See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/283155799

Development of a multi-criteria spatial

planning support system for growth potential
modelling in the Western Cape, South Africa
ARTICLE in LAND USE POLICY JANUARY 2016
Impact Factor: 3.13 DOI: 10.1016/j.landusepol.2015.09.014

READS

21

7 AUTHORS, INCLUDING:
Adriaan Van Niekerk

Danie Du Plessis

Stellenbosch University

Stellenbosch University

48 PUBLICATIONS 262 CITATIONS

20 PUBLICATIONS 34 CITATIONS

SEE PROFILE

SEE PROFILE

Sanette Lacea, Aletta Ferreira

Ronnie Donaldson

Stellenbosch University

Stellenbosch University

27 PUBLICATIONS 191 CITATIONS

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.

SEE PROFILE

Available from: Adriaan Van Niekerk

Retrieved on: 02 December 2015

Land Use Policy 50 (2016) 179193

Contents lists available at ScienceDirect

Land Use Policy

journal homepage: www.elsevier.com/locate/landusepol

Development of a multi-criteria spatial planning support system for

growth potential modelling in the Western Cape, South Africa
Adriaan van Niekerk a,e, , Danie du Plessis b , Ilze Boonzaaier a , Manfred Spocter c ,
Sanette Ferreira c , Lieb Loots d , Ronnie Donaldson c
a
Centre for Geographical Analysis, Department of Geography and Environmental Studies, Stellenbosch University, Private Bag X1, Matieland 7602, South
Africa
b
Centre for Regional and Urban Innovation and Statistical Exploration, Department of Geography and Environmental Studies, Stellenbosch University,
Private Bag X1, Matieland 7602, South Africa
c
Department of Geography and Environmental Studies, Stellenbosch University, Private Bag X1, Matieland 7602, South Africa
d
Department of Economics, University of the Western Cape, Private Bag X17, Bellville 7535, South Africa
e
School of Plant Biology, The University of Western Australia, 35 Stirling Hwy, Crawley WA 6009 Perth, Australia

a r t i c l e

i n f o

Article history:
Received 16 May 2015
Received in revised form 8 September 2015
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 denitive 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 generate 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 selection, weighting and normalisation. Two criteria sets, weighting schemes and normalisation methods were
considered. Two different classication techniques were also evaluated. A total of 16 scenarios were generated 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 and place in effective development policy is reinforced by the renewed focus on regional development
(Ascani et al., 2012). Continued systematic research on the role 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 identication 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).
http://dx.doi.org/10.1016/j.landusepol.2015.09.014
0264-8377/ 2015 Elsevier Ltd. All rights reserved.

growth and where investment and interventions such as land use

changes and infrastructure projects will have the greatest socioeconomic impact (Henderson and Wang, 2007). Although many
settlements have solid developmental bases and are experiencing
dynamic growth, some are experiencing reduced economic activities, poor service delivery and deteriorating infrastructure (Bowns,
2013). Decreasing social and economic service levels within settlements 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
specic areas require accurate and timely information.
Empirical analyses, such as the application of growth equations
and regression 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
not available at the appropriate spatial or temporal resolutions. For

180

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

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 population gures 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 experiencing 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 densication (Musakwa and Van
Niekerk, 2013).
An alternative approach to empirical modelling is to make use
of a range of growth-related factors and to analyse them in a deterministic 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 factors that may affect a region or settlements 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 correlated, which can lead to compensability problems. In addition, some
factors contributing to growth potential may be difcult to map
or quantify. For instance, entrepreneurial innovation often stimulates economic activities and social development but is extremely
difcult to predict.
The use of empirical and deterministic modelling of growth
potential is conceptually awed as growth potential modelling
is not a well-structured problem (Saaty, 1978). Well-structured
problems (e.g. mathematics-related problems) have single, correct and convergent answers, while ill-structured problems do
not have a nite 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 or
a predened sequence of operations. For instance, there is no
universally-agreed set of criteria (parameters) that can be combined in a particular way (rules) to provide a denitive 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 collecting 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 nding consensual agreement among a community about the most viable, most
defensible and preferred solution.
Computer systems are often used to support the process of nding solutions for ill-structured problems. Decision support systems
(DSS), for example, are computer systems that were specically
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 assemblages is normally to gather, structure, analyse, and communicate
information in planning. PSS are often loosely 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 systems (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 support if required, they are not solely dedicated to that use (Vonk et al.,
2007). The concept of PSS have evolved substantially since the early
urban models of the 1960s and 1970s that failed to meet the expectations of users (Batty, 1979) and the introduction of arguments in
the late 80s for thinking beyond GIS. PSS now include a wide range
of approaches such as rule-based accounting (e.g. What if?), cellular
automata (e.g. SLEUTH), and microsimulation (e.g. UrbanSim) models (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 analysis systems) which is of particular relevance to this research (Vonk
et al., 2007).
PSS normally incorporate predictive analysis to present decision
makers with different scenarios to explore the possible effects of
their decisions. This type of interactive exploration enables a decision 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 nd 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 powerful 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 problem (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) for solving ill-structured problems by combining MCDM,
GIS and PSS to generate a number of growth scenarios for settlements in the Western Cape province of South Africa. A 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, 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 potential 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 methodological considerations, a short discussion of the analysis results and its
value 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 classications were
analysed and interpreted to formulate a set of generic interventions

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

181

Fig. 1. Western Cape Province of South Africa.

for stimulating development in the non-metropolitan settlements

of the Western Cape (Donaldson et al., 2012a). Therefore, building
on that studys ndings the Western Cape province of South Africa
was chosen for demonstrating how the GPPSS can be used to support growth potential modelling at settlement level. The Western
Cape is South Africas fourth largest province, covering 11% of the
countrys land area (see Fig. 1). In 2011 the province accommodated approximately 5.8 million people, 11.2% of the national total
(Statistics South Africa, 2012). At 129 462 km2 it is about the same
size as England or Bangladesh.
The Western Cape is well known for its environmental and biological diversity. It comprises most of the Cape Floristic Region
(CFR), the only oral kingdom located entirely within the geographical connes of one country. The CFR is recognized globally as a
biodiversity hotspot which covers only 0.05% of the earths land surface, but as for biodiversity it contains three per cent of the worlds
plant species (Broennimann et al., 2006).
Most (85%) of the provinces economic activities are geographically concentrated in the Cape Metropolitan Area (CMA) and
the adjacent Cape Winelands District (CWD). According to the
Western Cape Government (2013), nance, insurance, real estate
and business services make up the core of the provinces economy.
Manufacturing is the second largest economic sector with a contribution 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 contributes 10%. Although the agricultural sector is relatively small (4%
of GDPR), it is the backbone of the non-metropolitan districts where

most of the land is used for cultivation and grazing. It is noteworthy

that about 11 million hectares (84%) of the provinces land surface
is currently producing more than 55% of South Africas total agricultural 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 generating more than 20% of South Africas 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). Population growth, together with an urbanization level of 90% (Kok and
Collinson, 2006), are causing increasing needs for housing, employment and food which place immense pressures on the provinces
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 gure increased to 192000 between
2001 and 2011. In-migration is mainly driven by productionism (in
search of employment, education and better services) with most
migrants being unmarried, young (2529 years) and unemployed
(or not economically active), with low incomes. A large proportion
(31.3%) end up living in informal dwellings in backyards or informal settlements largely concentrated in the CMA. A smaller but
prominent sub-stream of in-migrants consists of afuent, highly
skilled, mostly married individuals from other metropolitan cities,
especially Gauteng. These migrants are driven by environmentalism, and favour the CMA and adjacent municipalities, as well as
the intermediate sized settlements along the south coast. These

