ABSTRACT
Drought is a natural phenomenon that poses a significant threat to water resources in affected regions. The detrimental effects of this extreme weather event, regardless of its type, have had an impact, necessitating concrete resource allocation projects to mitigate drought. From this perspective, our research proposes a group multicriteria sorting model to support resource allocation for drought mitigation. A multicriteria sorting model was developed based on the PROMSORT method to assist the government in evaluating 14 municipalities of the Apodi-Mossoró river basin in the Rio Grande do Norte State in Brazil. This model classifies the municipalities into high, moderate, and low-priority categories, enabling targeted attention and allocation of resources for drought mitigation efforts. The research findings demonstrate that the proposed model can effectively support strategic public policies, allocate resources, and facilitate the implementation of appropriate actions, thereby focusing efforts on the cities most severely affected by drought and alleviating the adverse effects of this natural disaster
HIGHLIGHTS
We identified and sorted the cities most impacted by drought through group decision-making.
Structured approach for allocating public resources.
We facilitate introduction of strategic public policies.
We consider the particularities of municipalities in the Brazilian semi-arid region.
We propose decision rules to generate a group decision.
INTRODUCTION
Drought is a natural phenomenon with large negative deviations in long-term climate precipitation and mainly harms the affected region's water resources. As a reflection of climatic conditions, the hydrography of regions affected by drought is fragile, in its broad aspects, being insufficient to sustain large rivers that remain perennial in long periods of precipitation absence (Kaletova et al., 2021). Therefore, these regions that present this natural disaster that compromises water and energy security and subsistence agriculture need concrete projects to allocate resources to mitigate drought (Marengo et al., 2018).
Smakhtin & Schipper (2008) define drought as a seasonal and recurrent event caused by long-term climatic conditions like low precipitation and high evaporation. The scarcity of rainfall leads to the depletion of water reservoirs, thereby rendering water resources unable to meet the necessary demands. Thus, water scarcity occurs, negatively impacting the economy and society.
Depending on their intensity, frequency, and duration, drought events can be sorted into four types: (i) meteorological drought, which occurs in situations of a scarcity of rain and can lead to other types of droughts; (ii) hydrological drought, which is the lack of water in the hydrological system and can be determined by the reduction of average water levels in the affected region; (iii) agricultural drought results from rainfall shortages and soil water deficits in agriculture; and (iv) the socioeconomic drought, which is the junction of the three previously mentioned droughts, relating to offer and demand by something (Karamouz et al., 2015; Bae et al., 2019). However, the damage caused by this extreme weather event, regardless of type, has impacted many regions worldwide. Thus, these regions need concrete resource allocation projects to mitigate drought.
The daily search for water resources and the growing emissions of waste from society are directly related to the expansion of urban areas and population growth. As a result, consuming these resources leads to severe global water scarcity, further constraining socioeconomic development and threatening human and ecological health (Dong et al., 2016). As a result, the search for more efficient water management becomes increasingly constant due to legal rights and economic interests that must be considered during the decision-making process to mitigate drought.
Decision-making related to managing water resources is complex. Nevertheless, it has been increasingly aggravated in places that face prolonged droughts, because these decisions need to consider multiple objectives involving environmental, social, and economic impacts (Gonçalo & Morais, 2018). In Brazil, these decisions are made by specific committees that inform the National Water Resources Policy (Ministry of Environment) and are supported by simple majority voting. While simple majority voting offers advantages in terms of simplicity and efficiency, it also introduces a high level of subjectivity and susceptibility to external influences (Silva et al., 2010; De Almeida et al., 2019). Conversely, multicriteria techniques are capable of supporting the decision-making process by considering the intricacies and complexities inherent in complex problems like drought (Abdullah et al., 2021).
Droughts are typically managed reactively, addressing their impacts after a water crisis happens, unlike other disasters (Pischke & Stefanski, 2016; Roshni et al., 2022; Omari-Motsumi et al., 2023). However, a reactive approach needs to be reviewed and better coordinated. Applying a collaborative effort is crucial for effectively managing droughts and anticipating weather events (Mera, 2018). Researching how decision processes in drought management can be modeled and implemented is essential. For this purpose, the multicriteria decision-support techniques can serve as a timely ally.
According to Zeleny (1982), the multicriteria decision-support approach is a tool capable of solving the dilemma of multiple conflicting objectives based on the value judgments of one or more decision-makers, aiming to clarify a decision. Modeling existing subjectivity in decision-making processes through the multicriteria decision-support methodology allows contemplating different perspectives on the context, enabling better decision-making (Wu et al., 2016). Therefore, multicriteria methods can be applied to different contexts, mainly in water resources policy and management (Costa e Silva et al., 2020).
Alamanos et al. (2018) claimed that the multicriteria approach has been successfully implemented in water resources management, highlighting those involving participatory approaches. The main applications are related to water scarcity problems (Zhang et al., 2017; Uen et al., 2018), water supply policy and planning (Silva et al., 2010; Sikder & Salehin, 2015; Rivero et al., 2020), drought vulnerability mapping (Paegelow et al., 2021; Sivakumar et al., 2021) and in the water losses and management (Morais et al., 2014; Trojan & Morais, 2015; Medeiros et al., 2017). Efforts to develop methodologies that provide support for decision-making are still underway.
Gonçalo & Morais (2018) presented a multicriteria model of collective decisions to rank the municipalities (cities) devastated by the drought and, thus, guide public management efforts to mitigate the effects of drought in northeastern Brazil. Nevertheless, their approach deals with a multicriteria problem of ranking alternatives and not sorting possible critical zones for resource allocation.
Kim et al. (2014) developed an iterative framework to prioritize suitable locations for robust water resource allocation considering various climate change scenarios. Similarly, Sobhani et al. (2020) managed to model, analyze, and predict the drought in Iran to prioritize the affected areas through the TOPSIS methodology (Technique for Order Preference by Similarity to Ideal Solution), a decision-support approach developed by Hwang & Yoon (1981), combined with fuzzy logic indices. Despite significant contributions, the socioeconomic and population parameters were not considered in their evaluations. Our approach will incorporate these indicators.
The literature review by Hajkowicz & Collins (2006) and supplemented by Gebre et al. (2021) on the multicriteria approach to water resource planning, management, and allocation found no conclusive evidence favoring one method over another. However, only some of these models deal with the multicriteria group sorting, which could assess the criticality of water resources in each municipality, considering different individual perspectives and multiple objectives. This identification is particularly valuable, considering the complexity associated with drought management.
Studies on water resources policies and management present an important gap associated with resource allocation to mitigate drought. A complex task lies in defining the municipalities (cities) that could receive resources to face the drought, as this decision generally requires consideration of a group of decision-makers' value judgments and multiple perspectives. The current research aims to solve this gap through a structured methodology capable of sorting municipalities (cities) in a critical state for resource allocation to mitigate drought.
