Water use scenarios assessment using Multicriteria Analysis

JOURNAL OF MULTI-CRITERIA DECISION ANALYSIS J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) Published online 11 April 2011 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/mcda.457

Water Use Scenarios Assessment using Multicriteria Analysis ANTONIO BOGGIA and LUCIA ROCCHI DSEEA-University of Perugia, Perugia, Italy ABSTRACT Water management is a fertile terrain for research, and can be investigated using several methodological instruments and approaches. Multipurpose water resources, which allow for the contemporary presence of in-stream (fishing, swimming, etc.) and off-stream uses (agriculture, household, etc.), are particularly difficult to management due to increasing water demand and the conflict between consumptive and not consumptive uses. New scenarios in agricultural policy (Reform of the European Common Agricultural Policy) and new requests from local stakeholders (recreational activities, rising household water demand, etc.) involve use of decision support methods to find a balance between multiple potential uses. This study describes the application of multicriteria decision aid for choosing the best project for water management and local development among a set of alternatives, using the regime method. The regime method can integrate quantitative data with quality judgement and preference index. The study aims to verify if the regime approach is understandable for the decision maker and if it is suitable for use in ambiguous situations where no quantitative information is available. It is not always possible to get the ‘best alternative’ in multicriteria evaluation: in this study we found two top-rank alternatives, with minor differences, to submit to the decision maker. Copyright r 2011 John Wiley & Sons, Ltd. KEY WORDS:

Multipurpose water resource; multicriteria decision aid; regime method; sustainable local development; Stochastic Multicriteria Acceptability Analysis

1. INTRODUCTION The term ‘multi-purpose resource’ refers to the simultaneous use of a natural resource for several social and economic objectives (Munda et al., 1994). Water is, by definition, a multi-use resource and for this reason it is often involved in management plans, which can be defined as conflict analysis characterized by technical, socio-economic, environmental and political value judgements (Munda et al., 1994, Lahdelma et al., 2001). In management plans, conflicting economic values are often included (Daly, 1992); this means finding the right allocation (efficiency problem), distribution (equity problem) and scale (sustainability problem) for the resource object of the plan. Water management projects typically entail productive (e.g. irrigation), essential (e.g. household) and environmental (e.g. maintenance of aquatic ecosystem) uses. In some cases recreational and leisure uses (e.g. fishing, bathing, etc.) are also

*Correspondence to: DSEEA-University of Perugia, Borgo XX Giugno, 74, Perugia 06121, Italy. E-mail: [email protected]

Copyright r 2011 John Wiley & Sons, Ltd.

taken into account, as for multipurpose water basin. Such different uses of water generate conflict, because the resource is scarce and different interests and stakeholders are involved. Presence of contemporary in-stream and off-stream uses can be problematic; an in-stream use does not require water consumption, while an off-stream use does. By water consumption we mean a use that does not allow for further reuse of the same water. In this way, we have two types of consumption: quantitative drawing and qualitative degradation; both can reduce or eliminate the presence of in-stream use. The management can be complicated by the presence of opposing legal rights and economic interests. As an example, a river could be used both for drawing irrigation water (off-stream use) and for fishing (in-stream use). Due to a lack in energy production in that area, the local government could decide to create a hydropower plant on the river. In this case the economic need is in conflict with the presence of legal rights (drawing rights of farmers), although the use of water is not consumptive. Moreover, there is a conflict with the leisure activities (fishing) and the environment (increase of fishes death for the waterwheel and change in the water level). Received 29 November 2009 Revised 25 January 2011 Accepted 29 January 2011

126

A. BOGGIA AND L. ROCCHI

In this case, water management is involved in the local economic development: to find an instrument allowing a choice between the different scenarios of management and development is essential. This paper aims to contribute towards the solution of these situations, focusing in particular on water management and its importance for local sustainable development. The instrument chosen for the study is multicriteria analysis (MCA) and, in particular, the regime method. Our main objective is the analysis of the regime method in water management situations that involve the presence of both in-stream and offstream uses. We want to emphasize the hypothetical problems found in water management, the interactions between recreational and environmental needs and their capability in local development scenarios. We also provide evidence showing how the regime method can help in these situations. Then, we make a first application of Stochastic Multicriteria Acceptability Analysis (SMAA) to water management, to compare the outcome with the regime one. Section 2 describes the problem of water management, while in Section 3 the regime method is presented. Section 4 presents the case study. Section 5 shows the main results and discussion, and Section 6 reports the conclusion. 2. METHODS FOR DECISION AIDING IN WATER MANAGEMENT The literature on water management issues includes a wide variety of methods, each one focusing on a different aspect of the problem. There are two main questions in water management: the presence of new needs for water use and the decrease of water availability due to the consequences of climate change. We consider in this section the game theory approach, some technical models, random utility theory and MCA, because they represent the largest number of studies. Game theory is useful in cases of adversarial situations (Getirana et al., 2008; Salazar et al., 2007), involving several stakeholders who can mutually influence each other with their choices. In particular, game theory is suitable for international basin (Ansink and Rujs, 2008) or transboundary water management situations (Frisvold and Caswell, 2000), or in cases with a clear ‘rights scenario’ between stakeholders (Wang Copyright r 2011 John Wiley & Sons, Ltd.

