A Risk-Based Multi-Criteria Decision Analysis Approach to Evaluating Transboundary Water Development- The Case of Lower Mekong River Basin

The Lower Mekong River basin (LMB) covers the lower part of the Mekong river basin, including Laos, Thailand, Cambodia and Vietnam. Due to numerous pressures from high population growth and intensive hydropower development, the LMB has been facing significant challenges concerning its biodiversity and ecosystem. In 2017, Mekong River Commission (MRC), an intergovernmental organisation founded in 1995 among LMB countries, established the Council Study, which analysed the impacts of water development scenarios concerning the environmental, socioeconomic aspects of the LMB. This paper explores the nature of risks to the LMB water development and subsequently evaluates LMB’s water development scenarios described in the Council Study by using a multi-criteria decision analysis (MCDA) method. MCDA method has been widely applied in the field of water resource management in order to assist the decision-making process by systematically evaluating a certain number of alternatives against well-selected criteria through a preference rating scheme. By implementing a risk-based comprehensive assessment of the LMB transboundary water, this study provides insights into the impacts of the increasing risks to the ecosystem and human beings on the water development of the basin over time, which assists to change the awareness and the perspective toward humans’ risks and transboundary river ecosystem of decision-makers. This paper provides valuable recommendations for MRC to improve their policy concerning benefit-sharing scheme, water planning and risk mitigation strategies.


Introduction
There are around 280 international river basins worldwide which contribute to approximately 60 percent of global water flows and accommodate nearly two-fifth of the world's inhabitants [1] [2]. The transboundary river systems, which might be shared among several countries with high socio-economic and environmental interdependencies, strongly support people's lives and the global biodiversity [3]. Given the vital role of shared river basins and the differences of riparian countries' interest concerning their development priorities, managing these water resources in the context of environmental and social rapid changes is an enormous challenge. These problems are expected to become more severe over time, resulting from the increasing risk and uncertainty linked with population growth, climate change and other water-related evolution [4] [5]. Therefore, it is crucial for decision-makers in shared river basins to understand the nature of risk to the basin development to implement risk-based evaluation and then take essential steps to integrate these assessments into their water planning process.
The decision-making process related to water use and distribution often draws much attention to economic and political rather than social concerns [4]. When people cannot use water sustainably and equitably, the exploited ecosystem and humans dependent on them will be first and foremost affected seriously [6]. According to UN World Water Development Report, freshwater ecosystems face enormous negative impacts from direct sources, such as water quality degradation, and indirect sources arising from socio-economic changes [6] [7]. Hence, the ecosystem approach might be considered a promising approach to support the sustainability and integration of water management [8]. Decision-makers must also better understand and gradually change their perspectives toward humans' risks and transboundary river ecosystem. Additionally, risk-based comprehensive assessment of transboundary water development is essential, not only for providing insights into the impacts of risks on the final evaluation but also for supporting decision-making, planning and management processes in the long term.
The Mekong river, one of significant international freshwater, rises in China and then runs across other five countries before joining the South China Sea. of floods and droughts and growing population as well as threats to water quality, ecosystem, public health and human welfare [9] [13] [14]. Notably, in 2018, MRC has identified key challenges with their alarming rates, by which the level of concerns about these challenges was addressed [9]. The risks discussed in many MRC reports provide an overview of existing and potential risks without exactly pointing out the risk level, and more importantly, comparing the Mekong's situation with other river basins' to stress the need for taking these risks  A study evaluating the water development scenarios in the context of transboundary river basin can identify the existing and future impacts of water-related issues on the basin's long-term development. The literature has been focused solely on modelling the water development scenario [18], analysing basin's hydro political tensions [19] or generally assessing the basin's water policy [20], while less concentrated on carrying out a risk-based evaluation as in the LMB case. Also, previous studies have mainly analysed the water allocation and basin development scenarios regarding local river basins rather than water de-  [28]. These studies are essential to understand the existing and potential risks to the long-term development of the basin, which will support the water scenario evaluation process. However, they have not evaluated and ranked the basin development scenarios using the generated impacts. Particularly, there is no study utilising the scenarios described in the MRC Council Study to implement the water development evaluation.
Given the reasons above, this paper explores the nature of risks to the MRB water development and subsequently evaluates LMB's water development scenarios to answer the following research questions: 1) What are the characteristics of the risks to the transboundary river's ecosystem and human beings, and 2) How do river basin's risk levels affect the evaluation of transboundary water development scenarios.
This paper is organised as follows. The following section presents the literature relating to the water resource management of the LMB and previous studies on transboundary water resource evaluation. The research methodology, research design and data collection process are accordingly introduced. Then, the research results are presented, and the findings are discussed, identifying the key issues that MRC and all LMB countries should take into account during their water planning and evaluating process. The last sections involve conclusions with policy implications and recommendations for future water resource management and cooperation in the LMB.

