GIS-Based Multiple-Criteria Decision Analysis
Randal Greene1
*, Rodolphe Devillers1
, Joan E. Luther2
and Brian G. Eddy2
1
Department of Geography, Memorial University of Newfoundland
2
Canadian Forest Service, Natural Resources Canada
Abstract
Important and complex spatial decisions, such as allocating land to development or conservationoriented
goals, require information and tools to aid in understanding the inherent tradeoffs. They
also require mechanisms for incorporating and documenting the value judgements of interest
groups and decision makers. Multiple-criteria decision analysis (MCDA) is a family of techniques
that aid decision makers in formally structuring multi-faceted decisions and evaluating the alternatives.
It has been used for about two decades with geographic information systems (GIS) to analyse
spatial problems. However, the variety and complexity of MCDA methods, with their varying
terminologies, means that this rich set of tools is not easily accessible to the untrained. This paper
provides background for GIS users, analysts and researchers to quickly get up to speed on MCDA,
supporting the ultimate goal of making it more accessible to decision makers. A number of factors
for describing MCDA problems and selecting methods are outlined then simplified into a decision
tree, which organises an introduction of key methods. Approaches range from mathematical programming
and heuristic algorithms for simultaneously optimising multiple goals, to more common
single-objective techniques based on weighted addition of criteria values, attainment of criteria
thresholds, or outranking of alternatives. There is substantial research that demonstrates ways to
couple GIS with multi-criteria methods, and to adapt MCDA for use in spatially continuous
problems. Increasing the accessibility of GIS-based MCDA provides new opportunities for
researchers and practitioners, including web-based participation and advanced visualisation of
decision processes.
1. Introduction
People often make spatial decisions, in both personal and professional matters: what route
to take on a daily commute, where to locate a new branch office, or which forest stands to
harvest. Selecting an alternative usually requires trading off different considerations. Route
selection, for instance, may be a trade-off among distance, driving time, road quality
and scenery. Different people facing the same problem may apply different values and
motivations and reach different conclusions. As decisions increase in complexity and
importance, so does the need to formalise them using available information, and to document
the rationale.
Multiple-criteria decision analysis (MCDA) can be defined as ‘a collection of formal
approaches which seek to take explicit account of [key factors] in helping individuals or
groups explore decisions that matter’ (Belton and Stewart 2002, 2). For approximately
20 years, MCDA methods have been used for spatial problems by coupling them with
geographic information systems (GIS) (Carver 1991; Malczewski 2006a). The goal of this
paper is to make the GIS-based MCDA field more accessible to a wider audience. This
includes the GISciences community of researchers, analysts, and users, and ultimately
experts and decision makers in many fields. The GIS literature is filled with tools, scenarGeography
Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
ª 2011 The Authors and Crown in the right of Canada
Geography Compass ª 2011 Blackwell Publishing Ltd
ios and cases involving spatial decision support (Dragic´evic´ 2008; Nyerges and Jankowski
2010), so there is a major challenge for newcomers in even identifying GIS-based MCDA
research and tools. It requires an understanding of the concepts of non-spatial MCDA,
hence a related goal is to make sense of the sheer variety of MCDA methods and the
many ways they can be integrated with GIS.
Section 2 introduces MCDA, and Section 3 lists a number of factors used to categorise
decision scenarios and select formal methods. These selection factors are used
to build a methods decision tree in Section 4, which organises brief descriptions of
key MCDA methods. We then discuss the spatial extension of MCDA in Section 5,
particularly spatially continuous problems that are ideally suited to modelling with
GIS. Research trends in the field of GIS-based MCDA are also reviewed. Section 6
is geared towards the practitioner, covering available software and coupling strategies
for integrating MCDA with GIS. The Conclusion identifies opportunities related to
making the field more accessible.
2. MCDA Background
Multiple-criteria decision analysis aids decision makers in analysing potential actions or
alternatives based on multiple incommensurable factors⁄criteria, using decision rules to
aggregate those criteria to rate or rank the alternatives (Eastman 2009; Figueira et al.
2005; Malczewski 1999a). Although the decision criteria normally cannot all be maximised
in selecting an alternative or action, MCDA researchers and practitioners do not
view it simply as a quantitative optimisation problem that identifies the best potential
‘solutions’. Instead, the focus is on eliciting and making transparent the values and subjectivity
that are applied to the more objective measurements, and understanding their
implications (Belton and Stewart 2002; Roy 2005). The field is often referred to as multiple
criteria decision making, but decision ‘analysis’ or ‘aiding’ (MCDA) better reflects the
more subtle and broader-ranging intentions.
Multiple-criteria decision analysis grew out of and in reaction to single-criterion optimisation
techniques, most notably linear programming. These were developed during
World War II and honed in the early days of the business management field of Operations
Research, in both contexts without considering secondary consequences that require
multiple criteria (Zeleny 1982). Simple and somewhat crude approaches to reconciling
multiple criteria require alternatives to meet one, some or all criteria based on cut-off
values. These approaches are named non-compensatory methods, in that increases in the
value of one criterion cannot be offset by decreases in the value of another (Hwang and
Yoon 1981).
Among advocates of more sophisticated compensatory approaches that facilitate criteria
tradeoffs, two prominent schools of MCDA (American and European, summarised in
Table 1) evolved simultaneously but somewhat separately during the 1960s and 1970s.
Both schools shared the concepts of decision alternatives and criteria, but differed in their
philosophy and approach to aggregating criteria. The early American school of MCDA
followed the Operations Research tradition. One set of its methods used a value or utility
function based on multi-attribute utility theory (Keeney and Raiffa 1976), multiplying
weights by normalised criteria values (for instance converted to a continuous 0–1 scale)
and summing these to derive a score or rating for each alternative. Another set of methods
within the American school centred on the idea of specifying desirable or satisfactory
outcomes and using mathematical programming to come as close as possible to these in
criteria outcome space (multi-dimensional space where each dimension represents the
GIS-based MCDA 413
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
possible values of one criterion) (Dykstra 1984). The word ‘programming’ is used in the
sense of the program of action that is recommended as a result of the analysis. The European
school moved away from the Operations Research idea of obtaining an optimum,
and developed outranking relationships to help decision makers compare alternatives in a
pair-wise manner to rank their preferences for the alternatives in various ways (Roy
1968a cited in Roy and Vanderpooten 1996; Vincke 1992). A key assertion in this
approach is that decision makers do not have precise preconceptions of the relative
importance of the criteria, and that decision aiding should help them develop this insight.
A somewhat less prominent school of MCDA, based on fuzzy sets (Zadeh 1965) and
value-function aggregation, is the analytic hierarchy process (AHP) developed by Saaty
(1980). AHP uses pair-wise comparison of criteria to derive relative weights.
As MCDA has grown, the clear divisions among the schools have diminished. For
instance, subtleties introduced by the European school, such as recognition of subjectivity
and imperfect knowledge (Roy and Vanderpooten 1996), are now widely recognised and
are reflected in the accepted definitions of MCDA. The various techniques are considered
tools in the analyst’s toolkit to be applied as appropriate to different problems or phases
of the same problem. Consequently, the primary research challenges moved from
development of methods, to such issues as frameworks for method integration (Belton
and Stewart 2002) and application in distributed collaborative environments (Carver
1999; Malczewski 1999a). There has been a steady growth in MCDA’s range of application,
for instance in environmental and resource fields such as forest management (DiazBalteiro
and Romero 2008; Mendoza and Martins 2006).