et al., 2008) but with an interconnection of their gains. Moreover, game theory is also used for analyses of agreement on the management of a specific area (Ansink and Rujs, 2008) and in cases of pollution and restoration costs (Ni and Wang, 2007). Another common way to analyse water management issues is scenario simulation and modelling. With this technique, several scenarios are created through simulation to find the best technical use of the resource. In particular, the analysis is based on the evolution of future demand without regard to possible adversarial situations, but usually with a great attention to issues of uncertainty (Abolpour et al., 2007). These models can be very different from one another because they are built for a specific situation or problem (Bernhardi et al., 2000). It is possible to group these models into technical ones with (Koch and Gru¨newald, 2009) or without (Le´vite et al., 2003) an economic approach. In water management scenarios that include the management of environmental services and recreational uses, random utility theory has been applied. In these situations, there are neither clear acts or rights, nor an economic assessment and a value for the services. Random utility theory is used to find the preferences for water management and not to find the correct management (Johnson and Baltodano, 2004). MCA has been applied in several situations with very different aims. Romero and Rehman (1987) show how, among all the applications of MCA to natural resources, water management is the most common. The methods used are numerous: a good review of the main methods and widespread application of MCA is presented in Hajkowicz and Collins (2007), while Hajkowicz and Higgings (2008) compared the effectiveness of six of the most frequently used methods (weighted summation, range of value, PROMETHEE II, Evamix and compromise programming (CP)). MCA methods are used in very different situations, from planning under uncertainty (Bender and Simonovic, 2000), to evaluation of dairy effluent management options (Hajkowicz and Wheeler, 2008), or to land and water use under drought conditions (Munda et al., 1998). Hajkowicz and Collins (2007) reviewed 113 studies, showing how flexible the use of MCA can be. They focus their attention in particular on eight different areas of application, covering all aspects of water management. The most widely used category is fuzzy set analysis, followed by CP, J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

WATER USE SCENARIOS ASSESSMENT USING MULTICRITERIA ANALYSIS

127

Figure 1. Central weight vector.

analytic hierarchy process and the ELECTRE family (Hajkowicz and Collins, 2007). MCA is also often used to enforce and integrate other models, classified as technical approaches (Brouwer and van Ek, 2004), game theory (O¨zelkan and Duckstein, 1996) or before a costbenefit analysis (Wenzel, 2001). Hajkowicz and Collins (2007), in their analysis, did not find any evidence that one method is inherently better than another for the environmental valuation. We have found no instances of the regime method being applied to water management issues. However, we think it could be suitable for several reasons. The presence of an ordinal system for the weight vector means that the choice for the decision maker is easier, while finding a quantitative weight is quite difficult. The method is easily understood and transparent for the decision maker. Most MCA methods require preference information and only few of them can be used with ordinal (qualitative) criteria (Lahdelma et al., 2002). Regime allows both qualitative and quantitative information to be treated correctly. This characteristic seems to be present in water management contexts in which Copyright r 2011 John Wiley & Sons, Ltd.

clear data are not always available, but where preferences and priorities are often defined. In case of uncertain or inaccurate criteria and weights also, the family of SMAA methods (Tervonen and Figuera, 2008) is suitable particularly if there are several Decision Makers. We decide, for this reason, to compare briefly the results of our analysis with the regime with the output coming from one method of SMAA family, that allows use of ordinal information (Figure 1). 3. METHODOLOGY: THE REGIME APPROACH For our analysis we used a discrete multi-criteria assessment methodology: the regime method. This method is suitable for a wide range of assessment cases because it is able to work both with qualitative data (e.g. ordinal, binary, etc.) and with quantitative data, as well as with mixed data. It has been used to assess both projects and policies (Vreeker et al., 2002). The regime method is considered as an ordinal generalization of pairwise comparison methods, J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

128

A. BOGGIA AND L. ROCCHI

such as concordance analysis (Janssen, 1992). It is based on two kinds of input data: an effects table, which represents the (qualitative or quantitative) expected values of indicators for each of the alternative scenarios considered in the analysis, and a set of weights, representing the relative importance of each criterion, compared to the other ones. Considering i (i 5 1,y, I) alternatives and j (j 5 1,y, J) indicators. xji denotes the effect of alternative i according to indicator j. The priorities assigned to indicators are denoted in terms of weights wj (j 5 1,y, J). To establish a dominance relationship for each pair of alternatives, an index of concordance is used. It shows how much alternative i is better than alternative i0 . This index is obtained by summing the weights of the indicators included in the concordance set Cii00 that is the set of indicators for which alternative i is at least equal to alternative i 0 : X cii0 ¼ wj jeCii 0

