound spatial planning (Musakwa and Van Niekerk, 2013). Such
planningrequiresaccurate information about the suitability of land
and about the growth trends of settlements and their surrounding
hinterland. This information is also critical for formulating strate-
gies that can spur specific types of development in certain areas or
settlements. Because of the complexities involvedin growth poten-
tial modelling, innovative tools are needed to support decisions
about investment strategies in the province.
2.2. Growth potential modelling using a multi-criteria decision
making approach
The premise of growth potentialmodellingis that, in theabsence
of significant interventions, current and historical information can
be used to predict the future growth of a region or settlement. The
unavailability of empirical data measuring actual development is –
especially in data-scarce developing countries such as South Africa
– the main reason why empirical approaches are not always suit-
able for modelling growth potential. An alternative approach is to
build a model based on human reasoning or expert knowledge. Forinstance,it is well known that drivers of growthare often related to
entrepreneurial innovations thatstimulate economic activities and
social development. Althoughsuch actions cannot be predicted, the
likelihood of innovation and entrepreneurial investment in a set-
tlement is higher if it provides the necessary social capital (Iyer
et al., 2005) and institutional support (Arbia et al., 2010; Rodríguez-
Pose, 2013), while an initiative will only succeed if a settlement can
offer the necessary financial (economic) services, natural resources
and infrastructure (Cloke, 2013). Information relating to innova-
tion potential and growth preconditions can consequently be used
to model a settlement’s potential to attract and sustain future
investments (Zietsman et al., 2006). Such reasoning can be for-
malised as a set of deterministic rules in an expert system or as
factors in a MCDM to model the growth potential of a settlement.MCDM has been used in many types of applications including eco-
nomics (Al-Najjar and Alsyouf, 2003), noise pollution (Van der
Merwe and Von Holdt, 2006), forestry (Bruno et al., 2006; Varma
et al., 2000), conservation (Phua and Minowa, 2005; Wood and
Dragicevic, 2007), flood vulnerability (Yalcin and Akyurek, 2004),
transportation (Vreeker et al., 2002), tourism potential determin-
ing (Van der Merwe et al., 2008), and land use suitability analysis
(Van Niekerk, 2008).
Van der Merwe (1997) suggests a six-step procedure for apply-
ingMCDMfor spatial problems.The first step is to setthe objectives
of the evaluation. These objectives dictate which methodology or
decisionstrategywill be usedin the evaluation (e.g. multi-attribute,
multi-objective, individual,participative, deterministic,probabilis-
tic). In step two of the MCDM process, the appropriate criteria aredefined. Criteria can be either factors or constraints. Factors refer
to criteria that enhance or detract from an objective (e.g. growth
potential of a settlement), while constraints are meant to limit or
exclude cases for consideration(Malczewski, 1999). Once the crite-
ria are selected, the data for each criterion is collected and mapped
(step three), usually using a GIS. Because factors can be continu-
ous and measured in different scales (i.e. nominal, ordinal, interval
and ratio), step four of the MCDM process requires all factors to
be reformatted or normalized to a common measurement scale.
By nature different criteria do not have equal importance for a
particular objective. Access to infrastructure and labour might, for
instance, might be considered more important for spurring indus-
trial development than tourism or agricultural potential. To take
this into consideration, each criterion can be weighted according
to its relative importance. Weights can either be assigned by the
analyst or in consultation with stakeholders. Weight values of cri-
teria range from0 to1 and shouldbe specifiedso thattheirsumis 1.
The final step of the MCDM procedure involves analysingthe crite-
ria to produce composite maps. In MCDM, factors, constraints and
weights are combined using weighted linear combination (WLC).
This essentially involves calculating a composite value for a partic-
ular objective using Eq. (1).
S =
wi xi ×
c j (1)
Where S is the objective value; wi is the weight of factor i; xi is
the criterion score of factor i; c j is the Boolean criterion score of
constraint j; and is the product of criteria.
In contrast to the high-risk Boolean intersect (AND) and union
(OR) operations, WLC produces a risk-averse (Eastman, 2000) and
full trade-off solution (Mahini and Gholamalifard, 2006). If more
control over the level of trade-off is required, ordered weighted
averaging (OWA) can be applied as it employs an additional set of
weights, called order weights, that are assigned on a location basis
to manipulate trade-off (Malczewski, 2006). The result of MCDM
applied in a spatial context is a set of maps showing the relative
scores for each objective.
