Sustainable water resource planning at the basin scale with simultaneous goals of agricultural development and wetland conservation

Integrated water resource management is a systematic process for a sustainable development, allocation, and monitoring of water resources that are used for social, economic, and environmental purposes. Population growth, agricultural development, expansion of industries, heterogeneous temporal and spatial distribution of freshwater in terms of quantity, and increasing the quality problems of water resources have led to reliable water supply becoming a challenge in many countries.

Agriculture and the ecosystem are increasingly competing for water. Large amounts of water have been diverted to irrigate agriculture and other human activities in many river basins around the world. According to Calzadilla et al. (2010), approximately 70% of natural water resources are diverted annually from global river systems to irrigate agriculture. Human activities have been an important factor in the widespread loss of wetlands throughout history. For example, the Millennium Ecosystems Evaluation Report (MEA 2005) reports that between 80 and 98% of wetlands in or near important urban centers have disappeared worldwide. Such a trend has intensified over the years. Since 1985, 26% of the wetland areas have been destroyed by intensive agricultural drainage. Hoor al-Azim, meaning a very large wetland, is one of the largest wetlands in Iran. Hoor al-Azim Wetland has been experiencing water shortages in recent years, and it has caused a variety of social, economic, and environmental problems. One of the reasons is the overdevelopment of agricultural activities in the Karkheh basin. Therefore, one of the main purposes of this study is to achieve a balance between agricultural development and environmental protection in the basin.

Due to Iran's location in arid and semi-arid climates, the optimal use of water resources and better management in conditions of water shortage is essential to achieve sustainable development. This cannot be done without a systematic view of the basin and considering all the resources and demands. Planning and management at the basin level is a complicated and large-scale problem that emphasizes the effective application of modeling tools. Mathematical programming models can be divided into the simulation, optimization, and simulation–optimization models. Although simulation models are effective models in evaluating the performance of water resource systems in a variety of conditions, these models are not perfect for selecting and defining the best combination of system elements. This shortcoming can be addressed by using optimization tools and being equipped with intelligent search algorithms. These algorithms search for the best combination of variables in the automatic and evolutionary search process. The simulation–optimization approach has frequently been used to handle various issues related to water resource management (Cai et al. 2004) such as water allocation, due to its multidisciplinary nature and complexity. But most of these models have utilized economic objectives in water allocation modeling and left out the consideration of the environmental factors (Roozbahani et al. 2015).

Liu et al. (2010) presented a multi-objective model for the optimal allocation of water resources in saltwater intrusion areas. Objective functions include the following: maximizing economic interest and social satisfaction and minimizing the amounts of polluted water. The model was employed in the Pearl River Delta in China while ignoring the environmental flow satisfaction. Ahmadi et al. (2012) introduced a multi-objective genetic algorithm model for water allocation in the Aharchay watershed, Iran. The model provided desirable water quality and quantity for various sectors while maximizing agricultural production in the upstream region, mitigating the unemployment (social) impacts of land use changes, and providing reliable water supply to the downstream region. In this study, the quality of water was considered rather than the supply of environmental water demand.

Anghileri et al. (2013) developed a multi-objective optimization model for resolving disputes between the agricultural sector and electricity generation in the Alps watershed in Italy. The objective of this model has been to minimize the shortage of irrigation demand and to maximize hydropower generation. This was solved using the weight gain method, while the inflows of reservoirs were considered stochastic. In this study, satisfying the environmental water requirements was a firm constraint. Rezapour Tabari & Yazdi (2014) proposed a multi-objective model based on inter-basin water resources and restoration of outer-basin water resources. Three water allocation objectives were considered reducing water output from a basin boundary, supplying inter-basin water demand, and increasing water transfer to adjacent basins. These researchers failed to consider the environmental factors in their study.

Nabinejad et al. (2017) used the particle swarm optimization (PSO)–MODSIM model to determine optimal basin-scale water allocation. Optimal basin-scale water allocation was evaluated by incorporating reliability, vulnerability, reversibility, and equity sustainability indices into the PSO objective function. The extended model was applied for water allocation in the Atrak River Basin, Iran. An explanation of how the environmental flow requirements are included in the model and how it is met is not given.

Habibi Davijani et al. (2016) developed a two-objective socioeconomic model (aimed at job creation) that has been presented in this study for the optimum allocation of water resources to industry, agriculture, and municipal water sectors. In this study, the environmental factors have been completely ignored and no discussion has taken place.

Hatamkhani & Moridi (2019) employed a simulation–optimization approach to solve the problem of optimal planning at the basin scale. The objective functions of this problem were (1) maximizing the area under cultivation of agricultural development sectors and (2) maximizing the energy generation by the hydropower plant. In this study, environmental flow is included as an input in the simulation model and no details are provided about how it is supplied and the effect of development on it. Haavisto et al. (2018) determined the amount of payment for watershed services by developing a hydro-economic model to optimally allocate between agricultural and urban needs. The model consists of three parts, including the agricultural model, the urban water model, and the hydrological constraint. Alamanos et al. (2019) presented a holistic hydro-economic framework for sustainable water resource management, simply and understandably for policy-makers. It is examined under various management, climate, and pricing scenarios. The proposed framework is based on (a) the modeling of water balance and (b) the use of various hydro-economic outputs (e.g., irrigation water value, farmers’ utility, efficiency indexes, direct costs, etc.). Environmental factors do not play a role in the management policies of the developed model.