The sign of v ¼ ciii  cii i for each pair of alternatives is the focus of this method. If the sign is positive, i is preferred to i0 ; if it is negative, the result is just the reverse (Janssen, 1992). The problem in the regime method is that we have only ordinal information about the weights, or at most, qualitative scores, usually measured on a plus and minus (/111) scale (Janssen, 2001). Due to the lack of information on the cardinal value of the weights, we do not have information on the size of the difference between the alternatives. We would not be able to determine an unambiguous result, such as a complete ranking of the alternatives (Janssen et al., 1990). For this reason, ordinal weights are transformed in cardinal data, and a set of quantitative weights reflecting the qualitative priority information is defined. If the regime procedure always finds the same sign for v for all the values in the set, a complete ranking of alternatives is possible. If not, the regime procedure finds the subset of weights giving the same sign to v, and their sizes. The sizes could be interpreted as the probability for i preferred to i 0 . Thus, the dominance of i with respect to i 0 is calculated using a probability procedure, estimating the probability index Pii0 : Pii0 ¼ probðvii0 40Þ Copyright r 2011 John Wiley & Sons, Ltd.

The aggregate probability measure is given by: pi ¼

X 1  pii0 I  1 i6¼i0

It is the measure of the probability that alternative i is preferred to all other alternatives. In this way, the final ranking is given by the order of the values p for the alternatives: the higher p is for an alternative, the higher is the position of that alternative in the ranking. To test the sensitivity of the results, we used the method SMAA with ordinal criteria measurements (SMAA-O) (Lahdelma et al., 2003). SMAA-O belongs to the family of SMAA methods that have been developed for discrete multicriteria problems, in case of uncertain or inaccurate data. SMAA can be used also in the case of scarce or any information on the weight vector and when there are several decision makers (Lahdelma et al., 1998; Tervonen and Figuera, 2008). The original SMAA (Lahdelma et al., 1998) provides a support in classification problems. SMAA-2 extends SMAA and allows for ranking problems, while the SMAA-O increase the possibility of the method using ordinal information on criteria as well (Tervonen and Lahdelma, 2007). The main results of SMAA-O analysis are as follows: rank acceptability indices, central weight vector and confidence factor of alternatives. The rank acceptability indices represent the probability of an alternative to get a certain rank. If the acceptability index is equal to 1, the alternative is dominant for any distribution of weights; if it is equal to 0, the alternative is inefficient. The central weight vector describes the typical preferences favouring each alternative: it is defined as the expected centre of gravity of the favourable weight space (Tervonen and Figuera, 2008). The confidence factor is the probability for an alternative to be preferred according to the preferences expressed by the weight vector (Tervonen and Figuera, 2008). 4. CASE STUDY Lake Montedoglio is located in the Upper Tiber River Valley in Tuscany, Italy. Compared to the rest of the region, the area is less developed for tourism; however, the area has many monumental, artistic and natural amenities. Even if the lake is J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

WATER USE SCENARIOS ASSESSMENT USING MULTICRITERIA ANALYSIS

artificial, by means of a dam, it presents several interesting natural characteristics. The local private and public stakeholders would like to take advantage of its presence, in synergy with the amenities in that area. At present the lake is used very little for recreational activities and the main ones are developed in the first tract of the river, just after the dam. Until now, off-stream uses have not been completely developed. Household use is increasing, while irrigation involves a smaller area than the original one, as infrastructure is still not complete. Due to the presence of changing off-stream uses, some additional management problems have arisen, but this has increased the potential for the development of recreational activities. The presence of multiple uses for the lake water, both in-stream and off-stream, results in some conflicts. In particular, it is difficult to reconcile some essential off-stream uses, such as household ones, and in-stream uses, so as to allow for a higher level of natural environmental quality in the aquatic ecosystems. Moreover, a new sensitivity towards environmental services and public opinion supporting increased tourist development in Lake Montedoglio push public stakeholders towards valorization of the area. This does not fit well with the presence of some off-stream uses. In particular, irrigation causes a significant decrease of water volume during the summer, which is also the main period for tourist activities. Several stakeholders can be identified, all interested in using the lake primarily to promote the surrounded area or its economic development: the management of the dam and the lake, the Province government and the local municipalities; only three of the eight municipalities considered already used Montedoglio for household water. The ‘interest groups’ are the local fishers association, the association for the tourism and the private citizens. The latest can be split between residents in the surrounded municipalities, more involved in touristic and recreational strategies, and other municipalities, that are more interested in the environmental services, in particular about interactions with Tiber, and in the agricultural use. From a legal point of view, in Italy, household use has priority over all the others, while irrigation immediately follows. Area surrounding the lake has been interested since the 60s by cultivation of water demanding and profitable crops, as tobacco. Changes in the Common Agricultural Policy in the Copyright r 2011 John Wiley & Sons, Ltd.