Theoretical Framework
Water resource management is a complex and interdisciplinary issue, which involves multi-sectoral areas [29]. In water management, decision-makers face enormous challenges in making the final selection since the decision-making process is characterised by multiple conflicting objectives, including economic,  [30]. Due to the complexity of transboundary river, many significant factors need to be considered when carrying out an assessment of transboundary water development. Therefore, along with the increasing awareness of shared water bodies' vital role, many researchers have generated interest in employing a multi-criteria decision analysis (MCDA) approach to support the decision-making process relating to transboundary water planning and managing. MCDA, a widely used method to assist evaluation and management decisionmaking issues, is primarily concerned with situations in which decision-makers systematically evaluate a certain number of alternatives against well-selected criteria through a preference rating scheme [31] [32] [33]. MCDA is both an approach and a compilation of steps aiming at providing an aggregate ranking of alternatives, from the most favored to the least preferred one. The detailed steps of a MCDA model include: 1) Identify the decision objectives and the alternatives; 2) Identify the criteria for assessing the alternatives; 3) Score the criteria weights which reflect the relative importance of each criterion to the decision-making process; 4) Calculate the overall weights of alternatives against the criteria; 5) Rank the alternatives and make the final decision. There are two quantified steps which should be noted when applying MCDA assessment. First, the relative weights of criteria are calculated using normalization technique (Table 1).
Second, the overall preference score of each alternative is equal to the weighted average of its score across all criteria, which can be calculated by the following formula: where: ij s is the preference score of the i th alternative on the j th criterion, w j is the weight of the j th criterion, S i is the overall preference score of the i th alternative.
Over the past decades, a large body of literature has applied MCDA in the field of water resource management and planning ( [35] [36] [37]. Besides, the MCDA model has been applied in many other water-related sectors, such as groundwater management [38]   There is a growing trend in using the MCDA method to evaluate water management alternatives concerning local/national river basin issues [30] [46] [47] [48] rather than at international river basin issues. The past research has primarily applied the MCDA model to analyse water governance and water policy of international water organisations [20] [49]. Previous studies have yet to explore the possibility of utilising risk-based criteria in the MCDA model to assess water development scenarios in the transboundary river basin.

Data Collection
The data extracted from the river basins assessment of Transboundary  Based on a set of indicators classified into five main indicators and fifteen sub-indicators (Table 2), the TWAP assessment presents the assessment results and background information of 286 transboundary river basins at the basin and basin country unit level. The risk levels by indicators of each river basin were finally analysed by adopting five categories from "very low" to "very high" with the corresponding points from 1 to 5.