182

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

migration patterns and population dynamics hold important development implications at both provincial and municipal levels.
It is clear that the provinces natural, human and infrastructural resources cannot be managed sustainably without performing
sound spatial planning (Musakwa and Van Niekerk, 2013). Such
planning requires accurate information about the suitability of land
and about the growth trends of settlements and their surrounding
hinterland. This information is also critical for formulating strategies that can spur specic types of development in certain areas or
settlements. Because of the complexities involved in growth potential 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 potential modelling is that, in the absence
of signicant 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 suitable for modelling growth potential. An alternative approach is to
build a model based on human reasoning or expert knowledge. For
instance, it is well known that drivers of growth are often related to
entrepreneurial innovations that stimulate economic activities and
social development. Although such actions cannot be predicted, the
likelihood of innovation and entrepreneurial investment in a settlement is higher if it provides the necessary social capital (Iyer
et al., 2005) and institutional support (Arbia et al., 2010; RodrguezPose, 2013), while an initiative will only succeed if a settlement can
offer the necessary nancial (economic) services, natural resources
and infrastructure (Cloke, 2013). Information relating to innovation potential and growth preconditions can consequently be used
to model a settlements potential to attract and sustain future
investments (Zietsman et al., 2006). Such reasoning can be formalised 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 economics (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), ood vulnerability (Yalcin and Akyurek, 2004),
transportation (Vreeker et al., 2002), tourism potential determining (Van der Merwe et al., 2008), and land use suitability analysis
(Van Niekerk, 2008).
Van der Merwe (1997) suggests a six-step procedure for applying MCDM for spatial problems. The rst step is to set the objectives
of the evaluation. These objectives dictate which methodology or
decision strategy will be used in the evaluation (e.g. multi-attribute,
multi-objective, individual, participative, deterministic, probabilistic). In step two of the MCDM process, the appropriate criteria are
dened. 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 criteria are selected, the data for each criterion is collected and mapped
(step three), usually using a GIS. Because factors can be continuous 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 industrial 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 criteria range from 0 to 1 and should be specied so that their sum is 1.
The nal step of the MCDM procedure involves analysing the criteria 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 particular objective using Eq. (1).
S=

wi xi

cj

(1)

Where S is the objective value; wi is the weight of factor i; xi is

the criterion score of factor i; cj 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 to
spatially integrate and compare multiple geographically referenced 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 semistructured (Clarke, 1990), the MCDM procedures can be automated
within a GIS. The following sections explain how MCDM was implemented for modelling the growth potential of the settlements in the
study area.
2.3. MCDM implementation
2.3.1. Structuring framework
A structuring framework (Table 1) for growth potential modelling was designed as part of the rst step of the MCDM
procedure. The framework design was also based on a combination of international indicator guidelines (e.g. the United
Nations Indicators of Sustainable Development) and national governmental policy-driven initiatives (e.g. National Development
Plan 2030, National Spatial Development Perspective 2004). The
structuring framework consists of ve 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 similarity between the ve identied themes and those used in
the internationally recognised Environmental Sustainability Index:
Social/Cultural, Economic, Environmental, Political, and Institutional/Technological. Infrastructure was identied as 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 xed
production factors to a physical space to enhance its development
value and potential (Wong, 2002). Although recent sustainable
development indicator sets tend to move away from the traditional
four pillars (social, economic, environmental and institutional)
towards a more multi-dimensional view of sustainable development including cross-cutting themes such as poverty and natural

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

183

Table 1
Structuring framework for indicator selection.
#

Theme

Sub-themes

Modelling purpose

Economic

Preconditions for
Growth

Physical environment

Infrastructure

Human Capital

Institutional

Extent and diversity of retail and services sector

Tourism potential
Economic size and growth
Economic diversity
Market potential
Change in labour force
Property market
Availability and quality of water
Natural potential
Land availability and use
Transport and communication
Availability of municipal infrastructure
Poverty and inequality
Human resources
Population structure and growth
Quality of governance
Safety and security
Administrative and institutional function
Availability of community and public institutions

Innovation Potential

guidelines were used to draw up an initial set of candidate criteria shown in Table 2. All criteria were judged to be factors that will
either contribute to or detract from growth potential (i.e. none were
considered to be absolute constraints to further growth).

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 identication. In the context of MCDM, each thematic index 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 objective of the evaluation.
2.3.2. Criteria identication
The framework was used to identify 85 criteria relating to
growth potential. Data availability was a major factor in nding
suitable criteria, mainly because provincial-wide datasets at settlement 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 spectrum 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. Quantiable features that can be
monitored to establish performance trends and that are sensitive
enough to reect important changes in a settlements characteristics were favoured. The frequency and coverage of the selected
elements were also considered for timely identication of performance trends. The structuring framework and indicator selection

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 represented by a point (i.e. its centre) or a polygon (i.e. its urban edge).
In addition, because a town is inuenced by its surrounding hinterland activities, a town can also be dened 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 data
that was expected to inuence the growth potential of towns, it
was recognised that a combination of spatial entities (i.e. centre of
town, urban edge and Voronoi polygon) had to be used to represent
towns. For instance, to calculate a towns distance from major roads,
the town had to be represented by its centroid (i.e. point). Voronoi
polygons are preferred when the inuence of the surrounding hinterland, 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) and
Gong 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 inuence for each
settlement.
2.3.4. Normalization
Because criteria can be measured in different scales (i.e. nominal, 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

184

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

Table 2
Factors considered for each index.
Human capital index
Average per capita income 2011 (Rands) a , c , d
% of population receiving social grantsb
% of households living in informal housing 2011b
% change in economic empowerment 20012011a , c
Overcrowding 2011b
% Unemployment 2011 b , c , d
Matric pass rate 2012 (%) a , c
% 2065 years old with primary or no education 2011b
% 2065 year olds with at least grade 12 and highera , c
Ratio non-economically active population age 2011b , c , d
% Population growth rate 20012011a
% In-migrants past 10 years 2011a
Economic index
Tourism potential 2008 a , c
% Growth of economically active population 20012011 a , c , d
Growth in town extent 0508 (ha)a
Growth in town extent 0811 (ha)a
Growth in town extent 0511 (ha)a
Distance to PE, CT and 6 leader towns b , c
Total personal income 2011 (Rands million) a , c
% Growth in highly skilled labour 20012011 a , c
Number of property transactions 2010a
Value of property transactions 2010a , c
Property tax revenue 2010a , c
Number of formal retail outlets and service sector businesses 2010a , c , d
Number of formal retail outlets and service sector businesses 2010 per persona
Bio-physical index
Number of vacant residential stands 2010a
Number of vacant business stands 2010a
Number of 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 developments realised 2011 (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 cultivated 2012a , c , d
Growth in% area cultivated (20072012)a , c
Size and status of unexploited minerals 2010a , c
Groundwater quality (EC) 2011 (Ms/m)b
Biodiversityb
Infrastructure index
% Households with access to the internet 2011a , c , d
Distance to nearest scheduled airportb , c
Distance to nearest commercial harbourb , c
Distance to nearest small harbour and slipwaysb , c
Access to main and national roadsa , c , d
Access to railwaysa , c , d
% households with access to cellphone 2011a , c
% households with access to sanitation (ush) 2011a , c
% households with access to electricity (lighting) 2011a , c
% households with access to waste removal 2011a ,c
WWTW spare capacity per person 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
Qualied audits 2012a , c , d
Infrastructure backlog reduction 2010a , c , d
OPEX per capita 2010a
CAPEX per capita 2010a
Staff per cap ratio 2010b , c
% Posts lled 2010a , c , d
% Crime (all) occurrences change 20092012b , c , d
% Contact crime occurrences change 20092012b
% Property crime occurrences change 20092012b
Crime (all) occurrences (0912) per 100,000 populationb , c
Contact crime occurrences (0912) per 100,000 populationb
Property crime occurrences (0912) per 100,000 populationb
Small business support 2010a , c
Voter turnout 2010a
Number of Amenities 2010a , c , d
Number of Social service organisations 2010a , c
a
b
c
d