129

last few years have modified crop rotation and in several cases irrigation is no longer justified from an economic viewpoint. The presence of the lake in the area is still not completely exploited, but several projects are possible, each one reflecting the interests of one or more groups of stakeholders, which are also the promoters of the projects. As several institutional stakeholders are present in the area, the final decision has to be shared between the Management of the lake, the Province Government and the shore Municipalities. They are the Decision Makers and also the most important backers of each project. The following scenarios describe the projects that are being developed in the area. Recreational activities are central in all the scenarios, because they assure economic gain and because of a local preference for this kind of development in the area. However, all projects must respect household use. 4.1. The scenarios Proposals for future recreational development are oriented in four directions, each one corresponding to a particular scenario. Scenarios assessed in this paper are project proposed by local stakeholders and local governments to one or all the Decision Makers. We briefly present each scenario. 4.1.1. A—Sport fishing. Lake Montedoglio, like all large water reservoirs, is subjected to the phenomena of tail water: the release of water from the lake with a constant, quite cold, temperature all year round. The tail water phenomenon has changed the habitat of the river after the lake, increasing the quality of water and also of the aquatic ecosystems, and as a result there are more rare fish species. Moreover, some private stakeholders, organized in an association and assisted by the Province, have created a no-kill sanctuary on the river, just after the lake. The area is not completely developed, and some further investments are required. Every year, the area has 8000 visitors, but the number is increasing. 4.1.2. B—Sport and navigation. The lake is home to the small ‘Rowing and Sailing Club’. It is still quite small due to the lack of infrastructures on the shore. To increase the use of the lake by skiff, windsurf, kayak and other small vessels, the creation of a very small dock and some other complementary structures is required. J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

130

A. BOGGIA AND L. ROCCHI

4.1.3. C—Beach tourism. At present, swimming is not allowed in the lake, but this will soon be modified. After that, the lake will be used for swimming and as a beach. In this case, the presence of parking and complementary structures will also be required. As regards the increase of water activities, there are private interests that need public support. 4.1.4. D—Cultural and natural tourism. Lake Montedoglio is close to several cities famous for their artistic attractions: Arezzo, Citta` di Castello, Anghiari and San Sepolcro. The area is also interesting from a naturalistic viewpoint, and there are some natural routes that involve both the lake and the Tiber River. These are known as ‘Water Routes’, along an ancient mill and locks system. The tourist activities linked with ‘Water Routes’ are partially in conflict with the productive activities. In general, development of naturalistic tourist activities and the empowerment of cultural ones are much appreciated by the local public stakeholders and also by the sanctuary manager. This kind of tourism should be considered complementary to the fishing activities. 4.2. The criteria and their importance The following criteria represent the main objectives that the local government wants to achieve. They are compatible with the three dimensions of sustainable development: socio-political, environmental and economic (Pope et al., 2004). The full list is shown below:  Political criteria: n. of municipalities involved in each project; it represents the number of beneficiaries of each project: the larger the number of municipalities involved the better the project.

 Economic criteria: total cost of investment; the greater the expense, the worse the project is.  Socio-economic criteria: Private/public investment; it indicates the presence of a public or a private investments. Private investments are better than public ones.  Economic-environmental criteria: ratio of instream to off-stream uses; it is an evaluation of competing uses; if the index is low, the project increases competition and conflicts between in-stream and off-stream use.  Environmental criteria: water quality; for some projects a higher level of water quality is required; a higher level of water quality is better than a lower one.  Compatibility criteria: n. contemporary projects; a project that allows a larger number of other projects at the same time is preferred. As presented in the methodological section, the regime method allows for two different procedures, one using only qualitative information and ordinal data, the other using both quantitative and qualitative data. We had some cardinal information on the criteria n. 1 (number of municipalities involved), n. 2 (total cost of investment) and n. 6 (number of contemporary projects). For this reason we used the second procedure. The values of these indicators are reported in Table I, which shows the presence of ties in three vectors. Data have been processed using Definite 2.0 (Janssen et al., 2003)). We did not use a unique weight vector. In the study area there are different stakeholders with different needs and preferences on the scenarios, and the decision maker wanted to find a shared solution. Even if it is not possible to compare ordinal information on the weights and the final ranking between different groups of people, due to

Table I. Effect table, ordinal and cardinal scores Unit Municipalities involved Total cost of the investment Public/private investment In-stream/off-stream uses Water quality Combination of contemporary projects

n. Ordinal Ordinal Ordinal n.