Perhaps MCDA’s greatest strength is its ability to simultaneously consider both quantitative
and qualitative criteria, as long as the latter can be represented using an ordinal or
continuous scale. One result is that MCDA is an alternative to decision analysis based
solely on economic (monetary) valuation. There is substantial literature on economic valuation
of non-monetary phenomenon, such as ecosystem goods and services (van Kooten
and Bulte 2000; Turner et al. 2008). A practical challenge of such approaches is avoiding
dismissal by decision makers of these often very large and theoretical valuations when pitted
against hard economic criteria like jobs and exports. MCDA approaches can help
overcome economic biases (Herath and Prato 2006) by either using a non-monetary
common denominator (a continuous scale like 0–1) or avoiding altogether the need to
convert criteria from their original values.
Table 1. Early schools of multiple-criteria decision analysis.
American school European (French) school
Assumptions Precise knowledge and judgements,
optimal decisions
Imprecision in evaluating criteria, optimal
decisions not achievable
Goal Rating and selection of alternatives Ranking of alternatives
Aggregation
approaches
Value ⁄ utility function, multi-criteria
and multi-objective optimisation
Outranking
Key institutions Decision Sciences Institute – http://
www.decisionsciences.org/
Institute for Operations Research
and the Management Sciences –
http://www.informs.org
LAMSADE – http://www.lamsade.dau
phine.fr/
EURO Working Group – Multicriteria
Decision Aiding – http://www.cs.put.
poznan.pl/ewgmcda/
414 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
3. Method Selection Factors
One approach to succinctly categorising virtually all MCDA scenarios is their association
with various problem types, or proble´matiques. These include choice (making a single
selection or recommendation), ranking (establishing a preference order for some or all of
the alternatives), sorting (separating alternatives in classes or groups), description (learning
about the problem), design (developing new alternatives for possibly addressing the problem)
and portfolio (selecting a subset of alternatives) (Belton and Stewart 2002; Roy
1996).
Other factors that describe decision problems or affect the choice and implementation
of MCDA methods include:
1 Number of decision makers: MCDA techniques designed for individuals can be applied
for group decisions where consensus can be achieved through education or negotiation
(Malczewski 2006a). Otherwise, the methods must be extended using approaches
such as aggregated weighting (Malczewski 1999b) or voting (Hwang and Lin 1987).
Group approaches open up a variety of issues, often studied in Collaborative GIS
research (Balram and Dragic´evic´ 2006; Joerin et al. 2009; Rinner 2001).
2 Decision phase: The phase or phases of the decision process to be supported. There are
many ways to organise and describe decision phases (Anderson et al. 2003; Bouyssou
et al. 2006; Turban and Aronson 2001), with a critical distinction for MCDA
between the problem exploration⁄structuring phase and the evaluation⁄recommendation
phase.
3 Number of objectives: With a single objective (such as recommending the site for a new
fire station), the decision maker(s) can focus on relevant criteria or factors with measurable
attributes, and thus corresponding techniques are often called multiple-criteria
evaluation (MCE) or multiple-attribute decision making (MADM) (Jankowski 1995;
Malczewski 1999a). With multiple-objective decision making (MODM), it is necessary
to establish whether the objectives are in synergy or conflict (for instance allocating
urban land either to housing or green space) and to group the criteria by
objective (Eastman et al. 1995; Malczewski 2004).
4 Number of alternatives: Scenarios with a limited number of clear alternatives (like analysing
three pre-selected locations for a new fire station) are discrete problems that
usually culminate in a single selection (Chakhar and Mousseau 2007). A large or infinite
number of alternatives (like identifying all possible sites for the new fire station)
signifies a continuous problem usually characterised as screening, search, or suitability
rating (Eastman 2009; Malczewski 1999a).
5 Existence of constraints: Limitations on solutions, either in the form of alternatives⁄areas
to be excluded from consideration or conditions that the recommended solution must
meet. Common constraints in spatially continuous problems are that recommended
areas must be a minimum contiguous size (Eastman 2009) or provide corridors of
connectivity (Chakhar and Mousseau 2008).
6 Risk tolerance: The decision makers’ level of risk tolerance (Eastman 2009) and desire
to quantify the risk inherent in a choice (Chen et al. 2001; Eastman 2005). For
instance, when screening alternatives, a risk-tolerant decision maker might be willing
to accept alternatives that meet just a few criteria or even one criterion. A risk-averse
decision maker, on the other hand, may accept only alternatives that meet all criteria.
7 Uncertainty: Whether the criteria and weighting should be modelled with certainty
(i.e. deterministically) or uncertainty (i.e. probabilistically or fuzzily) (Jiang and EastGIS-based
MCDA 415
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
man 2000; Malczewski 1999a; Shepard 2005). This may be based on the nature of
the criteria or simply a matter of modelling preference. For instance, in a land-classification
problem, the transition from woodland to wetland could be modelled with
crisp boundaries (either one or the other) or fuzzy boundaries (with one or more classification
levels where the land is partially wooded and partially wet).
8 Measurement scales and units: Whether it is possible to convert heterogeneous criteria
based on various measurement scales (such as currency and qualitative survey results)
to a common scale, and whether decision makers are comfortable with representing
criteria numerically (Chakhar and Mousseau 2008; Joerin et al. 2001).
9 Experience: The training and experience of the analyst and decision makers (Belton
and Stewart 2002). Given the large number of methods and their vastly different
assumptions (see discussion of the early schools of MCDA in the Introduction), this is
a very practical consideration that results in technique biases.
10 Computational resource capacity: Another practical consideration is available software
(Malczewski 1999a; Weistroffer et al. 2005) and hardware, and these can have budget
implications.
11 Direction of problem solving: Typically, problems are worked forward in support of a
new decision. However, existing decisions can be worked backward to elucidate the
value judgements that would be needed to support them, in a process called preference
disaggregation (Jacquet-Lagre`ze and Siskos 2001; Siskos 2005).
4. MCDA Methods
Given the diversity of MCDA methods, selection of an appropriate method or combination
of methods depends on the context. The decision tree of Figure 1 is, therefore, not
intended to be comprehensive or definitive, but provides one approach to simplifying the
selection process. The clearest separation of methods is based on whether or not there are
multiple objectives. If the decision maker or analyst determines that the multiple objectives
are either complementary or can be prioritised, then MADM methods can be
applied repeatedly in a two-level or stepwise fashion (Eastman 2009; Malczewski 1999a).
If the multiple objectives are in conflict, MODM methods are required. The choice is
based on the number of alternatives, between mathematical programming for locating an
optimal solution, and heuristic methods for locating a satisfactory solution close to the
optimum. Unfortunately, there is no easy definition of what constitutes a ‘large’ number
of alternatives as it depends on the computational capacity of the software or algorithm
being used.
The MADM side of the tree is divided based on the question of trading off criteria.
Non-compensatory approaches are easier to understand and apply, but they require
including or excluding alternatives based on hard cut-offs. Compensatory approaches are
more realistic and subtle in their modelling, as they allow criteria outcomes to be traded
off against each other on a continuous scale, so that a loss in one criterion can be compensated
for by a gain in another. Note that MODM methods are generally compensatory
by nature, and therefore always support criteria tradeoffs. Like MODM, selection of
compensatory MADM methods is also differentiated based on the number of alternatives.