Factor contributed positively to the index (i.e. high values are preferred).
Factor contributed negatively to the index (i.e. low values are preferred).
Factor used in Criteria Subset A (see Section 2.3.5).
Factor used in Criteria Subset B (see Section 2.3.5).

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

185

Fig. 3. Weighted Voronoi polygons used as basis for data collection and analysis at settlement level.

method applied in MCDM as it uses the range of a variable as basis

for standardization. This is especially useful when different variables 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 and scenario-building purposes, linear scaling was
denoted as Normalization Method A, while z-score standardization
was referred to as Normalization Method B.
Xi =

Ri Rmin
m
(Rmax Rmin )

(2)

where: Xi is the standardised score; Ri is the raw score; Rmin represents the minimum score; Rmax is the maximum score; and M is
an arbitrary multiplier representing the upper standardised range
value.
zik =

(xik x k )
sdk

(3)

where, zik is the standardised score (also called z-score); xik is the
raw value of variable k for settlement i; x k is the mean value of
variable k for all settlements in the province; and sdk is the standard
deviation of variable k.
A mechanism whereby criteria can be inverted was also implemented as some criteria may have a positive or negative effect on
growth potential. For instance, a low crime rate is expected to have

a positive effect on growth potential, while a population with a low

level of education would be considered negative. Factors with positive and negative inuence on growth potential differentiated in
Table 2 with a + or sign. The GPPSS was designed to process such
cases by applying Eq. (4) in cases where a criterion has a negative
effect.
Yi = 1 Xi

(4)

where: Yi is the inverted score; and Xi represents the original score.

2.3.5. Criteria subset selection
A subset of the initial criteria set was selected through a participatory process involving structured meetings with municipal
representatives and provincial government ofcials, as well as
other stakeholders (e.g. urban and regional planners, business
owners). 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 reections of growth potential. These
included the indicators relating to the physical growth of settlements as it was argued that some settlements have densication
policies in place to limit physical urban expansion.1

1
A settlements rate of physical expansion is not necessarily on par with its economic or population growth rate. Stellenbosch, for example, experienced an annual
population growth rate of 8.5% from 2000 to 2010, while its physical expansion in
the same period averaged at 2.8% per annum (Musakwa and Van Niekerk, 2013).

186

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

The qualitative process of criteria selection was followed by a

quantitative data reduction process to produce a second subset of
factors. The danger of using a large number of factors in modelling is
that sets of factors that relate to a common theme will dominate the
outcome of a particular index. This may occur simply because more
data is available relating to a particular theme, resulting in multiple factors that essentially convey the same information. Another
potential danger of using large numbers of indicators in composite
indices is the inherent risk of trading off a poor result in one component against a strong performance or positive result in another
component (i.e. compensability). To guard against duplication
and compensability, statistical analyses can be used to remove any
aligned and inverted repetition within each index. Two approaches
were used in this study. The rst approach made use of a Pearsons
two-tailed bivariate correlation analysis (using IBM SPSS v22 software) to identify variables that are highly correlated. In cases where
two criteria had very high (>0.8) correlations, a qualitative judgement was made to decide which indicator to eliminate. A total of
26 variables were eliminated in this manner, resulting in Criteria
Subset A which consisted of 46 criteria.
The second approach used for removing duplication involved
the use of principal component analysis (PCA). Multivariate techniques such as PCA are often employed for removing data
duplication in large datasets involving multiple variables (Booysen,
2002; Grasso and Canova, 2008). The purpose of PCA is to determine
the number of latent variables underlying the data, to condense
the data, and to dene the content and meaning of the components that accounts for the variation in the data. PCA is based on the
premise that the variation observed in a variety of individual variables reects the patterns of a smaller number of some deeper or
more fundamental features (also referred to as the components).
The technique provides a reliable means of simplifying a variable
set and identifying, within each set, the common components of
association (Acton et al., 2009). In this study, PCA was applied as a
data reduction and criteria selection technique by identifying and
removing duplicated variables. Separate factor analyses were performed for each set of variables relating to a particular thematic
index (Table 1). The results of the PCA were studied to identify
the components that cumulatively represented 70% or more of the
underlying variation if a variable set. Highly correlated variables
within each principle component were identied by considering
the factor loadings within each component. Through this process,
a total of 55 variables were found to be redundant and were consequently removed from further analysis, resulting in Criteria Subset
B which consisted of 17 criteria. The PCA approach consequently
resulted in a much smaller set of core criteria compared to that of
the bivariate correlation method.
2.3.6. Weighting
The weight allocated to each criteria to reect its relative
importance can have a signicant effect on the MCDM outcome
(Malczewski, 2004). Deciding on which weights to allocate to each
criterion becomes more difcult as the number of criteria increases.
Methods such as the analytical hierarchy process (AHP) (Saaty and
Vargas, 1991) has been used to support this task (Subramanian
and Ramanathan, 2012), but the procedure is limited to a relatively small number of factors (Kousalya et al., 2012). In addition,
because human reasoning is used to specify the relative importance of a factor, the process of allocating weights can also lead
to inconsistencies. For instance, criterion A may be regarded as
more important than criterion B, while B might be considered more
important than criterion C. An inconsistency will occur if criterion C has been dened as being more important than criterion
A (Marinoni, 2004). To guard against such inconsistencies, Saaty
(1977) introduced a consistency ratio (CR) which can be calculated
from the principle eigenvector of the comparison matrix. A com-