C/B B C

B

A-sport fishing

B-sport and navigation

C-beach tourim

D-natural and cultural

3 5000 2 1 1 1

3 20 000 1 1 3 3

3 50 000 4 2 2 3

12 9000 3 2 3 7

Effect table represents the relative importance of each criterion in each scenario. Column 1 (Unit) shows qualitative (ordinal) and quantitative criteria, while Column 2 (C/B) shows whether the criterion is a cost or a benefit. Colon C/B is required to order the criteria from the biggest to the smallest or vice versa. Copyright r 2011 John Wiley & Sons, Ltd.

J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

131

WATER USE SCENARIOS ASSESSMENT USING MULTICRITERIA ANALYSIS

Table II. Stakeholders and their preferences Management

Public

Local private

Medium private

Tourism

Fishers

Average

Global

2 4 4 1 2 3

1 4 3 6 2 5

4 3 5 6 2 1

4 5 2 6 1 3

4 4 4 2 1 3

3 3 3 2 1 3

3 4 3 4 1 3

2 3 5 6 1 4

Municipalities involved Total cost of the investment Public/private investment In-stream/off-stream uses Water quality Combination of contemporary projects

Table shows the group of stakeholders present in the area and their preferences on the criteria. The table does not represent a weight table: numbers are the ordinal ranking of criteria. The best criteria for each group are indicated with number 1.

the ordinal nature of the analysis, we decided to make different rankings for each group. This choice was made to let the decision maker have a clear framework and understand possible overlappings in the ranking. The weight vectors were built from different starting points. Preferences of private, both the surrounding and the further, was derived from a questionnaire about the area citizens’ preference for the use of the lake. The questionnaire was built for a study based on choice experiments and asked about preferences over several scenarios set apart for in-stream/off-stream activities, recreational structure and water quality. The scenarios showed are not the same presented in this study, but the result was a high level of support for off-stream uses (low preference for criteria n. 5), a strong preference for the conservation of water quality (high preference for criteria n. 6) and recreational structure and activities (high preference for criteria n. 2). The questionnaire also showed differences between people living in the surrounding area and people in more remote areas, and between the sport fishers and all the others. Moreover, the same type of questionnaire submitted to tourists allows us to highlight their weight vector. The preference of the lake manager and of the public decision makers were directly asked. Eight groups of stakeholders are identified and are reported in Table II. Table II also shows stakeholders’ preferences in an ordinal way, where 1 represents the most preferred criterion. Preferences are taken into account for the construction of the weight vectors. The difference between local private and medium private groups is that the former represents people living in the surrounding area, while the latter are more remote, but are included in the area concerned with the lake. The average represents an ‘arithmetic mean’ of the preferences, Copyright r 2011 John Wiley & Sons, Ltd.

while global is a sort of weighted mean that takes into account how large each group of stakeholders is and tries to find a balance with the principles of sustainable development. As the information on weights are ordinal, both the averaged and global vectors represent an aggregation of preference and they are not mathematical means. The weight vectors are the following: wm ¼ðw4 ; w1 ; w5 ; w6 ; w2 ; w3 ÞT ¼ ð6; 5; 5; 4; 3; 3Þ wp ¼ðw1 ; w5 ; w3 ; w2 ; w6 ; w4 ÞT ¼ ð6; 5; 4; 3; 2; 1Þ wlp ¼ðw6 ; w5 ; w2 ; w1 ; w3 ; w4 ÞT ¼ ð6; 5; 4; 3; 2; 1Þ wmp ¼ðw5 ; w3 ; w6 ; w1 ; w2 ; w4 ÞT ¼ ð6; 5; 4; 3; 2; 1Þ wt ¼ðw5 ; w4 ; w6 ; w1 ; w3 ; w2 ÞT ¼ ð6; 5; 4; 3; 3; 3Þ wf ¼ðw5 ; w4 ; w6 ; w1 ; w3 ; w2 ÞT ¼ ð6; 5; 4; 4; 4; 4Þ wa ¼ðw5 ; w1 ; w3 ; w6 ; w2 ; w5 ÞT ¼ ð6; 4; 4; 4; 3; 3Þ wg ¼ðw5 ; w1 ; w2 ; w6 ; w3 ; w4 ÞT ¼ ð6; 5; 4; 3; 2; 1Þ As shown above, full weight vectors are presented only four times, due to indifference between some criteria for some groups of stakeholders. 5. RESULTS AND DISCUSSION The eight weight vectors produce results quite similar in seven cases, apart from the management, as shown in Table III. Alternative C, Beach tourism, the worst, had consistently very bad results. Alternatives A, Sport fishing, and D, Natural and cultural tourism, were dominant solutions: they usually ranked first or second, while B, Sport and navigation, is third. However, for the management, alternative B, Sport and navigation, improved its rank, while D, Natural and cultural tourism, weakened. Note in Table III J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