Our study takes into account the intrinsic characteristics and socioeconomic values directly linked to water resource scarcity. By sorting municipalities in advance, we aim to enhance resource allocation in areas severely affected by drought, ensuring that resources are allocated based on their actual needs. The primary objective of this study is to devise an approach that can proficiently manage this decision-making process.
For numerical validation, the study applied the multicriteria sorting methodology called PROMSORT, developed by Araz & Ozkarahan (2007), to verify, evaluate, and validate the different possible results of the proposed model and suggest the best results for the sorting of municipalities (cities) in allocating resources to mitigate drought. This article is organized into seven sections: (i) introduction, (ii) definitions of the PROMSORT method, (iii) model proposed, (iv) case study description, (v) results and discussions, (vi) managerial implications, and (vii) conclusions.
PROMSORT METHOD




The alternatives attribution/allocation in the categories is carried out from the outranking relationships of the alternatives with the limiting/reference profiles (), where:
The alternative a is compared with
to i=k, k−1,…,1
is the first limiting profile such that
is the first limiting profile such that
or aI
If h>t, then, a will go to category
Otherwise, the alternative must not be assigned to any category.





We propose a model that will incorporate the multicriteria sorting logic through the method described above, combined with some proposed innovations for group decision-making, to generate results for the decision problem. The next section presents the general model steps and their contributions.
PROPOSED GROUP SORTING MODEL
General steps of group sorting approach to resource allocation in drought mitigation.
General steps of group sorting approach to resource allocation in drought mitigation.
This model consists of three sequential steps. In general, the preferences of each identified decision-maker will be elicited individually, resulting in an individual sorting of the municipalities according to their priority degree. The final result will be represented by the aggregation of individual results using group decision procedures.
In the first step, we determine the decision-maker's number (N) to consult during the development of the study. The correct decision-makers' identification in projects involving water resources is crucial. It helps align their values, conditions, and preferences, leading to increased engagement, motivation, and better water resource management (Hargrove & Heyman, 2020). Furthermore, decisions involving catastrophic events such as droughts require diverse opinions and perspectives to formulate appropriate solutions.
In the second step, the preference elicitation process takes place. The chosen decision-makers are invited to evaluate the municipalities (alternatives) that primarily need the resource allocation in drought mitigation. To this, each decision-maker must establish the evaluation criteria () according to their preferences, which may be similar or different in each evaluation. Additionally, the alternatives evaluated must be common to all decision-makers involved and must be municipalities affected by the drought.
Subsequently, information about the value of the criteria weights will be incorporated, representing the criteria importance degree. The weight definition will occur through the surrogate weight procedure developed by Edwards & Barron (1994), called Ranking Ordered Centroid (ROC), in which it manages to represent the probable interpretation of the preferences expressed by a decision-maker, without establishing exact information (Danielson & Ekenberg, 2017).

Consequently, in the second step, each decision-maker must establish the limiting profiles (reference) between the categories to delimit the scope of each sorting priority assigned to each municipality. Do Carmo et al. (2021) recommend the limiting profile's definition in a joint agreement with the decision-makers or, if not possible, by the study's researchers, using intermediate numerical values of the scale. Additionally, it is important to note that the model allows the definition of the category and, consequently,
limiting profiles between categories that depend on the study problem and expected results. By definition, we have that, regardless of the number of defined categories, the
category is the most important than the
category, and so on until the
category is less important. The proposed sorting is shown in Table 1.
Municipality priorities sorting.
Category . | Description . |
---|---|
![]() | Municipalities with the highest priority in resource allocation to mitigate drought. |
![]() | Municipalities with the moderate priority in resource allocation to mitigate drought. |
… | … |
![]() | Municipalities with the lower priority in resource allocation to mitigate drought. |
Category . | Description . |
---|---|
![]() | Municipalities with the highest priority in resource allocation to mitigate drought. |
![]() | Municipalities with the moderate priority in resource allocation to mitigate drought. |
… | … |
![]() | Municipalities with the lower priority in resource allocation to mitigate drought. |
By utilizing the information obtained and parameters established by each decision-maker, it is possible to generate individual sorting of municipalities that require resource allocation to mitigate drought through the PROMSORT method.
In the third step, the individual's sorting obtained by the method applied may be aggregated by decision rules (attribution), allowing final sorts (group) to be generated and analyzed by decision-makers to establish a set of priority municipalities for resource allocation. Therefore, three (3) decision rules were defined and applied to unanimity or non-unanimity cases in the individual sorting obtained in the application of each method, as shown in Table 2.
Decision rules for group sorting.
Sorting type . | Decision rules . | Description . |
---|---|---|
Unanimity | First Decision Rule | Alternative ![]() ![]() ![]() ![]() |
Non-unanimity | Second Decision Rule | Alternative ![]() ![]() ![]() ![]() |
Third Decision Rule | The alternative ![]() ![]() |
Sorting type . | Decision rules . | Description . |
---|---|---|
Unanimity | First Decision Rule | Alternative ![]() ![]() ![]() ![]() |
Non-unanimity | Second Decision Rule | Alternative ![]() ![]() ![]() ![]() |
Third Decision Rule | The alternative ![]() ![]() |
Finally, a sensitivity analysis must be conducted to assess the stability of individual sorting for municipalities regarding their priority in resource allocation to mitigate drought. Thus, it is practicable to analyze the possible assessment inconsistencies of the municipalities in the criteria weights. Each criteria importance attributed by each decision-maker was increased and reduced by 20% and will be analyzed for possible new municipality sorting.
CASE STUDY DESCRIPTION
Historically, the Brazilian Northeast region has been significantly impacted by drought, particularly in its semi-arid areas. Factors such as low average annual rainfall (below 800 millimeters), aridity index of up to 0.5, and drought risk exceeding 60% contribute to classifying the Brazilian Northeast as a drought-prone area (Marengo & Bernasconi, 2015).
One sub-region of northeastern Brazil is the semi-arid, which is the most populous drought-affected region in the world, with approximately twenty-eight million inhabitants, equal to 13% Brazilian population (Alvalá et al., 2019; INSA, 2022). Habitually, the Brazilian semi-arid region presents irregular rainfall distribution and shallow and rocky soils, which give it a low water storage capacity, a small drainage system, and a high evapotranspiration rate, complicating water resources management.
Brazilian semi-arid municipalities are affected by drought, such that the main economic activity of the rural zone population is characterized by extensive livestock and agricultural subsistence with low technological investment and low productivity (Marengo, 2008). Furthermore, the Brazilian semi-arid presents a social context historically weakened by the drought, indisposing a minimum life quality. Second the Brazilian Institute of Geography and Statistics (IBGE, 2017), the Northeast region, where the semi-arid region predominates, presents the most critical situation in the country regarding food security, and 49.7% of the population shows signs of food restriction in their daily diet.