GIS are often used to conduct MCDM owing to its ability tospatially integrate and compare multiple geographically refer-
enced data sets. GIS can play an instrumental role in four of the
seven MCDM steps (i.e. map spatial criteria, set criteria weights,
multi-criteria evaluation and multi-objective evaluation). Most
GIS software packages provide tools and functions for performing
MCDM. One example is ESRI ArcGIS’ Weighted Overlay function
that allows users to specify a set of criteria (stored as raster layers)
and weights (importance ratings) that are analysed by the tool to
produce an aggregated result (ESRI, 2011).
Given that interactive scenario building is essential for decision
support, especially where objectives are vague and problems semi-
structured (Clarke, 1990), the MCDM procedures can be automated
within a GIS. Thefollowingsectionsexplain how MCDM was imple-
mented for modellingthe growth potentialof the settlements in thestudy area.
2.3. MCDM implementation
2.3.1. Structuring framework
A structuring framework (Table 1) for growth potential mod-
elling was designed as part of the first step of the MCDM
procedure. The framework design was also based on a com-
bination of international indicator guidelines (e.g. the United
Nations Indicators of Sustainable Development) and national gov-
ernmental policy-driven initiatives (e.g. National Development
Plan 2030, National Spatial Development Perspective 2004). The
structuring framework consists of five main themes, namely
human capital, economic, physical-environmental, infrastructural,and governance/institutional and are consistently present in
many of the documentation studied. There is a striking sim-
ilarity between the five identified themes and those used in
the internationally recognisedEnvironmental SustainabilityIndex:
Social/Cultural, Economic, Environmental, Political, and Institu-
tional/Technological. Infrastructure was identifiedas a stand-alone
factor (even though it can be regarded as a cross-cutting factor),
but the focus here was to apply infrastructure as the add-on fixed
production factors to a physical space to enhance its development
value and potential (Wong, 2002). Although recent sustainable
development indicator sets tendto moveaway from the traditional
four “pillars” (social, economic, environmental and institutional)
towards a more multi-dimensional view of sustainable develop-
ment including cross-cutting themes such as poverty and natural
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 6/16
A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193 183
Table 1
Structuring framework for indicator selection.
# Theme Sub-themes Modelling purpose
1 Economic Extent and diversity of retail and services sector
Tourism potential
Economic size and growth
Economic diversity
Market potential
Change in labour force
Property market
Preconditions for
Growth
2 Physical env ironment A vailability and qu ality of water
Natural potential
3 Infrastructure Land availability and use
Transport and communication
Availability of municipal infrastructure
4 Human Capital Poverty and inequality
Human resources
Population structure and growth
Innovation Potential
5 Institutional Quality of governance
Safety and security
Administrative and institutional function
Availability of community and public institutions
Fig. 2. Growth potential index construction.
hazards, these traditional spheres are still used to classify most
indicator sets (Tanguay et al., 2010).
The themes in Table 1 were used as main indices of growth
potential and as a framework for criteria identification. In the con-
text of MCDM, eachthematicindex is considered an objective, each
with a set of criteria. The thematic indices were then combined
to form intermediate indices relating to growth preconditions and
innovation potential (Fig. 2). Together, the intermediate indices
form a composite growth potential index which is the main objec-
tive of the evaluation.
2.3.2. Criteria identification
The framework was used to identify 85 criteria relating togrowth potential. Data availability was a major factor in finding
suitable criteria, mainly because provincial-wide datasets at set-
tlement level are limited. Preference were given to simple and
robust (i.e. statistically validated) indicators that are responsive
to policy interventions and resistant to manipulation. It was also
important that the selected criteria cover as wide as possible spec-
trum of human and economic activities, as well as the bio-physical
conditions of the settlements in the Western Cape, while having
minimal overlap with each other. Quantifiable features that can be
monitored to establish performance trends and that are sensitive
enough to reflect important changes in a settlement’s character-
istics were favoured. The frequency and coverage of the selected
elements were also considered for timely identification of perfor-
mance trends. The structuring framework and indicator selection
guidelines were used to draw up an initial set of candidate crite-
ria shown in Table 2. All criteria were judged to be factors that willeither contribute to or detract fromgrowthpotential (i.e.none were
considered to be absolute constraints to further growth).