parison matrix is considered inconsistent when its CR value is 0.1

or more. Correcting inconsistencies requires a process of adjusting
the relative importance of the individual factors, which is not only
time-consuming, but often subjective (Malczewski, 2004). However, the major problem with weighting schemes is the qualitative
and subjective way in which they are drawn up. This was of particular concern in this study because the results could be used
for informing investment strategies which might benet specic
regions or settlements in the study area. Any possibility of bias in
the allocation of weights had to be eliminated as it may reduce
the acceptance of the results. One strategy for avoiding bias during
importance ranking is to assume that all factors are equally important. This approach was used by Zietsman et al. (2006) for ranking
the growth potential of settlements using 83 indicators. Another
strategy is to make use of statistical analyses to objectively determine the relative importance of a particular factor (Booysen, 2002).
In the case of PCA, components can be weighted with the proportion of variance in the original set of variables explained by the rst
principal component of that particular component. The advantage
of this technique is that it produces a set of weights that explain
the largest variation in the original variables.
For the purpose of scenario generation, two alternative weighting schemes were implemented in the SDSS. The rst scheme
(Weighting Scheme A) allocates equal weights to all the factors,
while the second scheme (Weighting Scheme B) makes use of PCA
to automatically generate a weight for each criterion. This is done
by allocating a weight to the criterion with the highest loading on
each principle component in accordance to the proportion of the
overall variance explained by the particular component.
2.3.7. Classication
A settlements index value can be used to rank each town according to its relative performance. However, ranking towns from high
(1) to low (131) within each index makes comparison between settlements difcult. In addition, it introduces an articial ordering
of settlements due to the relatively small intervals between index
values. For instance, town A may be ranked several positions higher
than town B even though their index values differ by only a fraction (also compounded by the compensability problem referred
to above). Consequently, all index values were grouped into ve
classes (labelled very high, high, medium, low and very low) using
natural breaks (Classication Technique A) and quintiles (Classication Technique B). Natural breaks, also known as Jenks algorithm
(Jenks, 1967), uses statistical analyses to detect natural breaks in
the histograms of the raw index values, while quintile classication creates groups that are equal in size (number of cases). To
perform statistical analyses, the classes were transformed to ordinal numbers namely: 1 = Very Low, 2 = Low, 3 = Medium, 4 = High,
and 5 = Very High.
2.4. Scenario generation and statistical analyses
The GPPSS was developed using ArcView GIS (ESRI, 2002). SDSS
development within an existing GIS environment has been shown
to provide several advantages over developing stand-alone systems
(Basson, 2005; Bester, 2004; Mlisa, 2007; Van Niekerk, 1997; Varma
et al., 2000). The GPPSS consists of a spatial database, a knowledge base and an inference system (Van Niekerk, 2009). The spatial
database contains the various datasets relating to the criteria listed
in Table 2, while the knowledge base is a list of criteria and their
respective weights. During execution, the inference engine normalizes and combines (using WLC) all the criteria to produce a
result that can then classied. By automating this procedure, various growth potential scenarios can interactively be generated by:
(1) editing (updating) the raw datasets values; (2) disabling certain criteria (i.e. creating a subset of criteria); (3) setting a different

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

187

Table 3
Growth potential scenarios generated by setting different model settings.
Scenario #

Scenario code

Criteria subset

Weighting scheme

Normalisation method

Classication technique

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

AAAA
AABA
ABAA
ABBA
BAAA
BABA
BBAA
BBBA
AAAB
AABB
ABAB
ABBB
BAAB
BABB
BBAB
BBBB

A
A
A
A
B
B
B
B
A
A
A
A
B
B
B
B

A
A
B
B
A
A
B
B
A
A
B
B
A
A
B
B

A
B
A
B
A
B
A
B
A
B
A
B
A
B
A
B

A
A
A
A
A
A
A
A
B
B
B
B
B
B
B
B

normalization method; (4) modifying the relative weights of individual criteria; and (5) selecting a different classication method.
The system can also be set up to iteratively apply various permutations 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 classication techniques (Table 3).
The classications of all 16 scenarios were mapped and statistically compared. The median growth potential classication per
settlement (MEDIAN) was rst calculated to represent the overall
(consensus) classication 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 quantify 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 classication of
that settlement.
Pearsons two-tailed bivariate correlation analyses were carried
out in IBM SPSS v22 software and used to determine how closely
the growth potential classications of each scenario agree to the
consensus (median) classication. A correlation score (COR) was
recorded for each scenario.
The SD score, MEDIAN, and classication results of all 16 scenarios were mapped to enable spatial comparison. The results
were presented to a group of stakeholders, including government ofcials, local and district municipal representatives, town
and regional planners, social scientists, environmental managers,
economists and geographers. Stakeholders were requested to interpret the scenario results and to comment on how the classications
compare to their growth expectations for individual settlements.
The SD scores, COR values and the outcome of the qualitative assessment were considered in the selection of the most appropriate
solution (Voss and Post, 1988).

3. Results
3.1. Classications
The classications of the 16 growth potential scenarios are
shown in Table A1. Overall there is a good classication 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 classications were identical no matter which

parameter set (combination of criteria subset, weighting scheme,

normalization method and classication 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 classication 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 the settlements that were classied in Fig. 4a
as having a high or very high growth potential are clustered around
the City of Cape Town, most likely inuenced by their proximity to
the metropolis. This cluster includes the towns Malmesbury, Paarl
and Stellenbosch, which were all classied 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 Bettys Bay, Pringle Bay and Hermanus. A fourth
cluster of high potential municipalities and settlements are located
along the Garden Route, with Mosselbaai, George and Knysna being
classied as having a very high growth potential and Plettenberg
Bay as high. Most of the settlements in the interior, specically
the Karoo region, were classied as having a limited (i.e. very low
or low) growth potential, the only exception being Oudtshoorn
which received an overall classication of 3.5 (High). In terms
of SD scores (Fig. 4b) it seems that there was better agreement
(low SD scores) between the classications of the larger settlements, while smaller rural settlements generally recorded larger
(e.g. 0.6 or more) SD scores. Examples include Lutzville, Redelinghuys, 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 classication, with 12 (75%) having a correlation of
more than 0.9 (p < 0.001). Scenario 1 (AAAA) achieved the highest correlation (0.963) with the MEDIAN classication, 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 correlations of 0.7 or more are generally considered to be very strong, it can be concluded that all of
the scenarios produced classications that are in high agreement.
This demonstrates that the methodology is relatively insensitive to
different sets of criteria and weighting schemes.

188

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

Fig. 4. Geographic comparison of the (a) MEDIAN (consensus) and (a) SD (disagreement) scores per settlement.

3.3. Qualitative evaluation

These ndings were subjected to a rigorous public participation
process consisting of several brainstorming workshops. The results
were also made available to the ofcials 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
classications of a small number of towns were raised, but most
were dispelled when the data (criteria) used in the modelling were
scrutinized. The growth potential classications of the settlements
in Fig. 5 were perceived by many to be a true reection of input
data (see Section 4), which conrmed the quantitative (correlation)
analysis results.