Conversely, data from the National Water and Basic Sanitation Agency (ANA, 2021), for September 2021 report that several states belonging to the Brazilian semi-arid region are in a drought situation. It stands out that the Rio Grande do Norte State presents 100% of its territory in a widespread drought situation, with 65% of its territorial part already characterized as a severe drought, especially in the State's western region, due to short- and long-term negative precipitation anomalies. Therefore, prioritizing cities in allocating resources to mitigate drought needs to be done.
The selection of these municipalities as potential model alternatives is primarily motivated by their location in the semi-arid area of Brazil (geographical location). Additionally, the presence of a river basin, which is shared by all alternatives under consideration, and the presence of municipalities with distinct features will create a wealth of relevant information for decision-making.
RESULTS AND DISCUSSION
The initial methodological step requires the decision-makers to specify who will be consulted during the study's progress. Representatives of the State Secretary for the Environment and Water Resources (SEMARH) and the Water Management Institute (IGARN) of the Rio Grande do Norte State, Brazil, were selected as government representatives in the municipalities sorting process. They were chosen according to (i) the political influence level over the region in which they are located and (ii) the knowledge level about the problems faced by the region affected by drought. Table 3 presents the characteristics of the three decision-makers.
Decision-makers chosen to act in the municipalities sorting process that need the resource allocation in drought mitigation.
Decision-maker . | Description . |
---|---|
Decision-maker 1 | Water Resources Planning and Management Coordinator at SEMARH |
Decision-maker 2 | Environmental and Sanitation Coordinator at SEMARH |
Decision-maker 3 | Operations Management Coordinator at IGARN |
Decision-maker . | Description . |
---|---|
Decision-maker 1 | Water Resources Planning and Management Coordinator at SEMARH |
Decision-maker 2 | Environmental and Sanitation Coordinator at SEMARH |
Decision-maker 3 | Operations Management Coordinator at IGARN |
The first decision-maker was the water resources planning and management coordinator SEMARH. Its specific attributions are related to political and strategic assessment for water management, including developing actions to minimize/mitigate drought events. The second decision-maker identified was the environment and sanitation coordinator at SEMARH. Among their responsibilities, establishing activities and projects associated with the use and preservation of the environment and water resources stand out. Moreover, its activities include monitoring and restoration of degraded regions, trash management, and environmental education programs. Lastly, the third decision-maker considered is the operations management coordinator at IGARN. Additionally, it aims to coordinate and implement water resources management activities in the state, as well as perform research to set criteria and standards for the authorization and sensible use of water resources.
The second methodological step requires decision-makers to evaluate the municipality's performance and the parameters of the multicriteria methods. Initially, the decision-maker's objectives were identified, described, and clarified. According to the decision-makers group, applying the proposed model to that context will allow the development of public policies and action plans to mitigate drought more efficiently. Thus, three significant objectives were identified that should be considered as a means of prioritizing municipalities, as noted below:
Hydrological Objective: aims to measure and maximize the water impacts of the drought-affected region in order to identify the municipality's priority level in the resource allocation process;
Sanitary Objective: aims to identify and maximize the sanitary impacts in the municipalities considered in the process for resource allocation to mitigate drought;
Socioeconomic Objective: aims to identify and maximize the socioeconomic impacts of municipalities affected by drought, seeking to prioritize these in the resource allocation process.
It's important to note that decision-makers did not consider the environmental objectives (indicators) in the evaluation because of the caatinga vegetation prevailing in the semi-arid region of the Brazilian Northeast. Caatinga is xerophytic vegetation (Santos & Santos, 2008) comprising plants adapted to semi-arid and desert climates, with most plants shedding their leaves during water scarcity (Santos et al., 2010). The caatinga ecosystem has developed natural resilience mechanisms to water scarcity, decreasing the importance of ecological aspects in this study. Furthermore, we addressed the political objectives subjectively by incorporating the opinions of decision-makers, who are public political representatives. Hence, it is pertinent to focus on more immediate aspects, such as the socioeconomic vulnerability of communities, the lack of potable water, and health security.
To the study, we identified and presented relevant criteria for evaluation to the decision-makers based on a literature review that similarly addressed the topic (Hajkowicz & Collins, 2006; Alamanos et al., 2018; Gonçalo & Morais, 2018). In total, 10 criteria were pre-selected: (i) reservoir situation, (ii) total water service index, (iii) total sewage service index, (iv) population, (v) human development index, (vi) number of education institutes, (vii) hospitals, (viii) number of emergency health units and basic health units, (ix) gross domestic product (GDP) of municipalities, and (x) average amount made available to beneficiaries of social programs. Each decision-maker then had the flexibility to choose specific criteria for their evaluations. It is essential to note that the pre-selected criteria are not exhaustive, and decision-makers have the option to include additional criteria if deemed necessary, although no participant suggested the inclusion of criteria beyond the pre-selected list. Furthermore, this is an ad hoc decision, meaning that if the decision-makers change, the criteria may also change. Table 4 shows the evaluation criteria defined for each decision-maker.
Criteria Evaluation.
Objective . | Criteria . | Code . | Description . | Type . | Scale . | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . |
---|---|---|---|---|---|---|---|---|
Hydrological | Reservoir Situation | Cr1 | It seeks to determine the current level of the water reservoirs by each Municipality. | Minimization | Reservoir Level Percentage (%) | ✓ | ✓ | ✓ |
Total water service index | Cr2 | It seeks to determine the municipality's supply condition | Minimization | Percentage (%) of the municipality's total population served with water supply. | ✓ | ✓ | ✓ | |
Sanitary | Total sewage service index | Cr3 | It seeks to determine the sanitary sewage condition of the municipality population. | Minimization | Percentage (%) of the Municipality's total population served with sanitary sewage. | ✓ | ✓ | ✓ |
Socioeconomic | Population | Cr4 | Inhabitants Number residing in the Municipality. | Maximization | Inhabitants Number | ✓ | ✗ | ✓ |
Human Development Index (HDI) | Cr5 | The Human Development Index (HDI) is a municipality's average measure of basic human development achievements. | Minimization | Numerical scale between 0 and 1 | ✗ | ✗ | ✓ |
Objective . | Criteria . | Code . | Description . | Type . | Scale . | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . |
---|---|---|---|---|---|---|---|---|
Hydrological | Reservoir Situation | Cr1 | It seeks to determine the current level of the water reservoirs by each Municipality. | Minimization | Reservoir Level Percentage (%) | ✓ | ✓ | ✓ |
Total water service index | Cr2 | It seeks to determine the municipality's supply condition | Minimization | Percentage (%) of the municipality's total population served with water supply. | ✓ | ✓ | ✓ | |
Sanitary | Total sewage service index | Cr3 | It seeks to determine the sanitary sewage condition of the municipality population. | Minimization | Percentage (%) of the Municipality's total population served with sanitary sewage. | ✓ | ✓ | ✓ |
Socioeconomic | Population | Cr4 | Inhabitants Number residing in the Municipality. | Maximization | Inhabitants Number | ✓ | ✗ | ✓ |
Human Development Index (HDI) | Cr5 | The Human Development Index (HDI) is a municipality's average measure of basic human development achievements. | Minimization | Numerical scale between 0 and 1 | ✗ | ✗ | ✓ |
Note: The acceptable symbol (✓) indicates that the criterion belongs to the evaluation matrix of the respective decision-maker. The unacceptable symbol (✗) indicates that the criterion does not belong to the evaluation matrix of the respective decision-maker.