It is important to realise that the methods are not mutually exclusive, due to the complexities
and multiple phases of decision analysis. For instance, non-compensatory techniques
could be used for preliminary screening of alternatives, followed by a
compensatory method to support final selection. Multiple techniques can also be applied
416 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
in parallel as part of a strategy to validate the robustness of the recommendations (Carver
1991; Roy 2005). A more common approach to sensitivity analysis is to run multiple
iterations using the same method, each time making slight adjustments in the inputs (such
as the selection and weighting of criteria) to assess the sensitivity of the resulting outputs
(Feick and Hall 2004; Malczewski 1999b; Store and Kangas 2001).
4.1. NON-COMPENSATORY AGGREGATION METHODS
Often used for screening as well as selection, non-compensatory methods include:
1 Conjunctive: Accept alternatives if they meet a cut-off value on every criterion. Implementations
involving spatial problems often use binary overlay (Jankowski 1995;
McHarg 1969), where the objects or cells in each layer are set to 1 if they pass the
Fig. 1. Multiple-criteria decision analysis methods decision tree. Shaded action nodes (dark grey) indicate the numbered
subsection of the paper that describes the set of methods.
GIS-based MCDA 417
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
cut-off for that criterion and 0 otherwise. The layers are combined using an intersection
operation (logical AND) to identify ‘solution areas’ that meet criteria, as shown in
Figure 2. Conjunctive methods are risk averse because all criteria must be fully met
(Eastman 2009).
Geology criterion (granitoid lithography)
And
And
And
Equals
“Solu on” areas
Waterdistance criterion (≥ 200m from wate
Slope criterion (≤ 10° slope)
Roaddistance criterion (≤ 1km from road)
Fig. 2. Conjunctive example. Binary overlay for mineral exploration site identification, showing areas that meet the
selected cut-off on all criteria.
418 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
2 Disjunctive: Accept alternatives that meet a cut-off value on at least one criterion
(Hwang and Yoon 1981). It can also be implemented for spatial problems using binary
overlay, where the map criteria layers are combined using a union (logical OR) operation.
It is a risk-taking method, because only one criterion must be met (Eastman
2009).
3 Lexicographic: Rank⁄order the criteria, then eliminate alternatives hierarchically by comparing
them on the highest ranked criterion, followed by the second highest ranked,
etc. (Carver 1991; Jankowski 1995).
4 Elimination by aspects: Use a lexicographic approach, but also enforce a conjunctive cutoff
for each criterion (Malczewski 1999a).
5 Dominance: Look for dominant alternatives that score at least as high as every other
alternative on every criterion (Jankowski 1995).
4.2. WEIGHTING METHODS
The following methods are used to derive relative criteria weights⁄importance before
applying a compensatory aggregation method (Belton and Stewart 2002; Malczewski
1999a; Nyerges and Jankowski 2010):
1 Ranking: Ranks⁄orders the criteria, then converts the ranks to weights using:
(i) Rank sum – each rank value divided by the sum of all rank values.
(ii) Rank reciprocal – 1 divided by each rank value.
(iii) Rank exponent – a rank sum with the numerator and denominator raised to a
power between 0 and 1, thereby reducing the resulting weight differences.
2 Rating: Rates the criteria using a common scale (such as any value between 0 and 1) or
point allocation (for instance allocating 100 points among all criteria).
3 Trade-off analysis: Directly assesses tradeoffs between pairs of criteria to determine the
cut-off values at which they are considered equally important.
4 Analytic hierarchy process: Compares criteria pair-wise on a fuzzy-linguistic ratio scale and
subsequently computes overall relative weights based on aggregate calculations of all
pair-wise ratios (Eastman 2009; Saaty 2005; Schmoldt et al. 2001). AHP is more than a
criteria weighting method, as it also provides an additive, hierarchical aggregation of
criteria. Figure 3 shows AHP weighting of three of the criteria from the mineral exploration
example.
4.3. COMPENSATORY AGGREGATION METHODS
Compensatory decision rules not requiring pair-wise comparison of alternatives are of
two types:
1 Additive methods that normalise criterion scores to enable comparison of performance
on a common scale:
(i) Weighted linear combination (WLC): Also known as simple additive weighting, this
approach multiplies normalised criteria scores by relative criteria weights for each
alternative (Geldermann and Rentz 2007; Nyerges and Jankowski 2010). WLC
can sum all weighted criteria values in a single step, or proceed hierarchically so
that each group of related criteria (such as wildlife, tourism and agriculture in a
rural land-management problem) is first aggregated before being combined with
other groups. In Figure 4, the earlier mineral exploration example is analysed
using single-step WLC, showing criteria normalisation and weighting, and the
GIS-based MCDA 419
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
resulting map of aggregated suitability scores. Because it supports full trade-off or
compensation among criteria values, WLC is mid-way on the risk tolerance continuum
between conjunctive and disjunctive approaches and is thus considered a
risk-neutral technique (Eastman 2009).
(ii) Fuzzy additive weighting: Adapts WLC using non-crisp criteria and weight values
derived from fuzzy-linguistic quantifiers such as ‘high’, ‘medium’ and ‘low’ (Gemitzi
et al. 2007; Malczewski 1999a; Zadeh 1965).
(iii) Ordered weighted averaging (OWA): Also based on fuzzy methods, OWA extends
WLC using criteria-order weights to control the levels of criteria trade-off, allowing
decision makers to place themselves along a continuous spectrum of risk tolerance
(Jiang and Eastman 2000; Malczewski 2006b; Rinner and Malczewski 2002;
Yager 1988).
2 Non-additive methods that use the original criteria scores:
(i) Ideal point: Identifies a point in criteria outcome space by specifying the preferred
value of each criterion (Malczewski 2004; Nyerges and Jankowski 2010). This
ideal point may not be close to a feasible alternative, but there are a number of
methods for selecting one, such as the Technique for Order Preference by Similarity
to Ideal Solution (TOPSIS) (Chen et al. 2001; Liu et al. 2006).
(ii) Non-dominated set: Identifies the set of alternatives that score at least as high as
every other alternative on at least one criterion, also called the efficient set or
Pareto set (Lotov et al. 2004; Malczewski 1999a).
(iii) Reasonable goals method: Extends the non-dominated set to help visually select from
the alternatives using a series of two-dimensional graphs of criteria outcome space
(Jankowski et al. 1999).
Fig. 3. Weight derivation using analytic hierarchy process in IDRISI GIS (http://www.clarklabs.org/). First, the criteria
are compared pair-wise. For instance, WaterDistance is considered to be strongly less important than RoadDistance
(1 ⁄ 5). Then the eigenvector of the pair-wise comparisons is used to determine the overall criteria weights. The consistency
ratio ensures, in this example, that the comparisons Slope ⁄ RoadDistance (1 ⁄ 3) and WaterDistance ⁄ RoadDistance
(1 ⁄ 5) are sufficiently consistent with the comparison WaterDistance ⁄ Slope (1 ⁄ 1).