Methods
First, a comparative analysis approach is used to construct a complete risk pro-  MCDA is an organized and transparent approach to problem solving that improves objectivity and produces trusted outcomes with a reasonable satisfaction level [51]. The application of MCDA method has been rapidly increasing due to the fact that the decision-making process relating to urgent real-life problems requires the consideration of multiple criteria [52]. In addition, this approach is distinguished by its ability to improve the decision's quality as well as to handle a variety of conflicting objectives with different perspectives of many stake-holders involving in the policy-making process [53]. A

Weighting method
Ranking and Weighted sum method scenarios concerning the assessment criteria, which are categorised in the next step.  Selecting the criteria for assessment A set of assessment criteria is carefully selected considering the characteristics of the LMB's risk profile and the MRC Council Study's data availability. The criteria presented in this study address the most intense pressures on the sustainable development of the environment, ecosystem and livelihoods of the LMB (Table 3).
Additionally, because the TWAP indicators were constructed applicable to any of the 286 river basins, it was challenging and impractical to incorporate the complete set of TWAP indicators into the LMB case's MCDA model. Finally, the author took a total of four main criteria and nine sub-criteria used in the MCDA model from a set of five main indicators and fifteen sub-indicators of the TWAP river basins assessment (Table 4).  Determining the criteria weights In a basic MCDA model, the decision-makers involved in weighing criteria are selected based on their expertise, academic knowledge and practical experiences. The decision-makers then assign criteria weights after considering their preferences. However, this decision-making process has some drawbacks because the decision-makers often hold different perceptions of the same problem, making it difficult to reach the final decision on the relative weights of criteria. More importantly, inconsistency problems are likely to occur when making decisions subjectively. To avoid these difficulties, the author employs the objective instead of the subjective weighting process. The scores for main and sub-criterion were assigned using the MRB risk values. The relative weights of the main criteria are then obtained by taking the average of their respective sub-criteria risk values while those of the sub-criteria are equal to their risk values.

A Comparison between the MRB and the Other Basins Worldwide
According to the TWAP intensive assessment, the relative risks related to a wide range of sustainable development issues were analysed for 286 global transboundary rivers, including the MRB, to identify river basins with significant problems and high potential risks. The risk points by criteria were calculated from basin to basin, and the detailed risk points of member countries of each river basin were also obtained. By sequentially calculating and comparing the global average risk points with the MBR's risk points, the typical variation of risk points across criteria is revealed, and that portrays the fundamental differences between the MRB and global trend. As shown in Figure 3, the threats to the MRB are generally higher than those at the global average scale across fifteen sub-criteria, especially regarding exposure to floods and droughts, economic dependence on water resources, and threat to fish. While the global average risk points are ranging from the lowest of 1.95 for the "environmental water stress" criteria to the highest of approximately 4.0 for the "wastewater pollution" criteria, the findings show that the risk points of the MRB reach a peak of 5.0, which is 2.5 times higher than its lowest level. Of all the criteria, wastewater pollution has become one of the biggest challenges facing both the MRB and river basins worldwide. More importantly, compared to the heavy dependence of economic development on water resources, it might be a worrying matter that fisheries, a crucial source of livelihood for the MRB's riparian countries, reaches the highest risk level among other issues. With the average risk level ranked 30th out of 286 international river basins, there is no doubt that the MRB is a hot spot for major river-basin-related problems.  Risk projections for the future development of global transboundary river basins were made for 2030 and 2050, focusing on four primary sources of water stress. The differences between the projected risks of the MRB and the world are shown in Figure 4. The MRB's expected risk values are lower than those of the global river basins across all criteria except nutrient pollution, the issue with the MRB's highest projected risk value. The growing concern over nutrient balance in the LMB was also addressed in a study by Liljeström et al. (2012), which makes the constructive suggestion that the LMB should take urgent action to control water pollution and improve the basin water quality [54]. Additionally, the risk levels continue to increase over time in both cases, posing severe challenges to the long-term development of river basins around the world and necessitating a massive collaborative effort to solve.