2.3.3. Data collection and mapping
Because most of the criteria in Table 2 are spatial in nature, a GIS
was used to capture and manipulate the various datasets involved
(DeMers, 2009). A GIS was also used to analyse the data owing to
its capability to combine multiple disparate datasets in a spatial
manner. Depending on the mapping scale, a town can be repre-
sented by a point (i.e. its centre) or a polygon (i.e. its urban edge).
In addition, because a town is influencedby its surrounding hinter-
land activities, a town can also be defined as a Voronoi (Thiessen)
polygon. The latter ensures that any point within the polygon is
closest to its centre (i.e. town centre). Due to the nature of the datathat was expected to influence the growth potential of towns, it
was recognised that a combination of spatial entities (i.e. centre of
town, urban edge andVoronoi polygon)had to be used to represent
towns. For instance, to calculate a town’s distancefrom majorroads,
the town had to be represented by its centroid (i.e. point). Voronoi
polygons are preferred when the influence of the surrounding hin-
terland, for instance when relating its surrounding agricultural
activities, needs to be calculated. The data consequently dictates
what spatial entities should be used during data preparation, but
for analysis purposes all polygons were converted to points (i.e.
centroids) to enable easier comparison.
For many regional planning and geography applications the
capacity or functional extent of a settlement should be taken into
account when generating Thiessen polygons. Dong (2008) andGong et al. (2012) developed a methodology whereby the size
and shape of a Thiessen polygon can be manipulated (weighted)
according to an attribute of the source dataset (usually points).
This approach was followed to generate the Thiessen polygons (see
Fig. 3) used in this paper. The polygons were weighted according
to the population sizes of the settlements in the Western Cape,
thereby generating a more realistic sphere of influence for each
settlement.
2.3.4. Normalization
Because criteria can be measured in different scales (i.e. nomi-
nal, ordinal, interval and ratio), MCDM requires that all indicators
are reformatted (normalised) to a common scale (Malczewski,
1999). Linear scaling (Eq. (2)) is the preferred normalization
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 7/16
184 A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193
Table 2
Factors considered for each index.
Human capital index
Average percapita income2011 (Rands) a , c, d
% of population receiving social grantsb
% of households living in informal housing 2011b
% change in economic empowerment 2001–2011a , c
Overcrowding 2011b
% Unemployment 2011 b , c, d
Matric pass rate 2012 (%) a , c
% 20–65 years old with primary or noeducation 2011b
% 20–65 year olds with at least grade 12 and highera , c
Ratio non-economically active population age 2011b , c , d
% Population growthrate 2001–2011a
% In-migrants past 10 years 2011a
Economic index
Tourism potential 2008 a , c
% Growth of economically active population 2001–2011 a , c, d
Growth in town extent05–08 (ha)a
Growth in town extent08–11 (ha)a
Growth in town extent05–11 (ha)a
Distance toPE, CT and 6 leader towns b , c
Total personal income2011 (Randsmillion) a , c
% Growth in highlyskilled labour 2001–2011 a , c
Number of property transactions 2010a
Value of property transactions 2010a , c
Property tax revenue 2010a , c
Numberof formal retail outlets and servicesector businesses 2010a , c , d
Numberof formal retail outlets and servicesector businesses 2010 per persona
Bio-physical index
Numberof vacant residential stands2010a
Numberof vacant business stands 2010a
Numberof vacant industrial stands 2010a
Mean annual precipitation a , c , d
Projected short term (2020) surplus/shortfalls of peak summer GAADD considering internal reticulation storage 2011 (mcm/a) a , c
Projected medium term (2025) surplus/shortfalls of GAAD under high growth scenario plus 100% of future developmentsrealised2011 (million m3/a)a
Groundwater availability 2011 (mcm/a) a , c
Groundwater quality 2011 b , c, d
Potential evaporation (mm) b , c
Grazing capacitya , c
% Area cultivated2012a , c, d
Growth in% area cultivated (2007–2012)a , c Size and status of unexploited minerals 2010a , c
Groundwater quality (EC) 2011 (Ms/m)b
Biodiversityb
Infrastructure index
% Households with accessto theinternet2011a , c , d
Distance to nearest scheduled airportb , c
Distance to nearest commercial harbourb , c
Distance to nearest small harbour and slipwaysb , c