Five criteria have been selected and categorized into three significant constructs: (i) hydrological, (ii) sanitary, and (iii) socioeconomic. The first objective includes the ‘Reservoir Situation’ criteria, which indicates the current level of the water supply network in each municipality, and the ‘Total water service index’ which determines the percentage of the total population with water supply. All decision-makers involved considered both criteria in the assessments. The second objective is represented by the ‘Total sewage service index’ criteria, indicating the percentage of the municipality's total population served with sanitary sewage and considered by all decision-makers. The third objective includes the ‘Population’ criterion, which indicates the number of inhabitants residing in the municipality and is considered in evaluating the first and third decision-makers. Finally, this objective considers the HDI criterion, which indicates the average basic human development achievements and is only considered by the third decision-maker. The HDI covers Education, Health, and Income, the latter also being an economic indicator, which considers the gross income of the municipality (i.e., the total value of its financial resources and economy).
The next step, still in the second methodological stage, will be to build the matrix to evaluate the drought-affected municipalities based on the previously defined criteria by each decision-maker. Through the database provided by the (IBGE, 2021), the National Sanitation Information System (SNIS, 2021), and the State Secretary for the Environment and Water Resources (SEMARH, 2022), Table 5 presents the individual assessments.
Individual assessment matrix.
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Cr1 . | Cr2 . | Cr3 . | Cr4 . | Cr1 . | Cr2 . | Cr3 . | Cr1 . | Cr2 . | Cr3 . | Cr4 . | Cr5 . |
A1 | 13.27% | 62.29% | 28.8% | 10,923 | 13.27% | 62.29% | 28.8% | 13.27% | 62.29% | 28.8% | 10,923 | 0.529 |
A2 | 25.29% | 94.08% | 23.7% | 3,869 | 25.29% | 94.08% | 23.7% | 25.29% | 94.08% | 23.7% | 3,869 | 0.646 |
A3 | 30.42% | 100.00% | 22.73% | 30,600 | 30.62% | 100.00% | 22.73% | 30.62% | 100.00% | 22.73% | 30,600 | 0.614 |
A4 | 30.92% | 67.36% | 16.6% | 9,670 | 30.92% | 67.36% | 16.6% | 30.92% | 67.36% | 16.6% | 9,670 | 0.678 |
A5 | 65.33% | 97.89% | 4.7% | 10,485 | 65.33% | 97.89% | 4.7% | 65.33% | 97.89% | 4.7% | 10,485 | 0.621 |
A6 | 70.92% | 83.22% | 12.7% | 20,541 | 70.92% | 83.22% | 12.7% | 70.92% | 83.22% | 12.7% | 20,541 | 0.618 |
A7 | 35.93% | 65.61% | 62.76% | 4,025 | 35.93% | 65.61% | 62.76% | 35.93% | 65.61% | 62.76% | 4,025 | 0.638 |
A8 | 14.79% | 100.00% | 10.1% | 1,743 | 14.79% | 100.00% | 10.1% | 14.79% | 100.00% | 10.1% | 1,743 | 0.604 |
A9 | 72.24% | 92.17% | 16.3% | 5,697 | 72.24% | 92.17% | 16.3% | 72.24% | 92.17% | 16.3% | 5,697 | 0.629 |
A10 | 79.67% | 78,00% | 7.05% | 5,941 | 79.67% | 78,00% | 7.05% | 79.67% | 78,00% | 7.05% | 5,941 | 0.608 |
A11 | 74.67% | 80.4% | 30.2% | 8,325 | 74.67% | 80.4% | 30.2% | 74.67% | 80.4% | 30.2% | 8,325 | 0.609 |
A12 | 75.55% | 26.64% | 19.5% | 5,128 | 75.55% | 26.64% | 19.5% | 75.55% | 26.64% | 19.5% | 5,128 | 0.608 |
A13 | 70.82% | 55.7% | 57.9% | 3,614 | 70.82% | 55.7% | 57.9% | 70.82% | 55.7% | 57.9% | 3,614 | 0.584 |
A14 | 68.74% | 100.00% | 22.3% | 4,467 | 68.74% | 100.00% | 22.3% | 68.74% | 100.00% | 22.3% | 4,467 | 0.604 |
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
. | Cr1 . | Cr2 . | Cr3 . | Cr4 . | Cr1 . | Cr2 . | Cr3 . | Cr1 . | Cr2 . | Cr3 . | Cr4 . | Cr5 . |
A1 | 13.27% | 62.29% | 28.8% | 10,923 | 13.27% | 62.29% | 28.8% | 13.27% | 62.29% | 28.8% | 10,923 | 0.529 |
A2 | 25.29% | 94.08% | 23.7% | 3,869 | 25.29% | 94.08% | 23.7% | 25.29% | 94.08% | 23.7% | 3,869 | 0.646 |
A3 | 30.42% | 100.00% | 22.73% | 30,600 | 30.62% | 100.00% | 22.73% | 30.62% | 100.00% | 22.73% | 30,600 | 0.614 |
A4 | 30.92% | 67.36% | 16.6% | 9,670 | 30.92% | 67.36% | 16.6% | 30.92% | 67.36% | 16.6% | 9,670 | 0.678 |
A5 | 65.33% | 97.89% | 4.7% | 10,485 | 65.33% | 97.89% | 4.7% | 65.33% | 97.89% | 4.7% | 10,485 | 0.621 |
A6 | 70.92% | 83.22% | 12.7% | 20,541 | 70.92% | 83.22% | 12.7% | 70.92% | 83.22% | 12.7% | 20,541 | 0.618 |
A7 | 35.93% | 65.61% | 62.76% | 4,025 | 35.93% | 65.61% | 62.76% | 35.93% | 65.61% | 62.76% | 4,025 | 0.638 |
A8 | 14.79% | 100.00% | 10.1% | 1,743 | 14.79% | 100.00% | 10.1% | 14.79% | 100.00% | 10.1% | 1,743 | 0.604 |
A9 | 72.24% | 92.17% | 16.3% | 5,697 | 72.24% | 92.17% | 16.3% | 72.24% | 92.17% | 16.3% | 5,697 | 0.629 |
A10 | 79.67% | 78,00% | 7.05% | 5,941 | 79.67% | 78,00% | 7.05% | 79.67% | 78,00% | 7.05% | 5,941 | 0.608 |
A11 | 74.67% | 80.4% | 30.2% | 8,325 | 74.67% | 80.4% | 30.2% | 74.67% | 80.4% | 30.2% | 8,325 | 0.609 |
A12 | 75.55% | 26.64% | 19.5% | 5,128 | 75.55% | 26.64% | 19.5% | 75.55% | 26.64% | 19.5% | 5,128 | 0.608 |
A13 | 70.82% | 55.7% | 57.9% | 3,614 | 70.82% | 55.7% | 57.9% | 70.82% | 55.7% | 57.9% | 3,614 | 0.584 |
A14 | 68.74% | 100.00% | 22.3% | 4,467 | 68.74% | 100.00% | 22.3% | 68.74% | 100.00% | 22.3% | 4,467 | 0.604 |
A1: Tenente Ananias; A2: Pilões; A3: Pau dos Ferros; A4: Campo Grande; A5: Umarizal; A6: Caraúbas; A7: Lucrécia; A8 – Severiano Melo; A9: Encanto; A10: José da Penha; A11: Marcelino Vieira; A12: Rafael Fernandes; A13: Riacho da Cruz; A14: Rodolfo Fernandes.