420 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
4.4. OUTRANKING AGGREGATION METHODS
Outranking methods undertake pair-wise comparison of a discrete set of alternatives to
rank them based on concordance (the set of criteria for which one alternative dominates
another) and discordance (the opposite set) (Belton and Stewart 2002). The outranking
philosophy recognises that decision makers are subject to ambiguous and evolving value
judgements, even during the MCDA process. Well-known methods of this type
include:
1 ELECTRE: A family of outranking methods (ELECTRE I, II, III, IV and TRI) that
have evolved along with the European school of MCDA (Bouyssou et al. 2006; Joerin
et al. 2001). ELECTRE can handle various problem types (choice, ranking, sorting)
and approaches to decision modelling. It introduced thresholds for declaring indifference
or preference between two alternatives on a particular criterion, and support for
criteria that cannot be weighted (Belton and Stewart 2002).
2 PROMETHEE: An outranking method that supports various criterion preference functions
such as U-shaped, linear and flat (no threshold) (Brans and Mareschal 2005; Geldermann
and Rentz 2007; Marinoni 2006).
Fig. 4. Weighted linear combination example. Mineral exploration site identification based on the inputs from
Fig. 2, leaving the Geology criterion as a hard constraint, but using continuous values for the RoadDistance, Slope
and WaterDistance criteria. Continuous values are normalised to a 0–1 scale, with optional scale reversal for criteria
where less is better. Then they are weighted (in this case equally) and summed to produce the continuous output
shown. Darker areas are more suitable, with the highest rated area scoring 0.86 (of a possible maximum of 1).
GIS-based MCDA 421
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
4.5. MATHEMATICAL PROGRAMMING METHODS
The following methods attempt to find the optimal way to satisfy goals by solving systems
of equations:
1 Linear⁄integer programming: Mathematically optimises by maximising or minimising a single-criterion
value using constraints, commonly employed in Operations Research and
Management Science (Anderson et al. 2003; Wisniewski 2002). An example is to minimise
the driving time to visit a specific set of customers, subject to speed limit constraints.
To apply this approach, multi-objective problems are converted to a single
objective using value functions (in the case of deterministic models) or utility functions
(in the case of probabilistic models) (Malczewski 1999a).
2 Goal⁄compromise programming: Finds the alternative that minimises overall deviation or
distance from user-specified ideal points or aspiration⁄reservation levels simultaneously
for multiple objectives (Anderson et al. 2003; Baja et al.2007; Ghosh 2008).
3 Interactive programming (reference point): Uses successively refined aspiration⁄reservation
levels for each objective to select a feasible alternative (Janssen et al. 2008; Malczewski
1999a; Zeng et al. 2007).
4.6. HEURISTIC METHODS
Due to computational limitations, mathematical optimisation is not possible when there
are a large number of alternatives, such as in spatially continuous problems modelled
using raster layers, where every possible outcome of every raster cell is an alternative.
The following methods can be used to allocate cells among conflicting objectives, with
the aim of a close to optimal ‘solution’:
1 Multiple-objective land allocation (MOLA): Allocates each cell to the objective with the
closest ideal point. Objectives can optionally be weighted unequally, so that a cell may
be allocated to an objective with a higher weight even when there is an objective with
a closer ideal point (Eastman 2009; Eastman et al. 1995).
2 Genetic algorithms (GA): Allocates cells based on a trial-and-error process that introduces
small changes (evolutionary mutations) and tests for solution improvement (Aerts et al.
2005; Bone and Dragic´evic´ 2009; Malczewski 2004).
3 Simulated annealing (SA): Allocates cells based on an iterative random process that tests
for overall improvement at each step (Possingham et al. 2000; Duh and Brown 2007;
http://www.uq.edu.au/marxan/).
Genetic algorithms, SA and other techniques such as cellular automata (CA) (Malczewski
2004; Myint and Wang 2006; White et al. 2004) are collectively referred to as geocomputation
when used in spatial problems. They can be applied to related aspects of spatial decision
support, such as time series used to predict the future outcome of proposed
alternatives resulting from MCDA.
5. GIS-Based MCDA
The basic intention underlying spatialised applications of MCDA is to augment the
traditional question of ‘what’ with the additional question of ‘where’ (Malczewski
1999a). GIS-based MCDA also facilitates calculation and analysis of spatial criteria such
as distance, travel time and slope. Virtually all MCDA methods can be applied to spa-
422 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
tial problems, as shown by the examples and the many GIS-oriented references in the
methods just elaborated. As discussed earlier, many MCDA methods can only be
applied to a small number of alternatives due to computational limitations (in the case
of mathematical optimisation) or practical considerations (in the case of pair-wise
comparisons). This limits the choice of methods in spatially continuous problems,
which attempt to rate or allocate swaths of land (i.e. where every cell or parcel of
land is potentially part of the recommended solution). One approach to opening up
additional methods for these problems is to convert them to a smaller number of discrete
alternatives. For instance, strategic regional planning exercises (e.g. http://
www.geog.leeds.ac.uk/papers/99-8/; http://www.cbhvregionalplan.ca/) can employ
representative scenarios showing a few possible land configurations for debate and discussion.
A risk of this approach, though, is potentially biasing subsequent analyses by
excluding good alternative configurations (Belton and Stewart 2002). Another option
for spatially continuous study areas is classification into homogeneous zones based on
criteria values or categories (Chakhar and Mousseau 2008; van Herwijnen and Rietveld
1999; Joerin et al. 2001). This limits the number of alternatives to the combination
of possible outcomes for the zones, although often with a loss of spatial
resolution.
An important element of accessibility for any field is a vibrant research community.
Use of MCDA with and in GIS has been an active and growing topic of research since
the early 1990s (Malczewski 2006a,c). These literature reviews also reveal use of many
different combinations of methods and approaches. Leading application areas include
environment⁄ecology, transportation, urban⁄regional planning, waste management,
hydrology⁄water resources, agriculture and forestry. The reader is encouraged to refer to
Malczewski (2006a; http://publish.uwo.ca/~jmalczew/gis-mcda.htm) for case studies in
their areas of interest.
Despite the breadth of methods and applications, GIS-based MCDA can still be categorised
as a niche field. A field-specific research group and related journal (http://publish.
uwo.ca/~jmalczew/gimda/) did not survive. Non-GIS publications such as Journal of
Multi-Criteria Decision Analysis, Operations Research, Decision Sciences and Management Science
are important sources of information, but rarely publish GIS-oriented material. GIS-based
MCDA publishing typically occurs in the general GISciences literature or in applicationoriented
journals. These trends were confirmed with a search of the Scopus citation database
using the query [‘GIS’ AND (‘multiple criteria decision’ OR ‘multi-criteria decision’
OR ‘multicriteria decision’ OR ‘MCD*’ OR ‘multiple criteria evaluation’ OR ‘multicriteria
evaluation’ OR ‘multicriteria evaluation’ OR ‘MCE’)] resulted in 279 articles,
broken down by year in Figure 5. Other combinations of search terms could yield additional
relevant articles, but these results are representative of the steady progression of the
publications in the field.
Figure 6 lists the journals containing three or more of the 279 articles. They are overwhelmingly
in the GIS, Environmental and Planning fields, with the leader being the
International Journal of Geographical Information Science. One of the few noteworthy academic
conferences for the field is the Urban and Regional Information Systems Association
(http://www.urisa.org/) annual conference. Again, researchers have to look to
general GISciences, general decision research, application-specific fields or industry events
for dissemination. No academic institution is a clear leader in GIS-based MCDA,
although a selection of leading researchers based on their apparent prominence, stated
research interests and publications in the field is provided in Table 2.