Accessto main and national roadsa , c, d
Access to railwaysa , c, d
% households with access to cellphone2011a , c
% households with access to sanitation (flush)2011a , c
% households with access to electricity (lighting) 2011a , c
% households with access to waste removal 2011a ,c
WWTW spare capacity perperson 2011 (l/day/pop)a , c, d
State of WWTW infrastructure 2011a , c, d
Institutional index
Management experience and capacity 2010a , c
Debtors ratio 2010b
Debt rate 2010b
Qualified audits 2012a , c, d
Infrastructure backlog reduction 2010a , c, d
OPEX percapita 2010a
CAPEX per capita 2010a
Staff per cap ratio 2010b , c
% Posts filled 2010a , c , d
% Crime (all) occurrences change2009–2012b , c , d
% Contact crime occurrences change 2009–2012b
% Property crime occurrences change 2009–2012b
Crime (all) occurrences (09–12) per 100,000 populationb , c
Contact crime occurrences (09–12) per 100,000 populationb
Property crime occurrences (09–12) per 100,000 populationb
Small business support 2010a , c
Voter turnout 2010a
Numberof Amenities 2010a , c , d
Number of Social service organisations 2010a , c
a Factorcontributed positively to theindex (i.e. high values are preferred).b Factorcontributed negatively to theindex (i.e. low values arepreferred).c Factorused in Criteria Subset A (see Section 2.3.5).d Factorused in Criteria Subset B (see Section 2.3.5).
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 8/16
A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193 185
Fig. 3. Weighted Voronoi polygons used as basis fordata collection andanalysis at settlement level.
method applied in MCDM as it uses the range of a variable as basisfor standardization. This is especially useful when different vari-
ables are combined using weights (levels of importance) as the
range of the outcome can be determined prior to the evaluation
(Van der Merwe et al., 2013; Van der Merwe and Van Niekerk,
2013). An alternative approach is the use of standardized z -scores
(Eq. (3)), which compares each raw value in a dataset to the mean
of the dataset and sets the standard deviation equal to 1. Both of
these normalization methods were implemented in the GPPSS. For
implementation andscenario-buildingpurposes, linear scaling was
denoted as Normalization Method A, while z -score standardization
was referred to asNormalization MethodB.
X i =Ri − Rmin
(Rmax − Rmin) ×m (2)
where: X i is the standardised score; Ri is the raw score; Rmin rep-
resents the minimum score; Rmax is the maximum score; and M is
an arbitrary multiplier representing the upper standardised range
value.
z ik =( xik − ¯ xk)
sdk(3)
where, z ik is the standardised score (also called z -score); xik is the
raw value of variable k for settlement i; ¯ xk is the mean value of
variable k forall settlements in the province;and sdk is the standard
deviation of variable k.
A mechanism whereby criteria can be inverted was also imple-
mented as some criteria may have a positive or negative effect on
growth potential. For instance, a lowcrime rate is expected to have
a positive effecton growth potential, while a population with a lowlevel of education would be considered negative. Factors with pos-
itive and negative influence on growth potential differentiated in
Table 2 wi t h a + o r− sign. The GPPSS was designed to process such
cases by applying Eq. (4) in cases where a criterion has a negative
effect.
Y i = 1 − X i (4)
where:Y i is the inverted score; and X i represents the original score.
2.3.5. Criteria subset selection
A subset of the initial criteria set was selected through a par-
ticipatory process involving structured meetings with municipal
representatives and provincial government officials, as well as
other stakeholders (e.g. urban and regional planners, businessowners). Eight open days were also held in each of the district
municipalities of the Western Cape to raise awareness of the
research and to elicit comment from the general public. Some of
criteria were considered to be dated while others were eliminated
on the grounds of being poor reflections of growth potential. These
included the indicators relating to the physical growth of settle-
ments as it was argued that some settlements have densification
policies in place to limit physical urban expansion.1
1 A settlement’s rate of physical expansion is notnecessarily on par with itseco-
nomicor populationgrowth rate. Stellenbosch,for example, experienced an annual
population growth rate of 8.5% from 2000 to 2010, while its physical expansion in
the sameperiod averaged at 2.8% per annum (Musakwa and Van Niekerk, 2013).