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

189

Fig. 5. Classication 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 development and social welfare support. As demonstrated in this paper,
growth potential modelling is an ill-structured problem as it has
multiple possible solutions, solution paths, and criteria. MCDM provides a logical framework for analysing and aggregating the large
number of factors that affect growth potential. GIS are very effective for preparing, analysing and presenting the various datasets
and criteria, but are not exible enough for interactive or automated scenario generation. The GPPSS enabled the construction of a
series of growth potential modelling scenarios or the Western Cape
province that can help government ofcials, 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 potential as it can highlight
sensitivity to specic 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, the
GPPSS can potentially automatically generate hundreds of different
subsets. The user can also interactively select or deselect individual
factors to see what inuence such changes will have of the results
(i.e. carry out a data sensitivity analysis). Another benet of the use
of the GPPSS is that the growth potential modelling can be easily
updated by simply editing or replacing the datasets associated with
the individual factors.
When applied to the data of the Western Cape the GPPSS
produced meaningful growth potential classications. 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 classications are good reections of
the underlying data and that they are not signicantly affected by

methodological variations. This nding is of great value because it

increases the condence in the modelling process and creates the
perception that the growth potential results are good reections of
the underlying data.
The bivariate correlation analyses revealed that the combination of parameters used for Scenario 1 produced a classication that
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), and
employs natural breaks classication (Classication Technique A)
to generate the growth potential classications. The classication
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 classications 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 classied as having a low growth potential (2) by Scenarios 1,
2, 5, 6, 8, 9, 12 and 13, while in Scenarios 3, 4, 10, 11, 12 and 14 it was
classied as having a medium potential (3). In three cases (Scenarios 7, 15, 16) Suurbraak received a growth potential classication
of very low (1). The general agreement amongst stakeholders was
that this small isolated town has little prospect of dramatic development and that a low growth potential is the most appropriate
classication. Similarly, when the criteria of Rheenendal (SD = 0.97)
is inspected it is clear that the medium (3) classication of Scenario
1 is the best reection of its true growth potential. This settlement is located close to the rapidly growing towns of Knysna and
George and has experienced a 127% growth in highly skilled labour
between 2001 and 2011. It also has good access to high quality surface and ground water; is close to airports and harbours; and has a
relatively low and declining crime rate. These factors contributed
to its medium growth potential classication in the majority of the
scenarios (2-6, 9, 10, 13, 14, 16).
Similar evaluations were carried out for various other settlements. The conclusion from these assessments was that Scenario 1
consistently produced sensible classication results. This observa-

190

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

tion, in combination with the fact that this scenario had the highest
COR score, led to the selection of Scenario 1 as the most appropriate growth potential classication model for the non-metropolitan
settlements of the Western Cape.
The GPPSS is not without limitations. Although its development
within a desktop GIS environment enabled rapid development,
the required software license makes it only accessible to those
with the required license. The systems graphical interface is very
rudimentary as it was designed as a research tool and not for general distribution. Users will also require familiarity with ArcView
software. A logical extension of this research would be the redevelopment of the GPPSS as a web application that is accessible over
the Internet.

This paper combined MCDM, GIS and SDSS to generate 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
was 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 Potential SDSS (GPPSS). The various 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, normalisation technique and classication method. Although the GPPSS
was specically 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.

5. Conclusions
The dynamics and intricacies of the problems and challenges
relating to settlements in recession must be approached in a coordinated manner. Investment strategies should be put into place to
accelerate development in settlements with high growth potential,
whilst ensuring sustained social and institutional support to those
living in regions with relatively low growth potential. Development policies should direct specic 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 institutions need an appropriate methodology to inform policy decision
management.

Acknowledgements
We thank the Department of Environmental Affairs and Development Planning of the Western Cape Government for providing
nancial support for this study.
Appendix A

Table A1
Settlement growth potential classications for all 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
Arniston
Ashton
Aurora
Barrydale
Beaufort West
Bettys /Pringle Bay
Bitterfontein
Bonnievale
Botrivier
Bredasdorp
Buffelsbaai
Caledon
Calitzdorp
Ceres
Citrusdal
Clanwilliam
Darling
De Doorns
De Rust
Doringbaai
Dwarskersbos
Dysselsdorp
Ebenhaesar
Eendekuil
Elandsbaai
Elim
Franschhoek
Friemersheim
Gansbaai
Genadendal
George
Goedverwacht
Gouda
Gouritsmond
Graafwater
Grabouw
Greyton
Jongensfontein

3
2
3
2
3
2
5
1
3
3
3
3
4
1
4
2
2
3
2
1
1
3
1
2
3
3
2
3
3
4
3
5
2
3
3
2
4
3
3

3
2
3
2
3
2
5
1
2
3
3
3
4
1
4
2
2
3
2
1
1
3
1
2
3
2
2
3
3
4
3
5
2
3
3
2
4
3
3

3
2
3
2
3
3
5
1
2
3
3
4
4
1
3
2
2
3
2
1
1
3
1
2
2
2
2
4
3
4
3
5
2
2
3
2
4
3
4

3
2
3
2
3
3
5
1
2
3
3
4
4
1
4
2
2
3
2
1
1
3
1
2
2
2
2
4
3
4
3
5
2
2
3
2
4
3
4

3
2
3
2
3
2
5
1
3
3
3
3
4
1
4
2
2
3
2
1
1
3
1
2
3
3
2
3
3
4
3
5
2
3
3
2
4
3
3

3
2
3
2
3
2
5
1
2
3
3
3
4
1
4
2
2
3
2
1
1
3
1
2
3
2
2
3
3
4
3
5
2
3
3
2
4
3
3

2
2
3
3
2
2
5
1
2
4
3
3
5
1
3
3
2
4
2
1
1
3
1
2
3
3
1
3
1
3
3
5
3
2
3
2
4
4
2

2
2
3
3
2
2
4
1
2
4
3
4
4
1
4
3
2
3
2
1
1
3
1
2
3
3
2
3
2
3
3
5
3
2
3
2
4
4
2

3
2
3
2
3
2
5
1
3
4
3
4
5
1
4
2
2
4
2
1
1
3
1
2
3
3
1
4
3
5
4
5
2
3
4
2
5
4
4

3
2
3
2
3
2
5
1
3
3
3
4
4
1
5
2
2
4
2
1
1
3
1
2
3
3
2
4
3
5
4
5
2
3
4
2
5
4
4

3
2
3
1
3
3
5
1
2
4
3
4
4
1
4
2
2
4
2
1
1
3
1
2
2
2
2
5
3
5
4
5
2
2
4
2
5
4
4

3
2
3
1
3
3
5
1
2
4
3
4
4
1
4
2
2
4
2
1
1
3
1
2
2
2
2
4
3
5
4
5
2
3
4
1
5
4
4

3
2
3
2
3
2
5
1
3
4
3
4
5
1
4
2
2
4
2
1
1
3
1
2
3
3
1
4
3
5
4
5
2
3
4
2
5
4
4

3
2
3
2
3
2
5
1
3
3
3
4
4
1
5
2
2
4
2
1
1
3
1
2
3
3
2
4
3
5
4
5
2
3
4
2
5
4
4

2
1
3
4
2
2
5
1
3
5
3
4
5
1
4
3
2
5
2
1
1
4
1
2
4
3
1
3
1
3
3
5
3
2
3
3
5
5
2

2
2
3
3
2
2
5
1
2
5
4
4
5
1
4
3
2
4
2
1
1
4
1
2
4
3
1
4
1
3
3
5
3
2
3
2
5
4
2

3
2
3
2
3
2
5
1
2
3.5
3
4
4
1
4
2
2
4
2
1
1
3
1
2
3
3
2
4
3
4
3
5
2
3
3
2
4.5
4
2

0.25
0.00
0.75
0.50
0.50
0.25
0.00
0.66
0.71
0.25
0.56
0.56
0.00
0.50
0.50
0.00
0.71
0.00
0.00
0.00
0.35
0.00
0.00
0.61
0.61
0.56
0.71
0.90
0.79
0.61
0.00
0.50
0.66
0.61
0.35
0.50
0.66
0.66