A Comparison between the LMB Countries
There are some apparent similarities in the distribution of risks between the four LMB countries ( Figure 5). Compared with other issues, wastewater pollution, the threat to fish and exposure to floods and droughts remain at the highest risk levels. In contrast, environmental and agricultural water stress risks stand at the lowest levels for all countries. Concerning the economic dependence on the Mekong water, we are led to the conclusion that the larger the proportion of the basin country' area in the LMB is, the heavier the country depends on the river water resource economically. For example, Laos and Cambodia, two riparian countries covering the two largest MRB portions at 20% and 27%, respectively, have the highest economic dependence risk points (five points). Interestingly, a noticeable difference in the risks of human water stress can be seen in Laos and Vietnam in relation to their geographical location in the LMB. The most upstream country in the LMB, Laos, has a deficient low-risk level for human-related water use, whereas that of Vietnam, the most downstream country, is relatively high at three risk points.   The future trends in the change of risks to human beings and the LMB riparian countries' environment are clearly illustrated in Figure 6. There is no difference in projections of risk over time in Cambodia, Thailand and Vietnam, except for Laos, which experienced an upward trend in risks to environmental water stress. Compared to the current risk levels described in Figure 4, those of Cambodia, Laos, and Vietnam remain unchanged for about forty years. In contrast, Thailand expects to experience a steady increase in the risk points of both human and environmental water stress, with one point higher in each stress category.

Criteria Weights
C1 (water quantity) with a final weight of 0.1463 was the main criterion of the minor importance to the decision-making process in the LMB because it had the lowest risk point while C3 (ecosystem) and C4 (socioeconomics) stood together at the highest positions ( Table 5).
The sub-criteria local weights followed by their final weights are described adjacent to their corresponding main criteria to illustrate each sub-criterion relative importance to its main criterion. Of the nine sub-criteria, C3.2 (threat to fish) ranked first, which means that fisheries might be the sector facing the most severe risk over the development of water resources in the LMB. Interestingly, although associated with the third-ranked main criteria, C2.2 (wastewater pollution) with a final weight very close to that of C3.2 was the second important criterion. In contrast, C4.2 (societal-well-being), the sub-criterion that belongs to the first-ranked main criterion, was the least important criterion. The two other least significant criteria were C1.1 (human water stress) and C1.2 (agricultural water stress).

Final Rankings of the Scenarios
After determining the criteria weights concerning the MCDA model's goal, the scenarios' prioritisation results against each criterion were calculated and presented in Figure 7 and Figure 8 as examples. As to these two particular criteria (C1 and C2), in general, the top-ranked water development scenario concerning the potential risks to the environment and ecosystem in the LMB was M1-the baseline scenario, followed by M2 and M3. Nonetheless, there was a noticeable difference of the priority order of three scenarios regarding criterion C2.1 (nutrient pollution): Scenario M2 remained the same ranking, whereas M1 and M3 were placed in reverse order. It can be seen that the ranking of a particular scenario according to one criterion might be higher or lower than that according to the other criteria.   The preference levels of each development scenario across all criteria and the results of the final rankings of three main scenarios are summarised in Table 6.
The baseline scenario (M1) was the most preferred, the planned development    Table 6 indicates significant differences between the three scenarios in comparing each scenario's performance across all criteria. Scenario M1 delivered the best performance on criterion C3.2 (threat to fish) and the worst performance on criterion C2.1 (nutrient pollution) (with the performance score of 0.0829 and 0.0227, respectively), which means that in scenario M1, the LMB faced the lowest risk in relation to the threat to fish, and at the same time, it had to deal with the highest risk originated from nutrient pollution. Accordingly, the best and the worst performances of scenario M2 were on criteria C2.2 (wastewater pollution) and C3.1 (ecosystem impacts from dams), and those of scenario M3 were on criteria C3.2 (threat to fish) and C3.1, respectively. Note that the impacts of dams on ecosystems (C3.1), which created the significant risk to the water development of the LMB in 2020 (M2), is surprisingly becoming the minor risk to the river basin's future growth.