Cr1 – Reservoir Situation; Cr2 – Total water service index; Cr3 – Total sewage service index; Cr4 – Population; Cr5 – HDI.
To the municipality priority, it is necessary to define the categories that will be considered in the study. These categories indicate a set of municipalities that urgently need resource allocation in drought mitigation. We have defined three categories: C1 – High Priority; C2 – Moderate Priority; e C3 – Low Priority. The category's descriptions are in Table 6.
Priority categories/classifications for the resource allocation process in drought mitigation.
Category/Class . | Description . |
---|---|
C1 – High Priority | Municipalities sorted in this category will have high urgency/priority in resource allocation to mitigate drought. |
C2 – Moderate Priority | Municipalities sorted in this category will have moderate urgency/priority in resource allocation to mitigate drought. |
C3 – Low Priority | Municipalities sorted in this category will have lower urgency/priority in resource allocation to mitigate drought. |
Category/Class . | Description . |
---|---|
C1 – High Priority | Municipalities sorted in this category will have high urgency/priority in resource allocation to mitigate drought. |
C2 – Moderate Priority | Municipalities sorted in this category will have moderate urgency/priority in resource allocation to mitigate drought. |
C3 – Low Priority | Municipalities sorted in this category will have lower urgency/priority in resource allocation to mitigate drought. |
Consequently, the values about the limiting profiles between the categories and the evaluation type for each criterion are requested by the PROMSORT method. The decision-makers defined the evaluation type as Usual (Type I) for all criteria. This evaluation type indicates that any difference in performance between alternative pairs will represent a strict preference (Brans & Mareschal, 2005). Additionally, each decision-maker established the values referring to the limiting profiles between categories for each criterion, delimiting the sorting priority in numerical values. This information is extremely important to the model, given that the incorrect definition of these values may lead to inefficient results. The limiting profile for each decision-maker is in Table 7.
Limiting Profile/Reference between Categories.
. | Limiting Profile/Reference between Categories . | Cr1 . | Cr2 . | Cr3 . | Cr4 . | Cr5 . |
---|---|---|---|---|---|---|
Decision-maker 1 | Limiting Profile 1 (![]() | 30% | 40% | 25% | 10000 | ✗ |
Limiting Profile 2 (![]() | 60% | 80% | 50% | 6000 | ✗ | |
Decision-maker 2 | Limiting Profile 1 (![]() | 30% | 51% | 26% | ✗ | ✗ |
Limiting Profile 2 (![]() | 60% | 86% | 52% | ✗ | ✗ | |
Decision-maker 3 | Limiting Profile 1 (![]() | 40% | 30% | 30% | 10000 | 0,300 |
Limiting Profile 2 (![]() | 60% | 70% | 60% | 3000 | 0,600 |
. | Limiting Profile/Reference between Categories . | Cr1 . | Cr2 . | Cr3 . | Cr4 . | Cr5 . |
---|---|---|---|---|---|---|
Decision-maker 1 | Limiting Profile 1 (![]() | 30% | 40% | 25% | 10000 | ✗ |
Limiting Profile 2 (![]() | 60% | 80% | 50% | 6000 | ✗ | |
Decision-maker 2 | Limiting Profile 1 (![]() | 30% | 51% | 26% | ✗ | ✗ |
Limiting Profile 2 (![]() | 60% | 86% | 52% | ✗ | ✗ | |
Decision-maker 3 | Limiting Profile 1 (![]() | 40% | 30% | 30% | 10000 | 0,300 |
Limiting Profile 2 (![]() | 60% | 70% | 60% | 3000 | 0,600 |
Cr1 – Reservoir Situation; Cr2 – Total water service index; Cr3 – Total sewage service index; Cr4 – Population; Cr5 – HDI.
Note: The unacceptable symbol (✗) indicates that the criterion does not belong to the evaluation matrix of the respective decision-maker.
The next step is establishing the criteria priority order so that the proposed model can compute the representative weights by each decision-maker. This step will be conducted by the ROC procedure (Equation (10)), representing the criteria importance degree. The priority order and criteria weights for each decision-maker are shown in Table 8.
Order and weights criteria.
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | |||
---|---|---|---|---|---|---|
Ranking/Order . | Criteria . | Weight . | Criteria . | Weight . | Criteria . | Weight . |
1° | Cr4 | 0.5208 | Cr1 | 0.6111 | Cr2 | 0.4567 |
2° | Cr1 | 0.2708 | Cr2 | 0.2778 | Cr1 | 0.2567 |
3° | Cr2 | 0.1458 | Cr3 | 0.1111 | Cr4 | 0.1567 |
4° | Cr3 | 0.0625 | ✗ | ✗ | Cr3 | 0.09 |
5° | ✗ | ✗ | ✗ | ✗ | Cr5 | 0.04 |
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | |||
---|---|---|---|---|---|---|
Ranking/Order . | Criteria . | Weight . | Criteria . | Weight . | Criteria . | Weight . |
1° | Cr4 | 0.5208 | Cr1 | 0.6111 | Cr2 | 0.4567 |
2° | Cr1 | 0.2708 | Cr2 | 0.2778 | Cr1 | 0.2567 |
3° | Cr2 | 0.1458 | Cr3 | 0.1111 | Cr4 | 0.1567 |
4° | Cr3 | 0.0625 | ✗ | ✗ | Cr3 | 0.09 |
5° | ✗ | ✗ | ✗ | ✗ | Cr5 | 0.04 |
Cr1 – Reservoir Situation; Cr2 – Total water service index; Cr3 – Total sewage service index; Cr4 – Population; Cr5 – HDI.