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 9/16
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 10/16
A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193 187
Table 3
Growth potential scenarios generated by setting different model settings.
Scenario # Scenario code Criteria subset Weighting scheme Normalisation method Classification technique
1 AAAA A A A A
2 AABA A A B A
3 ABAA A B A A
4 ABBA A B B A
5 BAAA B A A A
6 BABA B A B A
7 BBAA B B A A8 BBBA B B B A
9 AAAB A A A B
10 AABB A A B B
11 ABAB A B A B
12 ABBB A B B B
13 BAAB B A A B
14 BABB B A B B
15 BBAB B B A B
16 BBBB B B B B
normalization method; (4) modifying the relative weights of indi-
vidual criteria; and (5) selecting a different classification method.
The system can also be set up to iteratively apply various permu-
tations of settings to produce different scenarios. To demonstrate,
16 different scenarios were generated in this paper by applying
two subsets of criteria, two weighting schemes, two normalization
methods and two classification techniques (Table 3).
The classifications of all 16 scenarios were mapped and statis-
tically compared. The median growth potential classification per
settlement (MEDIAN ) was first calculated to represent the overall
(consensus) classification of a settlement. The median instead of
the mean was used as it is known to be less sensitive to outliers
(Pearson, 2002). The purpose of the SD score is essentially to quan-
tify the level of disagreement between the different scenarios for
a particular settlement. If the SD is very high (e.g. more than 1)
it would indicate that, for the particular settlement, there is not a
good agreement (consensus) between the methods used to model
its growth potential. Settlements with high SD may consequently
be considered to be more sensitive to the methodological approach
used, which suggests a level of uncertainty of the classification of that settlement.
Pearson’s two-tailed bivariate correlation analyses were carried
out in IBM SPSS v22 software and used to determine how closely
the growth potential classifications of each scenario agree to the
consensus (median) classification. A correlation score (COR) was
recorded for each scenario.
The SD score, MEDIAN , and classification results of all 16 sce-
narios were mapped to enable spatial comparison. The results
were presented to a group of stakeholders, including govern-
ment officials, local and district municipal representatives, town
and regional planners, social scientists, environmental managers,
economists andgeographers.Stakeholderswere requestedto inter-
pret the scenarioresults and to comment on how theclassifications
compare to their growth expectations for individual settlements.TheSD scores,CORvalues andthe outcome of thequalitative assess-
ment were considered in the selection of the most appropriate
solution (Voss and Post, 1988).
3. Results
3.1. Classifications
The classifications of the 16 growth potential scenarios are
shown in Table A1. Overall there is a good classification agreement
between the different scenarios, with 24 (21%) of the settlements
having a SD score of zero. This implies that, for 24 settlements, all
the growth potential classifications were identical no matter which
parameter set (combination of criteria subset, weighting scheme,
normalization method and classification technique) was used. A
total of 71 (61%) settlements had a SD of 0.5 or less, indicating that
adjustments to the parameters resulted in only minor differences
in the classification outputs of the majority of settlements.
The spatial representation of the MEDIAN result and SD scores
are shown in Fig. 4. A number of observations can be made from
these results. Many of thesettlements that were classifiedin Fig.4a
as having a high or very high growthpotential are clusteredaround
the City of Cape Town, most likely influenced by their proximity to
the metropolis. This cluster includes the towns Malmesbury, Paarl
and Stellenbosch, which were all classified as having a very high
growth potential. A second cluster of very high and high potential
settlements occurs in the Saldanha Bay region, with Vredenburg
and Langebaan (very high growth potential) acting as the main
nodes. The third cluster of towns with high and very high growth
potential is located along the coast of the Overstrand municipal
area, in particular Betty’s Bay, Pringle Bay and Hermanus. A fourth
cluster of high potential municipalities and settlements are located
along the Garden Route, withMosselbaai, George and Knysna beingclassified as having a very high growth potential and Plettenberg
Bay as high. Most of the settlements in the interior, specifically
the Karoo region, were classified as having a limited (i.e. very low
or low) growth potential, the only exception being Oudtshoorn
which received an overall classification of 3.5 (High). In terms
of SD scores (Fig. 4b) it seems that there was better agreement
(low SD scores) between the classifications of the larger settle-
ments, while smaller rural settlements generally recorded larger
(e.g. 0.6 or more) SD scores. Examples include Lutzville, Redel-
inghuys, Aurora, Botrivier, Gansbaai/Franskraalstrand, Suurbraak,
Jongensfontein, Friemersheim and Rheenendal.