0.50

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

191

Table A1 (Continued)
Settlement

AELN

AEZN

AULN

AUZN

CELN

CEZN

CULN

CUZN

AELQ

AEZQ

AULQ

AUZQ

CELQ

CEZQ

CULQ

CUZQ

MEDIAN

SD

Haarlem
Heidelberg
Herbertsdale
Hermanus/Onrus
Hopeeld
Kalbaskraal
Klapmuts
Klawer
Kleinmond
Kliprand
Knysna/Brenton
Koekenaap
Koringberg
Kurland
Ladismith
Laingsburg
Lamberts Bay
Langebaan
Leeu Gamka
Lutzville
Malmesbury
Matjiesfontein
McGregor
Merweville
Montagu
Moorreesburg
Mosselbaai
Murraysburg
Napier
Natures Valley
Nuwerus
Op-die-Berg
Oudtshoorn
Paarl
Paternoster
Pearly Beach
Piketberg
Plettenberg Bay
Pniel/Kylemore
Porterville
Prince Albert
Prince Alfred Ham.
Rawsonville
Redelinghuys
Rheenendal
Riebeek-Kast./Wes
Rietpoort
Riversdale
Riviersonderend
Robertson
Saldanha
Saron
Sedgeeld
Slangrivier
St Helena Bay
Stanford
Stellenbosch
Stilbaai
Strandfontein
Struisbaai
Suurbraak
Swellendam
Touwsrivier
Tulbagh
Uniondale
Vanrhynsdorp
Velddrift
Villiersdorp
Volmoed
Vredenburg
Vredendal
Wellington
Wilderness
Witsand
Wolseley
Worcester
Yzerfontein
Zoar
COR

2
3
2
5
3
4
3
2
4
1
5
2
3
2
2
2
2
4
1
2
5
1
3
1
3
4
5
1
3
3
1
2
3
5
3
2
3
4
4
4
1
3
3
2
3
4
1
3
3
3
3
3
4
2
3
3
5
4
1
2
2
3
1
2
2
2
3
3
1
5
2
3
4
2
2
4
3
1
0.963

2
3
2
4
3
4
3
2
4
1
5
2
3
2
2
2
2
4
1
2
4
1
3
1
3
4
5
1
3
3
1
2
3
5
3
2
3
4
4
4
1
3
3
2
3
4
1
3
2
3
4
3
4
2
3
3
5
4
1
3
2
3
1
2
2
2
3
3
1
4
2
4
4
2
2
4
4
1
0.962

2
3
2
5
3
4
3
2
4
1
5
2
3
3
2
2
2
4
1
2
4
1
3
1
3
4
5
1
3
3
1
2
3
5
3
2
3
4
5
4
1
3
3
2
3
4
1
3
3
3
4
3
4
2
3
3
5
4
1
3
3
3
1
2
2
2
3
4
1
4
2
4
4
3
2
4
4
1
0.954

2
3
2
5
3
4
3
2
4
1
5
2
3
3
2
2
2
4
1
2
4
1
3
1
3
4
5
1
3
3
1
2
3
5
3
2
3
4
4
4
1
3
3
2
3
4
1
3
2
3
4
3
4
2
3
3
5
4
1
3
3
3
1
2
2
2
3
3
1
4
2
4
4
3
2
4
4
1
0.949

2
3
2
5
3
4
3
2
4
1
5
2
3
2
2
2
2
4
1
2
4
1
3
1
3
4
5
1
3
3
1
2
3
5
3
2
3
4
4
4
1
3
3
2
3
4
1
3
3
3
3
3
4
2
3
3
5
4
1
2
2
3
1
2
2
2
3
3
1
4
2
3
4
2
2
4
3
1
0.959

2
3
2
4
3
4
3
2
4
1
5
2
3
2
2
2
2
4
1
2
4
1
3
1
3
4
5
1
3
3
1
2
3
5
3
2
3
4
4
4
1
3
3
2
3
4
1
3
2
3
4
3
4
2
3
3
5
4
1
3
2
3
1
2
2
2
3
3
1
4
2
4
4
2
2
4
4
1
0.962

1
2
2
5
2
4
3
2
3
1
5
2
4
2
2
2
3
4
1
3
5
2
2
1
3
4
5
1
3
3
1
2
3
5
3
2
4
4
4
4
1
3
3
3
1
5
1
3
3
3
3
2
4
2
2
2
5
3
1
1
1
3
1
2
2
3
4
3
1
4
3
3
4
3
2
4
4
1
0.865

2
2
2
4
3
4
3
2
3
1
5
2
4
2
2
2
3
4
1
3
5
2
3
1
3
4
5
1
3
4
1
3
3
5
3
2
4
4
4
4
2
3
3
3
2
4
1
3
3
3
3
3
4
2
2
3
5
3
2
2
2
3
2
3
2
3
3
3
2
4
3
4
4
3
3
4
4
1
0.869

1
3
2
5
3
5
3
1
5
1
5
1
4
2
1
2
2
5
1
1
5
1
3
1
4
5
5
1
3
3
1
1
4
5
4
2
3
5
5
5
1
4
4
2
3
5
1
4
3
4
4
3
5
2
3
3
5
5
1
2
2
4
1
2
2
2
4
4
1
5
2
4
5
2
2
5
4
1
0.953

1
3
2
5
4
5
3
1
5
1
5
1
4
2
2
2
2
5
1
1
5
1
3
1
4
4
5
1
3
3
1
2
4
5
4
1
4
5
5
5
1
4
3
1
3
5
1
4
2
4
4
3
5
2
3
3
5
5
1
3
3
4
1
2
2
2
4
4
1
5
2
4
5
2
2
5
5
1
0.947

2
3
2
5
3
5
4
1
5
1
5
1
4
3
2
2
2
5
1
1
5
1
3
1
4
4
5
1
3
4
1
1
4
5
3
2
3
5
5
5
1
3
4
1
4
5
1
4
3
4
4
3
5
2
3
3
5
5
1
3
3
4
1
2
2
2
3
4
1
5
1
4
5
3
2
5
4
1
0.939

2
3
2
5
3
5
3
1
5
1
5
1
4
3
2
2
2
5
1
1
5
1
3
1
4
4
5
1
3
4
1
2
4
5
3
2
3
5
5
4
1
4
3
1
4
5
1
4
2
4
5
3
5
2
3
3
5
5
1
3
3
4
1
2
2
2
4
4
1
5
2
5
5
3
2
5
5
1
0.942

1
3
2
5
3
5
3
1
5
1
5
1
4
2
1
2
2
5
1
1
5
1
3
1
4
5
5
1
3
3
1
1
4
5
4
2
3
5
5
5
1
4
4
2
3
5
1
4
3
4
4
3
5
2
3
3
5
5
1
2
2
4
1
2
2
2
4
4
1
5
2
4
5
2
2
5
4
1
0.953

1
3
2
5
4
5
3
1
5
1
5
1
4
2
2
2
2
5
1
1
5
1
3
1
4
4
5
1
3
3
1
2
4
5
4
1
4
5
5
5
1
4
3
1
3
5
1
4
2
4
4
3
5
2
3
3
5
5
1
3
3
4
1
2
2
2
4
4
1
5
2
4
5
2
2
5
5
1
0.947

1
2
1
5
3
5
4
3
4
1
5
2
5
2
2
2
3
5
1
3
5
2
3
1
4
5
5
1
3
4
1
2
4
5
3
2
5
4
5
5
1
3
4
3
1
5
1
3
4
4
4
2
5
2
2
2
5
3
1
1
1
4
1
2
2
3
4
4
1
5
4
4
4
4
2
5
4
1
0.847