Note: The unacceptable symbol (✗) indicates that the criterion does not belong to the evaluation matrix of the respective decision-maker.
The priority order and weight for the first decision-maker is Population (Cr4) weighted at 0.5208, followed by Reservoir Situation (Cr1) at 0.2708, Total Water Service Index (Cr2) at 0.1458, and Total Sewage Service Index (Cr3) at 0.0625. For the second decision-maker, the priority is Reservoir Situation (Cr1) weighted at 0.6111, Total Water Service Index (Cr2) at 0.2778, and Total Sewage Service Index (Cr3) at 0.1111. Meanwhile, the third decision-maker prioritizes Total Water Service Index (Cr2) weighted at 0.4567, Reservoir Situation (Cr1) at 0.2567, Population (Cr4) at 0.2567, Total Sewage Service Index (Cr3) at 0.0900, and HDI (Cr5) at 0.0400.
All the information and parameters previously defined were applied in the proposed model to generate the Positive Flow(ϕ+), Negative Flow(ϕ-), and Net Flow (ϕ) for the PROMSORT method, which is exposed in Table 9.
Alternatives performance through net, positive and negative flows for each decision-maker.
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | ϕ . | ϕ+ . | ϕ- . | ϕ . | ϕ+ . | ϕ- . | ϕ . | ϕ+ . | ϕ- . |
A1 | 0.7133 | 0.8567 | 0.1433 | 0.7267 | 0.8633 | 0.1367 | 0.5828 | 0.7914 | 0.2086 |
A2 | −0.2653 | 0.3673 | 0.6327 | 0.2947 | 0.6473 | 0.3527 | −0.1287 | 0.4356 | 0.5644 |
A3 | 0.53 | 0.746 | 0.216 | 0.0753 | 0.4987 | 0.4233 | −0.0482 | 0.4455 | 0.4937 |
A4 | 0.3333 | 0.6667 | 0.3333 | 0.3333 | 0.6667 | 0.3333 | 0.3729 | 0.6865 | 0.3135 |
A5 | 0.264 | 0.632 | 0.368 | −0.0987 | 0.4507 | 0.5493 | −0.0931 | 0.4535 | 0.5465 |
A6 | 0.3307 | 0.6653 | 0.3347 | −0.2373 | 0.3813 | 0.6187 | 0.0086 | 0.5043 | 0.4957 |
A7 | −0.248 | 0.376 | 0.624 | 0.1427 | 0.5713 | 0.4287 | 0.1696 | 0.5848 | 0.4152 |
A8 | −0.372 | 0.304 | 0.676 | 0.3667 | 0.6647 | 0.298 | −0.2805 | 0.3281 | 0.6086 |
A9 | −0.288 | 0.356 | 0.644 | −0.408 | 0.296 | 0.704 | −0.2515 | 0.3743 | 0.6257 |
A10 | −0.2227 | 0.3887 | 0.6113 | −0.4587 | 0.2707 | 0.7293 | −0.1446 | 0.4264 | 0.571 |
A11 | −0.14 | 0.43 | 0.57 | −0.4947 | 0.2527 | 0.7473 | −0.2383 | 0.3809 | 0.6191 |
A12 | −0.2453 | 0.3773 | 0.6227 | −0.2267 | 0.3867 | 0.6133 | 0.2132 | 0.6053 | 0.3921 |
A13 | −0.4827 | 0.2587 | 0.7413 | −0.0933 | 0.4533 | 0.5467 | 0.0376 | 0.5188 | 0.4812 |
A14 | −0.4227 | 0.2787 | 0.7013 | −0.3573 | 0.3027 | 0.66 | −0.5089 | 0.2139 | 0.7228 |
r1 | 0.4967 | 0.7393 | 0.2427 | 0.5313 | 0.7453 | 0.214 | 0.4389 | 0.7195 | 0.2805 |
r2 | 0.0187 | 0.5093 | 0.4907 | −0.096 | 0.452 | 0.548 | −0.13 | 0.435 | 0.565 |
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | ϕ . | ϕ+ . | ϕ- . | ϕ . | ϕ+ . | ϕ- . | ϕ . | ϕ+ . | ϕ- . |
A1 | 0.7133 | 0.8567 | 0.1433 | 0.7267 | 0.8633 | 0.1367 | 0.5828 | 0.7914 | 0.2086 |
A2 | −0.2653 | 0.3673 | 0.6327 | 0.2947 | 0.6473 | 0.3527 | −0.1287 | 0.4356 | 0.5644 |
A3 | 0.53 | 0.746 | 0.216 | 0.0753 | 0.4987 | 0.4233 | −0.0482 | 0.4455 | 0.4937 |
A4 | 0.3333 | 0.6667 | 0.3333 | 0.3333 | 0.6667 | 0.3333 | 0.3729 | 0.6865 | 0.3135 |
A5 | 0.264 | 0.632 | 0.368 | −0.0987 | 0.4507 | 0.5493 | −0.0931 | 0.4535 | 0.5465 |
A6 | 0.3307 | 0.6653 | 0.3347 | −0.2373 | 0.3813 | 0.6187 | 0.0086 | 0.5043 | 0.4957 |
A7 | −0.248 | 0.376 | 0.624 | 0.1427 | 0.5713 | 0.4287 | 0.1696 | 0.5848 | 0.4152 |
A8 | −0.372 | 0.304 | 0.676 | 0.3667 | 0.6647 | 0.298 | −0.2805 | 0.3281 | 0.6086 |
A9 | −0.288 | 0.356 | 0.644 | −0.408 | 0.296 | 0.704 | −0.2515 | 0.3743 | 0.6257 |
A10 | −0.2227 | 0.3887 | 0.6113 | −0.4587 | 0.2707 | 0.7293 | −0.1446 | 0.4264 | 0.571 |
A11 | −0.14 | 0.43 | 0.57 | −0.4947 | 0.2527 | 0.7473 | −0.2383 | 0.3809 | 0.6191 |
A12 | −0.2453 | 0.3773 | 0.6227 | −0.2267 | 0.3867 | 0.6133 | 0.2132 | 0.6053 | 0.3921 |
A13 | −0.4827 | 0.2587 | 0.7413 | −0.0933 | 0.4533 | 0.5467 | 0.0376 | 0.5188 | 0.4812 |
A14 | −0.4227 | 0.2787 | 0.7013 | −0.3573 | 0.3027 | 0.66 | −0.5089 | 0.2139 | 0.7228 |
r1 | 0.4967 | 0.7393 | 0.2427 | 0.5313 | 0.7453 | 0.214 | 0.4389 | 0.7195 | 0.2805 |
r2 | 0.0187 | 0.5093 | 0.4907 | −0.096 | 0.452 | 0.548 | −0.13 | 0.435 | 0.565 |
A1 – Tenente Ananias; A2 – Pilões; A3 – Pau dos Ferros; A4 – Campo Grande; A5 – Umarizal; A6 – Caraúbas; A7 – Lucrécia; A8 – Severiano Melo; A9 – Encanto; A10 – José da Penha; A11 – Marcelino Vieira; A12 – Rafael Fernandes; A13 – Riacho da Cruz; A14 – Rodolfo Fernandes.
r1 – Limiting Profile 1; r2 – Limiting Profile 2.