3.2. Sensitivity analysis
The correlation analysis results (see bottom of Table A1)
revealed that most of the scenarios were highly correlated with
the MEDIAN classification, with 12 (75%) having a correlation of
more than 0.9 ( p< 0.001). Scenario 1 (AAAA) achieved the high-
est correlation (0.963) with theMEDIAN classification, although its
correlation difference to Scenarios 2 and 6 is marginal. The lowest
correlations (i.e. largest deviation) were recorded for Scenarios 7,
8, 15, and 16 with all of them having correlations of less than 0.9
( p< 0.001). However, given that correlationsof 0.7 or more aregen-
erally considered to be very strong, it can be concluded that all of
the scenarios produced classifications that are in high agreement.
This demonstrates that the methodology is relatively insensitive to
different sets of criteria and weighting schemes.
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 11/16
188 A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193
Fig. 4. Geographic comparison of the(a) MEDIAN (consensus) and (a) SD (disagreement) scores per settlement.
3.3. Qualitative evaluation
These findings were subjected to a rigorous publicparticipation
process consisting of several brainstorming workshops. The results
were also made available to the officials and general public of all
of the 26 municipalities in the Western Cape province (see Fig. 1).
Maps (Figs. 5 and A1 ) of all the scenarios were also produced to
support the evaluation process. Good feedback was received from
the stakeholders and participants and the general agreement was
that, compared to the other scenarios (Fig. A.1), Scenario 1 (Fig. 5)
provided the most meaningful results. Some concerns about the
classifications of a small number of towns were raised, but most
were dispelled when the data (criteria) used in the modellingwere
scrutinized. The growth potential classifications of the settlements
in Fig. 5 were perceived by many to be a true reflection of input
data(see Section 4), which confirmed the quantitative (correlation)
analysis results.
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 12/16
A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193 189
Fig. 5. Classification results of Scenario 1 (AAAA).
4. Discussion
Growth potential modelling is useful as it provides insight into
which settlements in a region are likely to experience growth and
which areas are likely to decline. Such information can be used to
support investment decisions relating to infrastructure develop-
ment and social welfare support. As demonstrated in this paper,
growth potential modelling is an ill-structured problem as it has
multiplepossiblesolutions, solutionpaths, andcriteria. MCDMpro-vides a logical framework for analysing and aggregating the large
number of factors that affect growth potential. GIS are very effec-
tive for preparing, analysing and presenting the various datasets
and criteria, but are not flexible enough for interactive or auto-
mated scenariogeneration.The GPPSS enabled theconstructionof a
series of growth potential modellingscenarios or the Western Cape
province that can help government officials, scientists and other
stakeholders to gain a regional outlook of development trends. The
generation of scenarios also reduces the risks associated with the
methodological uncertainties of growth potentialas it canhighlight
sensitivity to specific parameters. This is of particular importance
for growth potential modelling for which there is no universally
acceptable methodology. In this paper only two subsets of criteria
were considered in the scenario generation process. However, theGPPSS canpotentially automatically generatehundredsof different
subsets. The user can also interactivelyselect or deselect individual
factors to see what influence such changes will have of the results
(i.e. carry out a data sensitivity analysis). Another benefit of the use
of the GPPSS is that the growth potential modelling can be easily
updated by simply editing or replacingthe datasets associatedwith
the individual factors.
When applied to the data of the Western Cape the GPPSS
produced meaningful growth potential classifications. Through a
process of scenario building and comparison it was determined
that there is good agreement between different parameter sets.
This is an indication that the classifications are good reflections of
the underlying data and that they are not significantly affected by
methodological variations. This finding is of great value because it
increases the confidence in the modelling process and creates the
perception that the growthpotential results are good reflections of
the underlying data.