1
2
1
5
3
4
4
2
4
1
5
2
5
2
2
2
3
5
1
3
5
2
3
1
4
5
5
1
3
5
1
2
4
5
3
2
5
5
5
5
1
4
3
3
1
5
1
3
3
4
4
3
5
1
2
3
5
3
1
1
1
4
1
3
2
3
4
4
2
5
4
4
5
4
2
5
4
1
0.858

3
2
5
3
3.5
4
3
2
4
1
5
2
4
2
2
2
2
4.5
1
2
5
1
3
1
3.5
4
5
1
3
3
1
2
3.5
5
3
2
3
4
5
4
1
3
3
2
3
5
1
3
3
3.5
4
3
4.5
2
3
3
5
4
1
2.5
2
3.5
1
2
2
2
3.5
4
1
5
2
4
4
2.5
2
4.5
4
1

0.50
0.35
0.43
0.43
0.87
0.66
0.43
0.66
0.71
0.00
0.00
0.61
0.71
0.50
0.35
0.00
0.50
0.50
0.00
0.79
0.56
0.50
0.25
0.00
0.50
0.50
0.00
0.00
0.00
0.71
0.00
0.50
0.50
0.00
0.50
0.35
0.87
0.66
0.66
0.66
0.25
0.61
0.50
0.71
0.97
0.66
0.00
0.61
0.66
0.50
0.56
0.35
0.50
0.25
0.50
0.35
0.00
0.79
0.25
0.79
0.75
0.50
0.25
0.35
0.00
0.50
0.50
0.66
0.35
0.66
0.83
0.50
0.66
0.71
0.25
0.50
0.56
0.00

192

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

Fig. A1. Classication results of Scenarios 216.

References
Acton, C., Miller, R., Fullerton, D., Maltby, J., 2009. SPSS for Social Scientists, 2nd ed.
Palgrave Macmillan, Hampshire.

Agrell, P.J., Stam, A., Fischer, G.W., 2004. Interactive multiobjective agro-ecological
land use planning: the Bungoma region in Kenya. Eur. J. Oper. Res. 158,
194217.
Al-Najjar, B., Alsyouf, I., 2003. Selecting the most efcient maintenance approach
using fuzzy multiple criteria decision making. Int. J. Econ. 84, 85100.

A. van Niekerk et al. / Land Use Policy 50 (2016) 179193

Arbia, G., Battisti, M., Di Vaio, G., 2010. Institutions and geography: empirical test
of spatial growth models for European regions. Econ. Modell. 27, 1221.
Ascani, A., Crescenzi, R., Iammarino, S., 2012. Regional Economic Development: A
Review. Department of Geography and Environment, London School of
Economics and Political Science.
Ascough, J.C.I., Rector, H.D., Hoag, D.L., McMaster, G.S., Vandenberg, B.C., Shaffer,
M.J., Weltz, M.A., Ahjua, L.R., 2001. Multicriteria spatial decision support
systems for agriculture: overview, applications, and future research directions.
In: Rizzoli, A.E., Jakeman, A.J. (Eds.), Integrated Assessment and Decision
Support Proceedings of the 1st Biennial Meeting of the IEMSS. June 2427
2002. IEMSS, Lugano, pp. 175180.
Barro, R.J., 1991. Economic growth in a cross section of countries. Quart. J. Econ.
106, 407.
Basson, F.C., 2005. A Spatial Decision Support System for Groundwater Abstraction
Impact Assessment and Licensing. Stellenbosch University, Stellenbosch.
Battisti, M., Vaio, G., 2008. A spatially ltered mixture of -convergence
regressions for EU regions, 19802002. Empirical Econ. 34, 105121.
Batty, M., 1979. Progress, success, and failure in urban modelling. Environ. Plan. A
11, 863878.
Bester, F.J., 2004. Multi-criteria Decision Making and Geographical Information
Systems: An Extension for ArcView. Stellenbosch University, Stellenbosch.
Booysen, F., 2002. An overview and evaluation of composite indices of
development. Soc. Indic. Res. 59, 115151.
Bowns, C., 2013. Shrinkage happens [hellip] in small towns too!: responding to
de-population and loss of place in Susquehanna River Towns. Urban Des. Int.
18, 6177.
Broennimann, O., Thuiller, W., Hughes, G., Midgley, G.F., Alkemade, J.M.R., Guisan,
A., 2006. Do geographic distribution, niche property and life form explain
plants vulnerability to global change? Global Change Biol. 12, 10791093.
Bruno, R., Follador, M., Paegelow, M., Renno, F., Villa, N., 2006. Integrating remote
sensing, GIS and prediction models to monitor the deforestation and erosion in
Peten Reserve. In: Guatemala, Society for Mathematical Geology XIth
International Congress, Universit de Lige Belgium.
Chang, K., 2006. Introduction to Geographic Information Systems, 3rd ed. McGraw
Hill, New York.
Clarke, M., 1990. Geographical information systems and model-based analysis:
towards effective decision support systems. In: Scholten, H.J., Stillwell, J.C.H.
(Eds.), Geographical Information Systems for Urban and Regional Planning.
Kluwer Academic Publishers, Amsterdam, pp. 165175.
Cloke, P., 2013. An Introduction to Rural Settlement Planning. Taylor & Francis.
CNDV Africa, 2005. Western Cape Provincial Spatial Development Framework.
PGWC, Cape Town.
DeMers, M.N., 2009. Fundamentals of Geographic Information Systems, 3rd ed.
Wiley & Sons, New York.
Densham, P.J., 1991. Spatial decision support systems. In: Maguire, D.J., Goodchild,
M.F., Rhind, D.W. (Eds.), Geographical Information Systems: Principles and
Applications. Longman Harlow, pp. 403412.
Donaldson, R., Spocter, M., Du Plessis, D., Van Niekerk, A., 2012a. Towards generic
interventions to stimulate growth potential in small towns of the Western
Cape Province, South Africa. S. Afr. Geog. J. 94, 120136.
Donaldson, R., Van Niekerk, A., Du Plessis, D., Spocter, M., 2012b. Non-metropolitan
growth potential of Western Cape Municipalities. Urban Forum 23, 367389.
Dong, P., 2008. Generating and updating multiplicatively weighted Voronoi
diagrams for point, line and polygon features in GIS. Comput. Geosci. 34,
411421.
Eastman, J.R., 2000. Decision Strategies in GIS. Directions Magazine, Directions
Media, p. s. p.
ESRI, 2002. ArcView GIS online user guide. ESRI, Redlands.
ESRI, 2011, ArcGIS 10 with Service Pack 4 for Windows [Ofine Help].
Gong, Y., Li, G., Tian, Y., Lin, Y., Liu, Y., 2012. A vector-based algorithm to generate
and update multiplicatively weighted Voronoi diagrams for points, polylines,
and polygons. Comput. Geosci. 42, 118125.
Goodchild, M.F., Densham, P.J., 1990. Research initiative six, spatial decision
support systems: scientic report for the specialist meeting. In: Technical
Report 90-5, National Center for Geographic Information and Analysis, Santa
Barbara.
Grasso, M., Canova, L., 2008. An assessment of the quality of life in the European
Union based on the social indicators approach. Social Indic. Res. 87, 125.
Henderson, J.V., Wang, H.G., 2007. Urbanization and city growth: the role of
institutions. Reg. Sci. Urban Econ. 37, 283313.
Hong, N.S., 1998. The Relationship Between Well-structured and Ill-structured
Problem-solving in Multimedia Simulation. The Pennsylvania State University.
Iyer, S., Kitson, M., Toh, B., 2005. Social capital, economic growth and regional
development. Reg. Stud. 39, 10151040.
Jacobs, W., Du Plessis, D., 2015. A spatial perspective of the patterns and
characteristics of main- and substream migration to the Western Cape, South
Africa. In: Unpublished research paper, Centre for Regional and Urban
Innovation in Statistical Exploration (CRUISE). Stellenbosch University,
Stellenbosch.
Jankowiski, P., Nyerges, T., 2001. GIS-supported collaborative decision making:
Results of an experiment. Ann. Assoc. Am. Geogr. 91, 4870.
Jenks, G.F., 1967. The data model concept in statistical mapping. Int. Yearbook
Cartogr. 7, 186190.
Kaza, N., 2010. Methods, information systems and mapping. J. Am. Plann. Assoc. 76,
123.