GIS-based MCDA 423
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
6. GIS-Based MCDA Software
An important factor in the accessibility of research and methods is the availability of tools
that implement them. GIS-based MCDA software can be categorised based on the level
of integration of MCDA capabilities within GIS. Jankowski (1995, 2006) defines three
levels of GIS-MCDA coupling: full (a single software package provided by the vendor),
tight (a common user interface and data management, achieved through package customisation)
and loose (based on data exchange between packages). Most MADM techniques
can be implemented in most GIS packages without custom programming (Malczewski
Fig. 5. Geographic information system-based multiple-criteria decision analysis article count by year (from http://
www.scopus.com).
Fig. 6. Geographic information system-based multiple-criteria decision analysis article count by journal (from http://
www.scopus.com).
424 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
1999a). For instance, ESRI’s ArcGIS suite of products (http://www.esri.com) provides
the building blocks needed to implement WLC, including weighting overlay and map
algebra. There are numerous free and commercial ArcGIS add-ons implementing other
GIS-based MADM techniques (Boroushaki and Malczewski 2008; Marinoni 2004;
http://arcscripts.esri.com). Only two packages, IDRISI and CommonGIS, provide full
integration of MCDA (Nyerges and Jankowski 2010).
IDRISI (http://www.clarklabs.org) is a commercial GIS that includes decision-support
modules based on WLC, AHP, OWA, MOLA and CA, among others, plus a
wizard to assist in selection of appropriate decision techniques (Eastman 2009). It also
encourages and supports identifying contiguity of ‘solution’ areas, helping address the fragmentation
problem in raster-based MCDA. Figure 7 shows a spatially continuous example
of IDRISI’s WLC capabilities (Rinner 2003a). CommonGIS (http://www.commongis.
com), originally called ‘Descartes’, is a Java-based program that runs in a web browser or
as a desktop application, and provides a number of multi-criteria decision capabilities
including Ideal Point, WLC, OWA and Pareto Sets. Figure 8 shows a discrete
WLC example from Jankowski et al. (2001), depicting interactivity and map-graph
linking.
IDRISI is the only GIS package to have full coupling of a MODM method, the MOLA
heuristic described in Section 4.6. Mathematical optimisation is typically integrated by
loose coupling of GIS with packages or libraries such as those provided by Lindo (Malczewski
1999a; http://www.lindo.com/), or using custom programs to tightly couple the
algorithms (Ghosh 2008). An important question is the order of integration (van Herwijnen
and Rietveld 1999; Malczewski 2006a), as it can introduce biases related to the steps
performed by each tool. Fully integrated GIS-based MODM is required to flexibly address
this issue. Progress towards this goal has been made in the realms of nature conservation
and land-use planning, as several organisations have developed packaged add-ons tightly
coupled with ArcGIS (http://www.natureserve.org/prodServices/vista/overview.jsp;
http://gg.usm.edu/pat/overview.htm; http://www.placeways.com). Customisation and
integration generally also hide technical complexity, and therefore, work towards the goal
of accessibility. It is important, however, that the underlying methods and assumptions are
Table 2. Selected geographic information system-based multiple-criteria decision analysis
researchers.
Researcher Institution Link
Steve Carver University of Leeds http://www.geog.leeds.ac.uk/people/s.carver/
Salem Chakhar Universite´ Paris-Dauphine http://www.lamsade.dauphine.fr/~chakhar/
Suzana Dragic´evic´ Simon Fraser University http://www.sfu.ca/dragicevic
Ronald Eastman Clark University http://www.clarku.edu/academiccatalog
/facultybio.cfm?id=61
Piotr Jankowski San Diego State University http://geography.sdsu.edu/People/Faculty
/jankowski.html
Florent Joerin Universite´ Laval http://www.adt.chaire.ulaval.ca/1_chaire
/presentation_titulaire.php
Jacek Malczewski University of Western Ontario http://geography.uwo.ca/faculty/malczewskij
Oswald Marinoni Commonwealth Scientific and
Industrial Research Organisation
http://www.csiro.au/people/Oswald.Marinoni.html
Timothy Nyerges University of Washington http://faculty.washington.edu/nyerges
Claus Rinner Ryerson University http://www.ryerson.ca/~crinner
GIS-based MCDA 425
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
well documented, to avoid creating a black box that is not trusted. Trust in MCDA may
also be compromised if it is applied a posteriori to support preconceived decisions (Voogd
1983).
Fig. 7. IDRISI multiple-criteria evaluation example (from Rinner 2003a, reproduced with permission of the publisher).
Users specify criteria weights and optionally select constraints, then evaluate all locations within the study
area using a 0–255 rating scale. It employs a custom web-based interface to the non-Web IDRISI package.
426 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
7. Conclusion
This paper has provided an overview of the background and methods of MCDA, and its
spatial extension using GIS. Although research output, tools and applications in GIS-based
MCDA continue to expand, the field has not achieved widespread acceptance. One reason
is that it is often considered to be just an element of spatial decision support. Another reason
is the breadth and complexity of available methods, particularly when viewed from
the perspective of someone with little or no background in formal decision analysis. This
introduction to the field is but one step towards making GIS-based MCDA more accessible.
The need for cursory treatment of the methods selected for presentation here, and the
exclusion of many other techniques and important issues, speaks to the richness that awaits
those who choose to delve further into this field. In addition to continued refinement of
the underlying methods and improved integration of MCDA with GIS software, there are
many other opportunities for increasing accessibility. We conclude by highlighting two of
them: web-based delivery and improved visualisation.
The Internet is an obvious deployment platform for collaborative GIS-based MCDA
and decision support, and this approach is not new (Carver 1999; Mason and Dragic´evic´
Fig. 8. CommonGIS multiple-criteria evaluation example (from Jankowski et al. 2001, reproduced with permission
of the publisher), showing counties of Idaho measured on ten healthcare criteria. Interactivity includes the ability to
visually select counties in the map, and to set criteria weights using sliders. Links can be seen (i) between the
selected county and the textual information in the bottom right, (ii) between the highlighted counties in the map
and the parallel coordinates graph to the left, and (iii) among the criteria weights, the overall county score at the
bottom of the graph and the county shading in the map.
GIS-based MCDA 427
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
2006; Rinner 2003b; http://www.collaborativegis.com/). Web-based applications have
certainly helped the momentum of Participatory GIS (PGIS), a newer sub-discipline that
emerged from the GIS and society debates (Pickles 1995) as a broad research umbrella
regarding socio-political aspects of interest group engagement using GIS (Craig et al.
2002; Haklay and Tobo´n 2003; Jankowski and Nyerges 2001; Weiner and Harris 2008).
Researchers are beginning to explicitly combine MCDA and PGIS (Boroushaki and Malczewski
2010; Sima˜o et al. 2009) and it is possible that GIS-based MCDA will be increasingly
positioned as a component of PGIS. Regardless, an important element of PGIS that
GIS-based MCDA practitioners could embrace in order to promote broad acceptance is
incorporating traditional and local knowledge (McIntyre et al. 2008; Rantanen and Kahila
2009; Sheppard and Meitner 2005). Doing this effectively requires approaches that support
the exploration⁄structuring phase of decision processes, not just the evaluation⁄recommendation
phase, to avoid a biased pre-selection of criteria and alternatives (Ramsey
2009). Beyond the PGIS realm, GIS-based MCDA can look to Web 2.0 (Haklay et al.