Individual sorting of municipalities that need the resource allocation in drought mitigation.
Individual sorting of municipalities that need the resource allocation in drought mitigation.
According to each decision-maker's preferences, the municipalities were classified according to their priority in resource allocation. The first decision-maker obtained the following municipalities with High Priority: A1 – Tenente Ananias, and A3 – Pau dos Ferros. The municipalities considered with Moderate Priority were: A4 – Campo Grande, A5 – Umarizal, and A6 – Caraúbas. The municipalities considered with Low Priority were: A2 – Pilões, A7 – Lucrécia, A8 – Severiano Melo, A9 – Encanto, A10 – José da Penha, A11 – Marcelino Vieira, A12 – Rafael Fernandes, A13 – Riacho da Cruz, and A14 – Rodolfo Fernandes.
Conversely, the municipality with High Priority to the second decision-maker is A1 – Tenente Ananias. The municipalities considered with Moderate Priority were: A2 – Pilões, A3 – Pau dos Ferros, A4 – Campo Grande, A7 – Lucrécia, A8 – Severiano Melo, and A13 – Riacho da Cruz. The municipalities considered with Low Priority in resource allocation to mitigate drought were: A5 – Umarizal, A6 – Caraúbas, A9 – Encanto, A10 – José da Penha, A11 – Marcelino Vieira, A12 – Rafael Fernandes, and A14 – Rodolfo Fernandes.
Additionally, the municipality sorted with ‘High Priority’ to the third decision-maker is A1 – Tenente Ananias. The municipalities considered with Moderate Priority were: A2 – Pilões, A3 – Pau dos Ferros, A4 – Campo Grande, A5 – Umarizal, A6 – Caraúbas, A7 – Lucrécia, A12 – Rafael Fernandes, and A13 – Riacho da Cruz. The municipalities considered with ‘Low Priority’ were: A8 – Severiano Melo, A9 – Encanto, A10 – José da Penha, A11 – Marcelino Vieira, and A14 – Rodolfo Fernandes.
Group sorting of municipalities that need the resource allocation to mitigate drought.
Group sorting of municipalities that need the resource allocation to mitigate drought.
To the group decision-making, the municipality with High Priority in resource allocation is A1 – Tenente Ananias. The municipalities sorted as Moderate Priority for the decision-makers group were: A2 – Pilões, A3 – Pau dos Ferros, A4 – Campo Grande, A5 – Umarizal, A6 – Caraúbas, A7 – Lucrécia, and A13 – Riacho da Cruz. The municipalities classified with Low Priority in resource allocation were: A8 – Severiano Melo, A9 – Encanto, A10 – José da Penha, A11 – Marcelino Vieira, A12 – Rafael Fernandes, and A14 – Rodolfo Fernandes.
In the final part of the methodology, we undertook a sensitivity analysis to evaluate the sorting stability of each decision-maker. We conducted weight variations to each construct, resulting in an increase and decrease of 20% of the weight for each criterion for every decision-maker. Thus, it is possible to statistically detect, for each decision-maker, the changes in the priority of each municipality and possible inconsistencies in information. The first decision-maker generated nine sortings for each alternative. To the second and third decision-makers, seven and eleven sortings were generated for each city, respectively. Table 10 presents the data sensibility analysis.
Data sensibility analysis.
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | C1 . | C2 . | C3 . | C1 . | C2 . | C3 . | C1 . | C2 . | C3 . |
A1 | 100% | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% |
A2 | 0% | 0% | 100% | 0% | 100% | 0% | 0% | 55% | 45% |
A3 | 78% | 22% | 0% | 0% | 100% | 0% | 0% | 91% | 9% |
A4 | 0% | 100% | 0% | 0% | 100% | 0% | 9% | 91% | 0% |
A5 | 0% | 100% | 0% | 0% | 43% | 57% | 0% | 73% | 27% |
A6 | 0% | 100% | 0% | 0% | 0% | 100% | 0% | 100% | 0% |
A7 | 0% | 0% | 100% | 0% | 100% | 0% | 0% | 100% | 0% |
A8 | 0% | 0% | 100% | 0% | 100% | 0% | 0% | 9% | 91% |
A9 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% | 100% |
A10 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 36% | 64% |
A11 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% | 100% |
A12 | 0% | 0% | 100% | 0% | 14% | 86% | 0% | 100% | 0% |
A13 | 0% | 0% | 100% | 0% | 57% | 43% | 0% | 100% | 0% |
A14 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% | 100% |
. | Decision-maker 1 . | Decision-maker 2 . | Decision-maker 3 . | ||||||
---|---|---|---|---|---|---|---|---|---|
. | C1 . | C2 . | C3 . | C1 . | C2 . | C3 . | C1 . | C2 . | C3 . |
A1 | 100% | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% |
A2 | 0% | 0% | 100% | 0% | 100% | 0% | 0% | 55% | 45% |
A3 | 78% | 22% | 0% | 0% | 100% | 0% | 0% | 91% | 9% |
A4 | 0% | 100% | 0% | 0% | 100% | 0% | 9% | 91% | 0% |
A5 | 0% | 100% | 0% | 0% | 43% | 57% | 0% | 73% | 27% |
A6 | 0% | 100% | 0% | 0% | 0% | 100% | 0% | 100% | 0% |
A7 | 0% | 0% | 100% | 0% | 100% | 0% | 0% | 100% | 0% |
A8 | 0% | 0% | 100% | 0% | 100% | 0% | 0% | 9% | 91% |
A9 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% | 100% |
A10 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 36% | 64% |
A11 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% | 100% |
A12 | 0% | 0% | 100% | 0% | 14% | 86% | 0% | 100% | 0% |
A13 | 0% | 0% | 100% | 0% | 57% | 43% | 0% | 100% | 0% |
A14 | 0% | 0% | 100% | 0% | 0% | 100% | 0% | 0% | 100% |
A1: Tenente Ananias; A2: Pilões; A3: Pau dos Ferros; A4: Campo Grande; A5: Umarizal; A6: Caraúbas; A7: Lucrécia; A8 – Severiano Melo; A9: Encanto; A10: José da Penha; A11: Marcelino Vieira; A12: Rafael Fernandes; A13: Riacho da Cruz; A14: Rodolfo Fernandes.