193

Kitchener, K.S., King, P.M., 1981. Reective judgment: concepts of justication and
their relationship to age and education. J. Appl. Dev. Psychol. 2, 89116.
Kok, P., Collinson, M., 2006. Migration and urbanization in South Africa. In: Report
03-04-02. Statistics South Africa, Pretoria.
Kousalya, P., Reddy, M., Supraja, S., Prasad, V.S., 2012. Analytical hierarchy process
approach-An application, of engineering education. Math. Aeterna 2, 861878.
Mahini, A.S., Gholamalifard, M., 2006. Siting MSW landlls with a weighted linear
combination methodology in a GIS environment. Int. J. Environ. Sci. Technol. 3,
435445.
Malczewski, J., 1999. GIS and Multi-criteria Decision Analysis. John Wiley & Sons,
New York.
Malczewski, J., 2004. GIS-based land-use suitability analysis: a critical overview.
Prog. Plann. 62, 365.
Malczewski, J., 2006. Ordered weighted averaging with fuzzy quantiers:
GIS-based multicriteria evaluation for land-use suitability analysis. Int. J. Appl.
Earth Obs. Geoinf. 8, 270277.
Marinoni, O., 2004. Implementation of the analytical hierarchy process with VBA in
ArcGIS. Comput. Geosci. 30, 637646.
Mlisa, A., 2007. Spatial Decision Support System for Hydrological Studies in the
Table Mountain Group Aquifers, Geography & Environmental Studies.
Stellenbosch University, Stellenbosch.
Musakwa, W., Van Niekerk, A., 2013. Implications of land use change for the
sustainability of urban areas: a case study of Stellenbosch, South Africa. Cities
32, 143156.
Pearson, R.K., 2002. Outliers in process modeling and identication. Control Syst.
Technol. IEEE Trans. 10, 5563.
Phua, M.-H., Minowa, M., 2005. A GIS-based multi-criteria decision making
approach to forest conservation planning at a landscape scale: A case study in
the Kinabalu Area, Sabah, Malaysia. Landscape Urban Plann. 71, 207222.
Pike, A., Rodriguez-Pose, A., Tomaney, J., 2010. Handbook of Local and Regional
Development. Taylor & Francis.
Rodrguez-Pose, A., 2013. Do institutions matter for regional development? Reg.
Stud. 47, 10341047.
Saaty, T.L., 1977. A scaling method for priorities in hierarchical structures. J. Math.
Psychol. 15, 234281.
Saaty, T.L., 1978. Modeling unstructured decision problemsthe theory of
analytical hierarchies. Math. Comput. Simul. 20, 147158.
Saaty, T.L., Vargas, L.G., 1991. Prediction, Projection and Forcasting. Kluwer
Academic Publishers, Boston.
Statistics South Africa, 2012. Census 2011 Census in Brief. Statistics South Africa,
Pretoria.
Subramanian, N., Ramanathan, R., 2012. A review of applications of analytic
hierarchy process in operations management. Int. J. Prod. Econ. 138, 215241.
Tanguay, G.A., Rajaonson, J., Lefebvre, J.-F., Lanoie, P., 2010. Measuring the
sustainability of cities: an analysis of the use of local indicators. Ecol. Indic. 10,
407418.
Van der Merwe, J.H., 1997. GIS-aided land evaluation and decision-making for
regulating urban expansion: a South African case study. GeoJournal 43,
135151.
Van der Merwe, J.H., Ferreira, S.L.A., Van Niekerk, A., 2008. A spatial gap-analysis of
tourism development opportunity in the Western Cape province. In: Centre for
Geographical Analysis. Stellenbosch University, Stellenbosch.
Van der Merwe, J.H., Ferreira, S.L.A., Van Niekerk, A., 2013. Resource-directed
spatial planning of agritourism with GIS. S. Afr. Geogr. J. 95, 1637.
Van der Merwe, J.H., Van Niekerk, A., 2013. Application of geospatial technology for
gap analysis in tourism planning for the Western Cape. S. Afr. J. Sci. 109, 110.
Van der Merwe, J.H., Von Holdt, D.S., 2006. Environmental footprint of aircraft noise
exposure at Cape Town International Airport. S. Afr. Geogr. J. 88, 177193.
Van Niekerk, A., 1997. Die Ontwikkeling Van Geograese Inligtingstelsels Vir
Omgewingsbestuur in Die Wes-kaap. Stellenbosch University, Stellenbosch.
Van Niekerk, A., 2008. CLUES: A Web-based Land Use Expert System for the
Western Cape. Stellenbosch University, Stellenbosch, pp. 221.
Van Niekerk, A., 2009. Using CLUES to nd suitable areas for perennial crops.
Varma, V.K., Ferguson, I., Wild, I., 2000. Decision support system for the sustainable
forest management. For. Ecol. Manag. 128, 4955.
Vonk, G., Geertman, S., Schot, P., 2007. A SWOT analysis of planning support
systems. Environ. Plann. A 39, 16991714.
Voss, J., Post, T., 1988. On the solving of ill-structured problems. InM. TH Chi, R.
Glaser, & MJ Farr (Eds.), The nature of expertise (pp. 311-342). Hillsdale, NJ:
Erlbaum. VossOn the solving of ill-structured problems311The nature of
expertise1988.
Vreeker, R., Nijkamp, P., Ter Welle, C., 2002. A multicriteria decision support
methodology for evaluating airport expansion plans. Transp. Res. 7, 2747.
Western Cape Government, 2013. Provincial economic review and outlook. Report
PR301/2013. Provincial Treasury, Cape Town.
Wong, C., 2002. Developing indicators to inform local economic development in
England. Urban Stud. 39, 18331863.
Wood, L.J., Dragicevic, S., 2007. GIS-based multicriteria evaluation and fuzzy sets to
identify priority sites for marine protection. Biodivers. Conserv. 16, 25392558.
Yalcin, G., Akyurek, Z., 2004. Analysing ood vulnerable areas with multicriteria
evaluation.
Zietsman, H.L., Ferreira, S.L.A., Van Der Merwe, I.J., 2006. Measuring the growth
potential of towns in the Western Cape, South Africa. Dev. S. Afr. 23, 685700.