132

A. BOGGIA AND L. ROCCHI

the value of the first rank alternative: it always has a very high probability value. In the regime method it is not possible to make a sensitivity analysis for the weight vector. The reason is clear: the method is based on the assumption that the ranking is a result of the ordinal weight vector, image of the true cardinal vector. However, it is possible to apply both sensitivity analysis and uncertainty analysis to the scores. The scope of uncertainty analysis involves testing the scores, while considering them as not perfectly defined. Score uncertainty is expressed by a percentage value that represents the possibility of change. For example, if a score is equal to 100 and we test such data for a percentage of uncertainty of 10%, this means the score can assume values between 90 and 110 and this may affect the final ranking. Uncertainty can be tested only for some criteria or for all criteria, and for both ordinal and cardinal data. The result of the analysis is a new ranking and also new probability and dominance tables. The former measures the probability of achieving a rank for an alternative, while the latter measures the dominance of one alternative over the others. If the ranking does not change, the evaluation can be considered stable. The final results were based on a certain amount of random drawing, usually 2000 or more. The minimum number of Monte Carlo iterations derives from

(Milton and Arnold, 1995): Kw ¼

1:962 4d2b

In our case, we wanted to achieve error limits of 0.01 for uncertainty analysis with 95% confidence we need to perform at least 9604 Monte Carlo interactions; we decided to run 10 000. The levels of uncertainty used in our work were two: high (25–50%) and very high (75–100%) level of uncertainty. Apart from the results of stakeholders Medium and Global, the final ranking did not change, even using a very high level of uncertainty. Both for Medium and Global, using a high uncertainty level (75 and 100%), alternative A, Sport fishing, improved its position, joining alternative D (Tables IV–VI). The sensitivity analysis showed good stability of the ranking, at the average, with a few reversal points. Even if there were some differences among the results of the group, some consistencies were present. The most stable criteria were n. 3 (Private/ public investment), n. 5 (water quality) and n. 6 (In-stream/off-stream uses). Such criteria had no reversal points, or when it occurred they were very far from the real value of the score. On the contrary, the criterion n. 1 ‘n. of municipalities involved’ showed greater instability, with several reversal points between the top two solutions and the

Table III. Ranking results Management A-sport fishing B-sport and navigation C-beach tourim D-natural and cultural

0.91 0.54 0.11 0.43

1 2 4 3

Public 0.75 0.34 0.08 0.83

Local private

2 3 4 1

0.57 0.39 0.11 0.93

Medium private

2 3 4 1

0.69 0.45 0.13 0.72

2 3 4 1

Tourism

Fishers

Average

Global

0.93 0.35 0.23 0.49

0.93 0.39 0.22 0.46

0.79 0.38 0.15 0.98

0.81 0.24 0.12 0.83

1 3 4 2

1 3 4 2

2 3 4 1

2 3 4 1

Table shows both the scores obtained and the final ranking of each alternative, according to the different weight vector for each group of stakeholders. The first column represents the scores, and the second, the rank.

Table IV. Final ranking after uncertainty Management Public Local private Medium private Tourism Fishers Average Global A-sport fishing B-sport and navigation decreased C-beach tourism D-natural and cultural tourism

1 3 4 2

5 5 5 5

2 3 4 1

5 5 5 5

2 3 4 1

5 5 5 5

1 2 3 1

 5 5 5

1 3 4 2

5 5 5 5

1 3 4 2

5 5 5 5

2 3 4 1

5 5 5 5

1 2 3 1

 5 5 5

In the table are shown the new rankings after the analysis of uncertainty. For each group of stakeholders the rank obtained after the analysis is reported (first column). The second column presents an equal sign if the rank is not changed in comparison with the previous ranking, and a arrow in case of change. Only alternative A better its position in two cases (Medium private and Global). Copyright r 2011 John Wiley & Sons, Ltd.

J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

133

WATER USE SCENARIOS ASSESSMENT USING MULTICRITERIA ANALYSIS

Table V. Rank acceptability indices

A-sport fishing B-sport and navigation decreased C-beach tourism D-natural and cultural tourism

I rank

II rank

III rank

IV rank

0.26 0.05 0.00 0.69

0.47 0.32 0.01 0.21

0.22 0.60 0.07 0.10

0.05 0.02 0.92 0.01

In the table are reported the acceptability indices for the four alternatives. The acceptability index represent the possibilities for an alternative to get a rank according to all the possible weights distribution.