C1: High priority; C2: Moderate priority; C3: Low priority.
The sensitivity analysis showed that seven municipalities had variations in their classifications. Only the Pau dos Ferros (A3) changed its initial sorting for the first decision-maker. The results showed that 78% of the simulated cases maintained the stability of the original classification. Namely, out of the nine simulations, seven cases are High Priority, reflecting their initial sorting. Conversely, only two cases, or 22% of the simulated cases, are considered a lower priority sorting for the water resource allocation process, i.e., two cases are Moderate Priority, diverging from the initial sorting.
As the second decision-maker, the municipalities of Umarizal (A5), Rafael Fernandes (A12), and Riacho da Cruz (A13) had their indexes changed. To Umarizal City (A5), 57% of the simulated cases were considered Low Priority, equal to initial sorting. Conversely, only 43% of simulations have a higher rating than the original, with Moderate Priority. To Rafael Fernandes City (A12), 86% of simulations are Low Priority, and only 14% of cases are different from the original, with Moderate Priority. Finally, the Riacho da Cruz City (A13) remained stable in its classification in 57% of the cases as Moderate Priority and 43% of simulations with sorting of the lower priority level, with Low Priority.
As the third decision-maker, Pilões (A2), Pau dos Ferros (A3), Campo Grande (A4), Umarizal (A5), Severiano Melo (A8), and José da Penha (A10) had their indexes changed. To Pilões City (A2), 55% of cases are Moderate Priority, and 45% with Low Priority. To Pau dos Ferros city (A3), 91% of simulations are Moderate Priority and 9% with Low Priority. Conversely, the Campo Grande (A4), 91% of cases are Medium Priority, and 9% are High Priority. Additionally, in Umarizal (A5), 73% of simulations are Moderate Priority, and 27% are Low Priority. In Severiano Melo City (A8), 91% of cases are Low Priority, and only 8% are Moderate Priority. Finally, José da Penha City (A10) had 64% with Low Priority and 36% with Moderate Priority.
Therefore, most alternatives reproduced the same sortings of the first recommendation in 100% of the simulations in the sensitivity analysis. These results highlight the quality and reliability of the evaluations and results. However, most of the variations in the classifications of the alternatives indicated a higher priority in resource allocation to drought mitigation, which may indicate a possible worsening of the situation in the municipality. This information can provide valuable insights for formulating more efficient resource allocation strategies. Therefore, we recommend acting initially in high-priority municipalities. Afterward, the public administration can direct actions to municipalities with moderate and low priorities.
POLICY IMPLICATIONS
Based on the results, we classified the municipality of Tenente Ananias as a high priority. Unlike the approach by Gonçalo & Morais (2018), who ranked the municipalities based on drought, our municipalities categorization allows us to identify a set of cities in a critical drought state. By focusing on multiple municipalities instead of serving them individually, the public administration can concentrate its efforts and resources more efficiently. This approach enables us to address drought and mitigate multiple problems simultaneously, benefiting a group of affected cities.
The Rio Grande do Norte State, Brazil, heavily relies on reservoirs and private wells for its water resources. The state suffers from social, economic, and infrastructure weaknesses. The region's economic activity primarily revolves around irrigated agriculture. However, the sandy characteristics of the soil in the area result in high-water infiltration. Despite the high-water demand, the region has limited storage capacity, leading to long and frequent periods of drought. Hence, there is a constant need to develop public policies for this region. Applying this model offers a structured approach that enables a more effective direction of such policies.
It is worth mentioning that the alternatives considered in our case study showed little variation in other economic indicators, such as GDP. However, in different scenarios where alternatives significantly differ in this economic aspect, including these indicators becomes essential for a thorough evaluation. Our flexible model ensures consideration of all relevant variables when allocating resources to mitigate drought.
The Brazilian government is responsible for developing and implementing actions to minimize or mitigate the impact of catastrophic natural events, including drought. It's necessary to identify the needs of affected municipalities and allocate resources such as emergency financial resources, parliamentary amendments, and water resources to mitigate the adverse consequences of this disaster. The resource allocation may include drilling wells, installing emergency pipelines, and implementing alternative water supply programs such as water trucks for intercity transportation.
Effectively tackling drought necessitates considering various perspectives when developing and implementing actions. Our approach, based on group decision-making and multicriteria sorting, provides a valuable method for considering these perspectives and directing drought-coping projects toward priority regions. By applying this model, the public administration can efficiently allocate scarce resources to areas impacted by drought, resulting in a more effective and targeted response to this challenge.
Dealing with drought and implementing effective measures typically requires the participation of a large number of decision-makers. Our research recognizes that drought management can be approached from a variety of angles, thereby enhancing the formulation of public water policies. Even though the model was applied numerically to only three decision-makers, it is notable that the proposed model can be applied to contexts with more decision-makers, allowing it to be used more effectively in the context of resource distribution for the entire state of Rio Grande do Norte.
CONCLUSIONS
Drought is a global natural disaster and needs to be effectively managed by the public administration using a structured methodology capable of incorporating the value judgments and diverse perspectives of the individuals involved in this decision. The objective of this study was to present an approach that can sort municipalities requiring resource allocation to address drought through the group multicriteria sorting methodology. As a result, this research established a model that can assist in the decision-making process of public policies, enabling the construction and structuring of strategic public policies and focusing efforts and resources on communities most impacted by drought based on their actual needs.
An incorrect definition of municipalities in a critical drought situation can lead to ineffective public resources and efforts allocation. Through the proposed model, it is possible to categorize drought-affected municipalities, both individually and collectively. As a result, competent deployment of efforts and resources can minimize and alleviate the negative effects of this natural calamity, offering valuable input for future research.
The city's identification affected by drought often relies on political decisions and involves a high degree of subjectivity. The study introduced a more impartial, structured, and transparent approach, including scientific indicators such as hydro-sanitary and socioeconomic factors. Consequently, the study offers a fairer structure, directing available resources towards the areas in greatest need and improving the impact of drought mitigation measures.
Although our study considered the perspective of only three decision-makers, the approach enables large-scale group decision-making, allowing for the consideration of multiple perspectives in formulating more balanced and effective public water policies. As a future proposal, the proposed model could be applied to all municipalities in the semi-arid region of Brazil and other countries to foster the development of national public policies for resource allocation mitigation drought from multiple perspectives.
ACKNOWLEDGMENT
This work was carried out with the support of the Coordination of Improvement of Higher Education Personnel – Brazil (CAPES) with financing code 001 and also by the National Council for Scientific and Technological Development – Brazil (CNPq).
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.