2008; http://oreilly.com/web2/archive/what-is-web-20.html) for developments like
crowdsourcing (Hudson-Smith et al. 2009; Poore 2010), whereby members of the public
could suggest novel alternatives in a decision problem.
Geographic information systems and map-based applications have always provided visual
appeal. However, the visual element of the platform is far from stagnant, being driven by
the increasing expectations of web users and those performing advanced interactive analysis.
GIS-based MCDA could add to its limited visualisation research (such as Jankowski et al.
2001; Lidouh et al. 2009; Rinner 2007), by considering how to incorporate visualisation
advances from other fields.
Acknowledgement
In addition to their respective organizations, the authors gratefully acknowledge the support
of the Humber River Basin Project, the Institute for Biodiversity, Ecosystem Science
and Sustainability, MITACS⁄Accelerate, the Natural Sciences and Engineering Research
Council of Canada, and the Canada Foundation for Innovation. The authors also thank
Caroline Simpson of Natural Resources Canada and two anonymous referees for feedback
on the manuscript.
Short Biographies
Randal Greene recently completed an MSc in the Department of Geography at Memorial
University of Newfoundland, Canada. His thesis title is ‘Addressing Accessibility Challenges
of GIS-based Multiple-Criteria Decision Analysis for Integrated Land Management:
Case study in the Humber region of Newfoundland and Labrador, Canada’, which
involved selection and integration of GIS-based exploratory visualization and MCDA
techniques. After receiving a Bachelor of Commerce, he worked for 7 years developing
business software and databases. Then he worked for 10 years in marine navigation and
surveillance software, developing a keen interest in GIS. Randal currently works with
Nature Conservancy of Canada.
Rodolphe Devillers is an Associate Professor in the Department of Geography at
Memorial University of Newfoundland, Canada. He is also cross-appointed with the
Department of Earth Sciences at Memorial University and adjunct professor at the
Department of Geomatics Sciences at Laval University, Quebec. Devillers received a
PhD in Geomatics Sciences from Laval University in 2004. His research interests range
428 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
from more theoretical research on spatial data quality and decision-support systems to
more applied research in marine applications of GISciences for fisheries, conservation
biology and geology.
Joan E. Luther is a Research Scientist with Natural Resources Canada, Canadian Forest
Service, an Adjunct Professor at the De´partement de Ge´omatique Applique´, Universite´
de Sherbrooke, and a Professional Associate with Sustainable Resource Management,
Memorial University of Newfoundland. She specializes in the application of earth-observation
data and geospatial information for sustainable resource management. She completed
Bachelor and Master of Science degrees at the Department of Geography,
Memorial University of Newfoundland. Her research interests include the development
of remote sensing and geospatial methods for modelling and mapping forest inventory
parameters, monitoring landscape changes and associated impacts on ecosystem goods and
services, and providing information products to support ecosystem-based management.
Brian G. Eddy is a Research Scientist with Natural Resources Canada, Canadian Forest
Service. His research focuses on ecological risk analysis and decision support, and the
application of geomatics technology for ecosystem-based management. Brian obtained a
PhD in Geography and Environmental Studies from Carleton University, and a Master in
Earth Sciences from the University of Ottawa.
Note
* Correspondence address: Randal Greene, Department of Geography, Memorial University of Newfoundland,
St. John’s, NL, Canada A1B3X9. E-mail: rgreene@mun.ca.
References
Aerts, J. C. J. H., van Herwijnen, M., Janssen, R. and Stewart, T. J. (2005). Evaluating spatial design techniques
for solving land-use allocation problems. Journal of Environmental Planning and Management 48 (1), pp. 121–142.
Anderson, D. R., Sweeney, D. J. and Williams, T. A. (2003). An introduction to management science: quantitative
approaches to decision making. 10th ed. Mason, OH: South-Western College Publishing.
Baja, S., Chapman, D. M. and Dragovich, D. (2007). Spatial based compromise programming for multiple criteria
decision making in land use planning. Environmental Modeling and Assessment 12, pp. 171–184.
Balram, S. and Dragic´evic´, S. (eds) (2006). Collaborative geographic information systems. Hershey, PA: Idea Group
Publishing.
Belton, V. and Stewart, T. (2002). Multiple criteria decision analysis: an integrated approach. Boston, MA: Kluwer
Academic Publishers.
Bone, C. and Dragic´evic´, S. (2009). GIS and intelligent agents for multiobjective natural resource allocation: a reinforcement
learning approach. Transactions in GIS 13 (3), pp. 253–272.
Boroushaki, S. and Malczewski, J. (2008). Implementing an extension of the analytical hierarchy process using
ordered weighted averaging operators with fuzzy quantifiers in ArcGIS. Computers and Geosciences 34 (4), pp.
399–410.
Boroushaki, S. and Malczewski, J. (2010). Measuring consensus for collaborative decision-making: a GIS-based
approach. Computers, Environment and Urban Systems 34 (4), pp. 322–332.
Bouyssou, D., et al. (2006). Evaluation and decision models with multiple criteria: stepping stones for the analyst. New
York: Springer.
Brans, J. P. and Mareschal, B. (2005). PROMETHEE methods. In: Figueira, J., Greco, S. and Ehrgott, M. (eds)
Multiple criteria decision analysis: state of the art surveys. New York: Springer, pp. 163–196.
Carver, S. J. (1991). Integrating multi-criteria evaluation with geographical information systems. International Journal
of Geographical Information Systems 5 (3), pp. 321–339.
Carver, S. J. (1999). Developing web-based GIS ⁄ MCE: improving access to data and spatial decision support tools.
In: Thill, J. C. (ed.) Spatial multicriteria decision making and analysis: a geographic information sciences approach. Aldershot,
UK ⁄ Brookfield, VT: Ashgate, pp. 49–76.
Chakhar, S. and Mousseau, V. (2007). An algebra for multicriteria spatial modeling. Computers, Environment and
Urban Systems 31 (5), pp. 572–596.
Chakhar, S. and Mousseau, V. (2008). GIS-based multicriteria spatial modeling generic framework. International Journal
of Geographical Information Science 22 (11), pp. 1159–1196.
GIS-based MCDA 429
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
Chen, K., Blong, R. and Jacobson, C. (2001). MCE-RISK: integrating multicriteria evaluation and CIS for risk
decision-making in natural hazards. Environmental Modelling and Software 16 (4), pp. 387–397.
Craig, W. J., Harris, T. M. and Weiner, D. (eds) (2002). Community participation and geographic information systems.
London ⁄ New York: Taylor & Francis.
Diaz-Balteiro, L. and Romero, C. (2008). Making forestry decisions with multiple criteria: a review and an assessment.
Forest Ecology and Management 255 (8–9), pp. 3222–3241.
Dragic´evic´, S. (2008). Modeling and visualization for spatial decision support. Cartography and Geographic Information
Science 35 (2), pp. 75–76.
Duh, J. D. and Brown, D. G. (2007). Knowledge-informed pareto simulated annealing for multi-objective spatial
allocation. Computers, Environment and Urban Systems 31 (3), pp. 253–281.
Dykstra, D. P. (1984). Mathematical programming for natural resource management. New York: McGraw-Hill.
Eastman, J. R. (2005). Multi-criteria evaluation and GIS. In: Longley, P., Goodchild, M. F., Maguire, D.
J. and Rhind, D. W. (eds) Geographical information systems. 2nd ed. New York ⁄ Toronto: Wiley, pp. 493–
502.