The bivariate correlation analyses revealed that the combina-
tionof parametersused for Scenario1 produceda classificationthat
was very similar to the MEDIAN result. This scenario incorporates
all of the criteria (Criteria Set A), applies equal weights (Weighting
Scheme A), performs linear scaling (Normalization Method A), andemploys natural breaks classification (Classification Technique A)
to generate the growth potential classifications. The classification
result of Scenario 1 was compared to other scenarios that received
relatively high (>0.95) COR scores. This evaluation focussed on the
settlements with high SD scores as the classifications under the
different parameter scenarios are most likely to deviate in these
towns. For instance, Suurbraak received an SD score of 0.75 and
was classified as having a low growth potential (2) by Scenarios 1,
2,5, 6,8, 9,12 and 13, while inScenarios3, 4,10,11,12 and 14it was
classified as having a medium potential (3). In three cases (Scenar-
ios 7, 15, 16) Suurbraak received a growth potential classification
of very low (1). The general agreement amongst stakeholders was
that this small isolated town has little prospect of dramatic devel-
opment and that a low growth potential is the most appropriateclassification. Similarly,when the criteria of Rheenendal (SD=0.97)
is inspectedit is clear that the medium(3) classification of Scenario
1 is the best reflection of its true growth potential. This settle-
ment is located close to the rapidly growing towns of Knysna and
George andhas experienced a 127% growthin highlyskilled labour
between 2001 and 2011. It also has good access to high quality sur-
face and ground water; is close to airports and harbours; and has a
relatively low and declining crime rate. These factors contributed
to its medium growthpotential classification in the majority of the
scenarios (2-6, 9, 10, 13, 14, 16).
Similar evaluations were carried out for various other settle-
ments. The conclusion from these assessments was that Scenario 1
consistently produced sensible classification results. This observa-
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 13/16
190 A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193
tion, in combination with the fact that this scenario hadthe highest
COR score, led to the selection of Scenario 1 as the most appropri-
ate growth potential classification model for the non-metropolitan
settlements of the Western Cape.
The GPPSS is not without limitations. Althoughits development
within a desktop GIS environment enabled rapid development,
the required software license makes it only accessible to those
with the required license. The system’s graphical interface is very
rudimentary as it was designed as a research tool and not for gen-
eral distribution. Users will also require familiarity with ArcView
software. A logical extensionof this research would be the redevel-
opment of the GPPSS as a web application that is accessible over
the Internet.
5. Conclusions
The dynamics and intricacies of the problems and challenges
relating to settlements in recession must be approached in a coor-
dinated manner. Investment strategies should be put into place to
accelerate development in settlements with highgrowth potential,
whilst ensuring sustained social and institutional support to those
living in regions with relatively low growth potential. Develop-
ment policies should direct specific types of investments to certain
areas or settlements. Industrial development should, for instance,
not be encouraged in settlements reliant on tourism or in areas
that are environmentally sensitive. To do the above state institu-
tions need an appropriate methodology to inform policy decision
management.
This paper combined MCDM, GIS and SDSS to generate a num-
ber of growth scenarios for settlements in Western Cape province
of South Africa. A new framework and methodology for select-
ing, structuring and analysing multiple growth potential criteria
was proposed. The framework, based on the principles of innova-
tion potential and growth preconditions, is applied to demonstrate
how it can be used to identify a series of candidate criteria relat-
ing to the growth potential of settlements. Scenarios are generated
using a newly-developed Growth Potential SDSS (GPPSS). The vari-
ous scenarios were then quantitatively and qualitatively evaluated
to select the most appropriate solution. The scenario-building also
highlighted the sensitivity of growth potential models to variations
in parameters such as the criteria set, weighting scheme, normal-
isation technique and classification method. Although the GPPSS
was specifically developed for the Western Cape province, it can
be applied for any group of settlements for which suitable data is
available. It can also be applied on other spatial entities such as
wards, municipalities, districts, counties, countries and region as
long as the data is available in GIS format.
Acknowledgements
We thank the Department of Environmental Affairs and Devel-
opment Planning of the Western Cape Government for providing
financial support for this study.
Appendix A
Table A1
Settlement growthpotential classifications forall scenarios (1= Very Low; 2 = Low; 3 = Medium; 4 =High; 5 = Very High).