MCDA Model Analysis for the LMB Member Countries
The four member countries' MCDA models were developed to analyse and rank the water development scenarios in priority order. Similar to the LMB's MCDA model, the basin country model used the same set of the main criteria and their corresponding sub-criteria except those of main criteria C2 because the two sub-criteria of C2 (nutrient pollution and wastewater pollution) are pollutionrelated issues, which cannot be clearly distinguished by country level. Consequently, its two sub-criteria (C2.1 and C2.2) were not incorporated into the basin country model. The results of criteria weights and the priority order of three water development scenarios are described below.

Criteria Weights
When considering the relative importance of the main criteria to the decision-making process, C1 (Water quantity) was the least favorable criterion to all countries. In contrast, the most favorable one was varied slightly but only switching between C3 (ecosystems) and C4 (socioeconomics) (Figure 9).
The global weights of all sub-criteria by countries were illustrated in Figure   10.

Final Rankings of the Scenarios
The overall weights of the criteria and the final ranking of water development scenarios for the four LMB countries are summarised in Tables 7-10. Surprisingly, the preference order of development scenarios were considerably different from country to country and can be categorised into Group 1 (Laos and

Discussion
As previous sections revealed, the risks concerning water-related issues, ecosystem, environment and human beings are relatively different between the MRB and the basins worldwide, particularly in terms of the threat to fish, exposure to flood and drought and economic dependence on water resources. The findings highlight the risk originated from the threat to fish, which poses a more significant danger to the LMB's future development than the other river basins. This issue was also addressed by MRC (2019) that the growing pressures on fisheries negatively impacts on the environmental condition of the LMB, and immediate actions are needed to manage fisheries better [9]. The high level of risk in fisheries might be explained by the four riparian countries' large fisheries production.
According to MRC (2019), all member countries' fisheries production in the river basin accounted for 45% of their total national fisheries production in 2015, equivalent to approximately 17 billion US dollars [9].
In addition, the risk of floods and droughts in the LMB was noticeably higher than that at the global scale and recognised as a prominent source of risk to the LMB development. Compared with other river basins, strong evidence indicated that severe floods and droughts resulting from climate change substantially threatened the LMB, making it the region of high vulnerability to climate change [55]. The fact that the LMB have higher economic dependence on water resources than the average level of river basins worldwide could result from the large shares in the total basin population of the four riparian countries. As estimated in 2015, more than one-third of approximately 65 million people living in the LMB come from Thailand and Vietnam, while the Cambodian accounts for one fifth and Laos take the rest of the LMB population [9]. This leads to the heavy dependence on the Mekong River water resources for food and livelihoods. Similarly, the economic reliance on river water is also greatly influenced by each country's share in the basin population.
Relating to the projected risk levels, the estimated values of nutrient pollution risk of the LMB were noticeably higher than those of the global average level.
These findings are consistent with a previous study of Maavara et al. (2015) that expressed severe concern about the increase in the projected Phosphorous concentration in dam reservoirs in major transboundary river basins like Mekong and Amazon that would expand surface water eutrophication [56].  In conclusion, the results suggest several important implications for future practice. Firstly, the MRC and other stakeholders in the LMB should give more concern to high risk-level issues such as the threat to fish or damages of floods and droughts during the post-event evaluation and in the early stage of basin planning at both local and river basin level. It would be valuable to consider the trade-offs among monetary benefits, the degradation of the environment and ecosystem, and the harm to human beings in the long-term for the better sustainable development of the LMB. Secondly, the benefit-sharing mechanism in the LMB, which mainly focused on ensuring the equitable distribution of benefits derived from utilising basin water resources, should be revised more effectively by taking the potential risks to each country in the long-term as constraints. Therefore, the member states have another incentive to establish a complete benefit-sharing scheme covering the river basin development aspects. In the future, a significant challenge facing MRC will be to improve its policy measures and adequately invest in appropriate infrastructure to mitigate the negative impacts of high-risk issues.