Table VI. Confidence factor and central weight vector CF C-beach tourism B-sport and navigation decreased A-sport fishing D-natural and cultural tourism

Municipalities Total cost of the Public/private In-stream/off- Water Combination of involved investment investment stream uses quality contemporary projects

0.06 0.55

0.00 0.07

0.02 0.10

0.01 0.14

0.17 0.45

0.60 0.12

0.20 0.11

0.95 1.00

0.08 0.21

0.19 0.16

0.23 0.14

0.17 0.14

0.25 0.14

0.08 0.21

The first column (CF) represents the confidence factor. CF is defined as the probability for an alternative to be the most preferred, while the central weight vector show which is the best alternative according to each factor. It represents the choice of a generically stakeholder.

bottom two, for all the alternatives. In some cases, for this criterion there were also reversal points between the top and bottom solutions. The other criteria presented an intermediate level of sensitivity. The most stable group of preference was represented by the stakeholder Global. This result is quite good because Global is an artificial group made up using the ‘weighted average’ of all the others. After the analysis with the regime, we decided to make a comparison using another MCA method, as several studies use to test the sensitivity of the results (Hajkowicz and Collins, 2007). Few MCA methods allow for using ordinal criteria and uncertainty in the score. A suitable method is SMAA-O. We used both acceptability index and the weight vector to compare the outcome of regime. The acceptability indices clearly show as worst alternative the C, Beach tourism, with 92% of probability. The other three alternatives have lower values; according to the acceptability indices, alternative D ranks first position, A the second and B the third. The weight vector shows which alternative is preferred according to each criterion. Alternative D, the most preferred, is first only according to the first criteria while it never has the worst position in any case. The high value of CF for the alternative A does not allow to exclude it. Copyright r 2011 John Wiley & Sons, Ltd.

Analysis via regime method shows a clear rank for the alternatives B, Sport and navigation, and C, Beach tourism, the bottom ones, but not for the top solutions. Alternatives A, Sport fishing, and D, Natural and cultural tourism, were clearly the most preferred but neither is the best. The public Decision Makers prefer alternative D, Natural and cultural tourism. However, the results of analysis show interest also for alternative A, Sport fishing, by a large group of stakeholders and after the analysis of uncertainty alternative A raises the position of alternative D. Also the analysis with SMAA-O shows a not clear predominance of alternative D over alternative A. Although the acceptability index for alternative D is higher than for A, the confidence indices for both alternatives are very close. For these reasons the decision makers are inclined to integrate part of scenario A in scenario D. Decision makers were aware of the whole assessment process, since they have been involved through depth interviews at each stage of the analysis, to share information and achieve a better result. 6. CONCLUSIONS The regime method is one of the MCA methods used in the case of ordinal scores. This method is J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

134

A. BOGGIA AND L. ROCCHI

based on the use of an ordinal weight vector, which is easier to understand and explain to the decision maker. The method can also be used with cardinal information on the score, and, although it was created to treat the ordinal weight vector, it is possible to also treat the cardinal weight vectors. However, in our application, we were interested in an ordinal weight vector because of its transparency and flexibility. We found that the regime method fits particularly well when decision makers need simple input information, such as ordinal weights, due to the complexity of the decision. In this case, the complexity is related not only to the matter (water uses are always contested and conflict-ridden) but also to the presence of several stakeholders in the area with distinct needs. Using the regime it is possible for the decision maker to understand and to take into account all these needs. Future applications also using the family of SMAA seem to be recommended due to their capability to handle situation with any information at all about preferences. In MCA applications it is not always possible to identify a unique solution, while it is easier to find a group of best solutions. In this case, we identified two top-rank solutions; however, the alternative D seems to be better than A, but only slightly. Moreover, the uncertainty analysis shows no significant differences between the two in the case of a high level of uncertainty.

REFERENCES Abolpour B, Javan M, Karamouz M. 2007. Water allocation improvement in river basin using Adaptive Neural Fuzzy Reinforcement Learning approach. Applied Soft Computing 7: 265–285. Ansink E, Rujs A. 2008. Climate change and the stability of water allocation agreements. Environmental Resources Economics 41: 249–266. Bender MJ, Simonovic SP. 2000. A fuzzy compromise approach to water resource systems planning under uncertainty. Fuzzy Sets and Systems 115: 35–44. Bernhardi L, Beroggi G, Moens MR. 2000. Sustainable water management through flexible method management. Water Resources Management 14: 473–495. Brouwer R, van Ek R. 2004. Integrated ecological and social impact assessment of alternative flood control policies in the Netherlands. Ecological Economics 50: 1–21. Daly HE. 1992. Allocation, distribution, and scale: towards an economics that is efficient, just, and sustainable. Ecological Economics 6: 185–193. Copyright r 2011 John Wiley & Sons, Ltd.