Settlement AELN AEZN AULN AUZN CELN CEZN CULN CUZN AELQ AEZQ AULQ AUZQ CELQ CEZQ CULQ CUZQ MEDIAN SD
Albertinia 3 3 3 3 3 3 2 2 3 3 3 3 3 3 2 2 3 0.50
Arniston 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 2 2 0.25
Ashton 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 0.00
Aurora 2 2 2 2 2 2 3 3 2 2 1 1 2 2 4 3 2 0.75
Barrydale 3 3 3 3 3 3 2 2 3 3 3 3 3 3 2 2 3 0.50Beaufort West 2 2 3 3 2 2 2 2 2 2 3 3 2 2 2 2 2 0.50
Betty’s /Pringle Bay 5 5 5 5 5 5 5 4 5 5 5 5 5 5 5 5 5 0.25
Bitterfontein 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.00
Bonnievale 3 2 2 2 3 2 2 2 3 3 2 2 3 3 3 2 2 0.66
Botrivier 3 3 3 3 3 3 4 4 4 3 4 4 4 3 5 5 3.5 0.71
Bredasdorp 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 4 3 0.25
Buffelsbaai 3 3 4 4 3 3 3 4 4 4 4 4 4 4 4 4 4 0.56
Caledon 4 4 4 4 4 4 5 4 5 4 4 4 5 4 5 5 4 0.56
Calitzdorp 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.00
Ceres 4 4 3 4 4 4 3 4 4 5 4 4 4 5 4 4 4 0.50
Citrusdal 2 2 2 2 2 2 3 3 2 2 2 2 2 2 3 3 2 0.50
Clanwilliam 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.00
Darling 3 3 3 3 3 3 4 3 4 4 4 4 4 4 5 4 4 0.71
De Doorns 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.00
De Rust 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.00
Doringbaai 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.00
Dwarskersbos 3 3 3 3 3 3 3 3 3 3 3 3 3 3 4 4 3 0.35Dysselsdorp 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0.00
Ebenhaesar 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0.00
Eendekuil 3 3 2 2 3 3 3 3 3 3 2 2 3 3 4 4 3 0.61
Elandsbaai 3 2 2 2 3 2 3 3 3 3 2 2 3 3 3 3 3 0.61
Elim 2 2 2 2 2 2 1 2 1 2 2 2 1 2 1 1 2 0.56
Franschhoek 3 3 4 4 3 3 3 3 4 4 5 4 4 4 3 4 4 0.71
Friemersheim 3 3 3 3 3 3 1 2 3 3 3 3 3 3 1 1 3 0.90
Gansbaai 4 4 4 4 4 4 3 3 5 5 5 5 5 5 3 3 4 0.79
Genadendal 3 3 3 3 3 3 3 3 4 4 4 4 4 4 3 3 3 0.61
George 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 0.00
Goedverwacht 2 2 2 2 2 2 3 3 2 2 2 2 2 2 3 3 2 0.50
Gouda 3 3 2 2 3 3 2 2 3 3 2 3 3 3 2 2 3 0.66
Gouritsmond 3 3 3 3 3 3 3 3 4 4 4 4 4 4 3 3 3 0.61
Graafwater 2 2 2 2 2 2 2 2 2 2 2 1 2 2 3 2 2 0.35
Grabouw 4 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 4.5 0.50
Greyton 3 3 3 3 3 3 4 4 4 4 4 4 4 4 5 4 4 0.66
Jongensfontein 3 3 4 4 3 3 2 2 4 4 4 4 4 4 2 2 2 0.66
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 14/16
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 15/16
192 A. van Niekerk et al. / Land Use Policy 50 (2016) 179–193
Fig. A1. Classification results of Scenarios 2–16.
References
Acton,C., Miller, R., Fullerton, D., Maltby, J., 2009. SPSS forSocial Scientists, 2nd ed.Palgrave Macmillan, Hampshire.
Agrell, P.J., Stam, A., Fischer, G.W., 2004. Interactive multiobjective agro-ecologicalland useplanning: the Bungoma region in Kenya. Eur. J. Oper. Res. 158,194–217.
Al-Najjar, B.,Alsyouf, I., 2003. Selecting the most efficient maintenance approachusing fuzzy multiple criteria decision making. Int. J. Econ. 84, 85–100.
7/23/2019 Development of a Multi-criteria Spatial Planning Support System
http://slidepdf.com/reader/full/development-of-a-multi-criteria-spatial-planning-support-system 16/16