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: avn@sun.ac.za (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