Frisvold GB, Caswell MF. 2000. Transboundary water management: game-theoretic lessons for projects on the US-Mexico border. Agricultural Economics 24: 101–111. Getirana A, Malta V, Soares de Azevedo J. 2008. Decision process in a water use conflict in Brazil. Water Resources Management 22: 103–118. Hajkowicz S, Collins K. 2007. A review of multiple criteria analysis for water resource planning and management. Water Resources Management 21: 1553–1566. Hajkowicz S, Higgings A. 2008. A comparison of multiple criteria analysis techniques for water resource management. European Journal of Operational Research 184: 255–265. Hajkowicz SA, Wheeler SA. 2008. Evaluation of dairy effluent management option using multiple criteria analysis. Environmental Management 41: 613–624. Janssen R. 1992. Multiobjective Decision Support for Environmental Management. Kluwer Academic Publishers: Dordrecht. Janssen R. 2001. On the use of multi-criteria analysis in environmental impact assessment in The Netherlands. Journal of Multi-Criteria Decision Analysis 10: 101–109. Janssen R, Nijkamp P, Rietveld P. 1990. Qualitative multicriteria methods in the Netherlands. In: Readings in Multiple Criteria Decision Aids, Bana CA, Costa E (eds). Springer: Heidelberg; 383–409. Janssen R, van Herwijnen M, Beinat E. 2003. Definite: Case Studies and User Manual. Report Number R-03/ 03. Institute for environmental studies, Amsterdam. Johnson NL, Baltodano ME. 2004. The economics of community watershed management: some evidence from Nicaragua. Ecological Economics 49: 57–71. Koch H, Gru¨newald U. 2009. A comparison of modelling systems for the development and revision of water resources management plans. Water Resources Management 23: 1403–1422. Lahdelma R, Hokkanen J, Salminen P. 1998. SMAA— stochastic multiobjective acceptability analysis. European Journal of Operational Research 106: 137–143. Lahdelma R, Miettinen K, Salminen P. 2003. Ordinal criteria in stochastic multicriteria acceptability analysis (SMAA). European Journal of Operational Research 1: 117–127. Lahdelma R, Salminen P, Hokkanen J. 2001. Using multicriteria methods in environmental planning and management. Environmental Management 26: 595–605. Lahdelma R, Salminen P, Hokkanen J. 2002. Locating a waste treatment facility by using stochastic multicriteria acceptability analysis with ordinal criteria. European Journal of Operational Research 142: 345–356. Le´vite H, Sally H, Cour J. 2003. Testing water demand management scenarios in a water-stressed basin in South Africa: application of a WEAP model. Physics and Chemistry of the Earth 28: 779–786. J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda

WATER USE SCENARIOS ASSESSMENT USING MULTICRITERIA ANALYSIS

Milton JS, Arnold JC. 1995. Introduction to Probability and Statistics (3rd edn). McGraw-Hill International editions, New York. Munda G, Nijkamp P, Rietveld P. 1994. Qualitative multicriteria evaluation for environmental management. Ecological Economics 10: 97–112. Munda G, Parruccini M, Rossi G. 1998. Multicriteria evaluation methods in renewable resource management: integrated water management under drought conditions. In Multicriteria Analysis for Land-Use management, Beinat E, Nijkamp P (eds). Kluwer Academic Publishers: Dordrecht. Ni D, Wang Y. 2007. Sharing a polluted river. Games and Economic Bahavior 60: 176–186. Pope G, Annandale D, Morrison-Saunders A. 2004. Conceptualising sustainability assessment. Environmental Impact Review 24: 595–616. Romero C, Rehman T. 1987. Natural resource management and the use of multicriteria criteria decision making techniques: a review. European Review of Agriculture Economics 14: 61–89. Salazar R, Szidarovszky F, Coppola E, Rojano A. 2007. Application of game theory for a roundwater conflict

Copyright r 2011 John Wiley & Sons, Ltd.

135

in Mexico. Journal of Environmental Management 84: 560–571. Tervonen T, Figuera JR. 2008. A survey on stochastic multicriteria acceptability analysis methods. Journal of Multi-Criteria Decision Analysis 15: 1–14. Tervonen T, Lahdelma R. 2007. Implementing stochastic multicriteria acceptabillity analysis. European Journal of Operational Reseach 178: 500–513. Vreeker R, Nijkamp P, Ter Welle C. 2002. A multicriteria decision support methodology for evaluating airport expansion plans. Transportation Research Part D: Transport and Environment 1: 27–47. Wang L, Fang L, Hipel KW. 2008. Basin-wide cooperative water resources allocation. European Journal of Operational Research 190: 798–817. Wenzel V. 2001. Integrated assessment and multicriteria analysis. Physics and Chemistry of the Earth 26: 541–545. O¨zelkan EC, Duckstein L. 1996. Analysing water resources alternatives and handling criteria by multi criterion decision techniques. Journal of Environmental Management 48: 69–96.

J. Multi-Crit. Decis. Anal. 17: 125–135 (2010) DOI: 10.1002/mcda