Eastman, J. R. (2009). IDRISI Taiga: guide to GIS and image processing. Worcester, MA: Clark Labs.
Eastman, J. R., Weigen, J., Kyem, P. A. K. and Toledano, J. (1995). Raster procedures for multi-criteria ⁄ multiobjective
decisions. Photogrammetric Engineering and Remote Sensing 61 (5), pp. 539–547.
Feick, R. D. and Hall, G. B. (2004). A method for examining the spatial dimension of multi-criteria weight sensitivity.
International Journal of Geographical Information Science 18 (8), pp. 815–840.
Figueira, J., Greco, S. and Ehrgott, M. (eds) (2005). Multiple criteria decision analysis: state of the art surveys. New
York: Springer.
Geldermann, J. and Rentz, O. (2007). Multi-criteria decision support for integrated technique assessment. In:
Kropp, J. and Scheffran, J. (eds) Advanced methods for decision making and risk management in sustainability science.
New York: Nova Science Publishers, pp. 257–273.
Gemitzi, A., et al. (2007). Combining geographic information system, multicriteria evaluation techniques and fuzzy
logic in siting MSW landfills. Environmental Geology 51 (5), pp. 797–811.
Ghosh, D. (2008). A loose coupling technique for integrating GIS and multi-criteria decision making. Transactions
in GIS 12 (3), pp. 365–375.
Haklay, M., Singleton, A. and Parker, C. (2008). Web mapping 2.0: the neogeography of the GeoWeb. Geography
Compass 2 (6), pp. 2011–2039.
Haklay, M. and Tobo´n, C. (2003). Usability evaluation and PPGIS: towards a user-centred design approach. International
Journal of Geographical Information Science 17 (6), pp. 577–592.
Herath, G. and Prato, T. (eds) (2006). Role of multi-criteria decision making in natural resource management. In:
Using multi-criteria decision analysis in natural resource management. Aldershot, UK: Ashgate, pp. 1–10.
van Herwijnen, M. and Rietveld, P. (1999). Spatial dimensions in multicriteria analysis. In: Thill, J. C. (ed.) Spatial
multicriteria decision making and analysis: a geographic information sciences approach. Aldershot, UK ⁄ Brookfield, VT:
Ashgate, pp. 77–99.
Hudson-Smith, A., Batty, M., Crooks, A. and Milton, R. (2009). Mapping for the masses: accessing web 2.0
through crowdsourcing. Social Science Computer Review 27 (4), pp. 524–538.
Hwang, C. L. and Lin, M. J. (1987). Group decision making under multiple criteria: methods and applications. Berlin ⁄ New
York: Springer-Verlag.
Hwang, C. L. and Yoon, K. (1981). Multiple attribute decision making: methods and applications, a state-of-the-art survey.
Berlin ⁄ New York: Springer-Verlag.
Jacquet-Lagre`ze, E. and Siskos, Y. (2001). Preference disaggregation: 20 years of MCDA experience. European Journal
of Operational Research 130 (2), pp. 233–245.
Jankowski, P. (1995). Integrating geographical information systems and multiple criteria decision-making methods.
International Journal of Geographical Information Systems 9 (3), pp. 251–273.
Jankowski, P. (2006). Integrating geographical information systems and multiple criteria decision-making methods:
ten years after. In: Fisher, P. (ed.) Classics from IJGIS: twenty years of the international journal of geographical information
science and systems. London: CRC Press, pp. 291–296.
Jankowski, P., Andrienko, N. and Andrienko, G. (2001). Map-centred exploratory approach to multiple criteria
spatial decision making. International Journal of Geographical Information Science 15 (2), pp. 101–127.
Jankowski, P., Lotov, A. V. and Gusev, D. (1999). Application of a multiple criteria trade-off approach to spatial
decision making. In: Thill, J. C. (ed.) Spatial multicriteria decision making and analysis: a geographic information sciences
approach. Aldershot, UK ⁄ Brookfield, VT: Ashgate, pp. 127–148.
Jankowski, P. and Nyerges, T. L. (2001). Geographic information systems for group decision making: towards a participatory,
geographic information science. London ⁄ New York: Taylor & Francis.
Janssen, R., van Herwijnen, M., Stewart, T. J. and Aerts, J. C. J. H. (2008). Multiobjective decision support for
land-use planning. Environment and Planning B 35 (4), pp. 740–756.
Jiang, H. and Eastman, J. R. (2000). Application of fuzzy measures in multi-criteria evaluation in GIS. International
Journal of Geographical Information Science 14 (2), pp. 173–184.
430 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
Joerin, F., Desthieux, G., Beuze, S. B. and Nembrini, A. (2009). Participatory diagnosis in urban planning: proposal
for a learning process based on geographical information. Journal of Environmental Management 90 (6), pp. 2002–
2011.
Joerin, F., The´riault, M. and Musy, A. (2001). Using GIS and outranking multicriteria analysis for land-use suitability
assessment. International Journal of Geographical Information Science 15 (2), pp. 153–174.
Keeney, R. L. and Raiffa, H. (1976). Decisions with multiple objectives: preferences and value tradeoffs. New York:
Wiley.
van Kooten, G. C. and Bulte, E. H. (2000). The economics of nature: managing biological assets. Malden, MA: Blackwell
Publishers.
Lidouh, K., De Smet, Y. and Zimanyi, E. (2009). GAIA map: a tool for visual ranking analysis in spatial multicriteria
problems. Proceedings of the International Conference on Information Visualisation 13, pp. 393–402.
Liu, C., et al. (2006). Catchment-wide wetland assessment and prioritization using the multi-criteria decision-making
method TOPSIS. Environmental Management 38 (2), pp. 316–326.
Lotov, A. V., Bushenkov, V. A. and Kamenev, G. K. (2004). Interactive decision maps: approximation and visualization
of Pareto frontier. Boston, MA: Kluwer Academic Publishers.
Malczewski, J. (1999a). GIS and multicriteria decision analysis. New York: J. Wiley & Sons.
Malczewski, J. (1999b). Spatial multicriteria decision analysis. In: Thill, J. C. (ed.) Spatial multicriteria decision making
and analysis: a geographic information sciences approach. Aldershot, UK ⁄ Brookfield, VT: Ashgate, pp. 11–48.
Malczewski, J. (2004). GIS-based land-use suitability analysis: a critical overview. Progress in Planning 62 (1), pp. 3–
65.
Malczewski, J. (2006a). GIS-based multicriteria decision analysis: a survey of the literature. International Journal of
Geographical Information Science 20 (7), pp. 703–726.
Malczewski, J. (2006b). Ordered weighted averaging with fuzzy quantifiers: GIS-based multicriteria evaluation
for land-use suitability analysis. International Journal of Applied Earth Observation and Geoinformation 8 (4), pp.
270–277.
Malczewski, J. (2006c). Multicriteria decision analysis for collaborative GIS. In: Dragic´evic´, S. and Balram, S. (eds)
Collaborative geographic information systems. Hershey, PA: Idea Group, pp. 167–185.
Marinoni, O. (2004). Implementation of the analytical hierarchy process with VBA in ArcGIS. Computers and Geosciences
30 (6), pp. 637–646.
Marinoni, O. (2006). A discussion on the computational limitations of outranking methods for land-use suitability
assessment. International Journal of Geographical Information Science 20 (1), pp. 69–87.
Mason, B. C. and Dragic´evic´, S. (2006). Web GIS and knowledge management systems: an integrated design for
collaborative community planning. In: Balram, S. and Dragic´evic´, S. (eds) Collaborative geographic information systems.
Hershey, PA: Idea Group Publishing, pp. 263–284.
McHarg, I. L. (1969). Design with nature. Garden City, NY: Natural History Press.
McIntyre, N., Moore, J. and Yuan, M. (2008). A place-based, values-centered approach to managing recreation on
Canadian crown lands. Society and Natural Resources 21 (8), pp. 657–670.
Mendoza, G. A. and Martins, H. (2006). Multi-criteria decision analysis in natural resource management: a critical
review of methods and new modelling paradigms. Forest Ecology and Management 230 (1–3), pp. 1–22.
Myint, S. W. and Wang, L. (2006). Multicriteria decision approach for land use land cover change using Markov
chain analysis and a cellular automata approach. Canadian Journal of Remote Sensing 32 (6), pp. 390–404.
Nyerges, T. L. and Jankowski, P. (2010). Regional and urban GIS: a decision support approach. New York: Guilford
Press.
Pickles, J. (ed.) (1995). Ground truth: the social implications of geographic information systems. New York: Guilford Press.
Poore, B. S. (2010). WALL-E and the ‘‘many, many’’ maps: toward user-centred ontologies for the national map.
Cartographica 45 (2), pp. 113–120.
Possingham, H., Ball, I. and Andelman, S. (2000). Mathematical methods for identifying representative reserve
networks. In: Ferson, S. and Burgman, M. (eds) Quantitative methods for conservation biology. New York: SpringerVerlag,
pp. 291–305.
Ramsey, K. (2009). GIS, modeling, and politics: on the tensions of collaborative decision support. Journal of
Environmental Management 90 (6), pp. 1972–1980.
Rantanen, H. and Kahila, M. (2009). The SoftGIS approach to local knowledge. Journal of Environmental Management
90 (6), pp. 1981–1990.
Rinner, C. (2001). Argumentation maps: GIS-based discussion support for on-line planning. Environment and
Planning B 28 (6), pp. 847–863.
Rinner, C. (2003a). Teaching spatial decision analysis with IDRISI online. Paper presented at the 6th AGILE conference
on geographic information science, Lyon, France, 24–26 April.
Rinner, C. (2003b). Web-based spatial decision support: status and research directions. Journal of Geographic Information
and Decision Analysis 7 (1), pp. 14–31.
Rinner, C. (2007). A geographic visualization approach to multi-criteria evaluation of urban quality of life. International
Journal of Geographical Information Science 21 (8), pp. 907–919.
GIS-based MCDA 431
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd
Rinner, C. and Malczewski, J. (2002). Web-enabled spatial decision analysis using ordered weighted averaging
(OWA). Journal of Geographical Systems 4 (4), pp. 385–403.
Roy, B. (1996). Multicriteria methodology for decision aiding. Dordrecht: Kluwer Academic Publishers.
Roy, B. (2005). An overview of MCDA techniques today: paradigms and challenges. In: Figueira, J., Greco, S. and
Ehrgott, M. (eds) Multiple criteria decision analysis: state of the art surveys. New York: Springer, pp. 3–24.
Roy, B. and Vanderpooten, D. (1996). The European school of MCDA: emergence, basic features and current
works. Journal of Multi-Criteria Decision Analysis 5 (1), pp. 22–37.
Saaty, T. L. (1980). The analytic hierarchy process: planning, priority setting, resource allocation. New York ⁄ London:
McGraw-Hill International Book Co.
Saaty, T. L. (2005). The analytic hierarchy and analytic network process for the measurement of intangible criteria
and for decision-making. In: Figueira, J., Greco, S. and Ehrgott, M. (eds) Multiple criteria decision analysis: state of
the art surveys. New York: Springer, pp. 345–408.
Schmoldt, D. L., Kangas, J., Mendoza, G. A. and Pesonen, M. (2001). The analytic hierarchy process in natural resource
and environmental decision making. Dordrecht ⁄ Boston, MA: Kluwer Academic Publishers.
Shepard, R. B. (2005). Quantifying environmental impact assessments using fuzzy logic. New York: Springer.
Sheppard, S. R. J. and Meitner, M. (2005). Using multi-criteria analysis and visualisation for sustainable forest management
planning with stakeholder groups. Forest Ecology and Management 207 (1–2), pp. 171–187.
Sima˜o, A., Densham, P. J. and Haklay, M. (2009). Web-based GIS for collaborative planning and public participation:
an application to the strategic planning of wind farm sites. Journal of Environmental Management 90 (6), pp.
2027–2040.
Siskos, Y. (2005). UTA methods. In: Figueira, J., Greco, S. and Ehrgott, M. (eds) Multiple criteria decision analysis:
state of the art surveys. New York: Springer, pp. 297–343.
Store, R. and Kangas, J. (2001). Integrating spatial multi-criteria evaluation and expert knowledge for GIS-based
habitat suitability modelling. Landscape and Urban Planning 55 (2), pp. 79–93.
Turban, E. and Aronson, J. E. (2001). Decision support systems and intelligent systems. 6th ed. Upper Saddle River, NJ:
Prentice Hall.
Turner, R. K., Georgiou, S. G. and Fisher, B. (2008). Valuing ecosystem services: the case of multi-functional wetlands.
London ⁄ Sterling, VA: Earthscan.
Vincke, P. (1992). Multicriteria decision-aid. Chichester, UK ⁄ Toronto: Wiley.
Voogd, H. (1983). Multicriteria evaluation for urban and regional planning. London: Pion.
Weiner, D. and Harris, T. M. (2008). Participatory geographic information systems. In: Wilson, J. P. and Fotheringham,
A. S. (eds) Handbook of geographic information science. Malden, MA: Blackwell Publishing, pp. 466–480.
Weistroffer, H. R., Smith, C. H. and Narula, S. C. (2005). Multiple criteria decision support software. In: Figueira,
J., Greco, S. and Ehrgott, M. (eds) Multiple criteria decision analysis: state of the art surveys. New York: Springer, pp.
989–1010.
White, R., Straatman, B. and Engelen, G. (2004). Planning scenario visualization and assessment: a cellular automata
based integrated spatial decision support system. In: Goodchild, M. F. and Janelle, D. G. (eds) Spatially integrated
social sciences. Oxford, UK ⁄ Toronto: Oxford University Press, pp. 420–442.
Wisniewski, M. (2002). Quantitative methods for decision makers. 3rd ed. Toronto: Financial Times ⁄ Prentice Hall.
Yager, R. R. (1988). Ordered weighted averaging aggregation operators in multicriteria decision making. IEEE
Transactions on Systems, Man and Cybernetics 18 (1), pp. 183–190.
Zadeh, L. (1965). Fuzzy sets. Information and Control 8, pp. 338–353.
Zeleny, M. (1982). Multiple criteria decision making. New York: McGraw-Hill.
Zeng, H., Peltola, H. and Kelloma¨ki, S. (2007). Interactive clear-cut regime selection for risk management of wind
damage using reference point approach. Ecological Informatics 2 (2), pp. 150–158.
432 GIS-based MCDA
ª 2011 The Authors and Crown in the right of Canada Geography Compass 5/6 (2011): 412–432, 10.1111/j.1749-8198.2011.00431.x
Geography Compass ª 2011 Blackwell Publishing Ltd