Musa (2021) employed a multi-objective model for the optimal allocation of water from limited resources to meet increasing demands in multiple sectors. The goal programming technique is used to formulate the model with multiple objectives. These objectives were (1) water demands satisfaction; (2) water quality control; (3) maximizing allocation of surface water and treated wastewater; and (4) minimizing extraction of groundwater. The water users of the study were domestic, agricultural and industrial sectors, and environmental factors were completely neglected. Hatamkhani & Moridi (2021) presented a water allocation model that considers the interaction of water supply and demand according to economic and social factors. To achieve this, a multi-objective optimization–simulation model has been used. The objective functions of the problem were as follows: (1) maximizing GDP of agricultural sectors and (2) maximizing social equality in different provinces of the basin. In this research, environmental requirements are considered a constraint in the Water Evaluation and Planning (WEAP) model.

In some studies, environmental factors are considered as an objective function such as minimizing shortages. Roozbahani et al. (2015) introduced a multi-objective model, which leads to sustainable water allocation of transboundary basins. Five water allocation objectives were considered for this model in which three of them represented the social factors and others addressed the economic and environmental preferences. Wang et al. (2015) constructed an optimal water resource allocation model to achieve the maximum comprehensive benefit for the river basin socioeconomic and eco-environmental systems through allocating limited water resources. The objective functions of the problem included economic objective, social objective, and ecological objectives. This paper used ecological water deficit to measure ecological objective.

Based on the previous studies in water resource optimization models, the objective functions usually include agricultural, urban, industrial, or hydropower sectors and the environmental and ecological factors are either not considered or considered as constraints. Even limited studies that have considered the environmental requirements as an objective function usually investigated the issue of allocation to existing demands, and this problem has been less studied in the planning and design phase. This paper attempts to propose a new perspective for optimal development planning in a water-deficient river basin in order to achieve the desired goals, sustainable development and conduct an accurate and feasible program for water resource allocation between agriculture and environment. To achieve these, a long-term multi-objective optimization model is employed that considers the environmental factors and agriculture development simultaneously. The objective functions of the problem include maximizing the total area under cultivation of agricultural development projects downstream of the Karkheh Dam, and maximizing the supply reliability of the environmental flow for the ecological conservation of Hoor al-Azim wetland. The model proposed in this study is reliability-based. Agricultural development units are estimated such that they are supplied with a minimum acceptable level of reliability, which leads to a sustainable and efficient development in the basin.

Using the multi-objective particle swarm optimization (MOPSO) and its integration with WEAP simulation model, the simulation–optimization model is developed and employed in the problem of water resource development planning of the Karkheh basin.

In this research, the objectives of the two-objective optimization problems include maximizing the total area under cultivation of agricultural development sectors downstream of the Karkheh Dam and maximizing the reliability of providing the environmental requirement of Hoor al-Azim Wetland. It should be noted that the cultivated area of agricultural units is estimated according to the reliability of their supply so that the reliability does not fall below 80%. Decision variables of the problem are the area under cultivation of agricultural development projects (⁠⁠).

The mathematical programming presented is a nonlinear and complex problem. The optimization–simulation approach was employed for solving this problem. Figure 1 shows the flowchart of the proposed model.

In the simulation–optimization process, first, the decision variables are generated by the MOPSO algorithm (coded in MATLAB) and written in Excel. Then, these variables are read by the WEAP model and the water allocation model is executed. According to the water allocated to the environment (wetland) and agricultural sectors, the objective functions are calculated. It should be noted that the objective functions are coded in the MOPSO algorithm and the problem constraints are considered internally in the WEAP model. If the MOPSO stop criteria are met, the algorithm stops. Otherwise, the particle velocity and location are updated and new decision variables are generated. This process is repeated until the optimal trade-off (Pareto front) is obtained.

The WEAP model is a computer tool for integrated water resource planning. This software has been developed using the comprehensive approach to simulation of water systems by the Stockholm Environment Institute to bridge the gap between resource management and hydrology of the basin, which is a comprehensive, flexible, and user-friendly software for policy analysis. WEAP is a suitable tool for testing water development and management strategies. This software is widely used for integrated programming. WEAP has river, economic, and agricultural demand management sub-models (Sieber & Purkey 2011).

WEAP is based on basic water balance equations and can be used in urban and agricultural systems, stand-alone basins, or complex water systems. WEAP also addresses a wide range of issues such as sector demands analysis, water conservation, water allocation priorities, surface and groundwater simulation, reservoir operation, hydropower generation, pollution routing, vulnerability assessment, and benefit-cost analysis of the projects (Hatamkhani et al. 2021).

The PSO algorithm was first introduced by Eberhart and Kenney in 1995 (Kennedy & Eberhart 1995). This algorithm, like other evolutionary computing techniques, uses a population or swarm that contains potential solutions to the problem to explore the search space. In PSO, each member of the population has an adaptive speed according to which they move in the search space, in addition, each of them also has a memory, i.e. the best position they reach in the search space. A group of particles is created randomly at the beginning of the algorithm and tries to find the optimal solution by updating the generations. At each iteration, each particle is updated using the two best values. The first is the best situation the particle has ever been able to reach, which is represented by pbest. The other best value used by the algorithm is the best position ever obtained by a particle (gbest).

The MOPSO algorithm of this study is similar to the method of Coello et al. (2004). In the MOPSO, a concept called archive or repository has been added to the PSO algorithm. Choosing the best overall answer and the best personal memory for each particle is a fundamental step in the particle swarm optimization algorithm. When particles want to make a move, they choose a member of the repository as a leader. This leader must be a member of the repository as well as a non-dominated solution. The members of the repository represent the Pareto front. So instead of gbest, one of the members of the repository is selected. This is why there is no repository in the PSO because there is only one objective. But in MOPSO several particles are non-dominated and are in the solution set (Hatamkhani et al. 2020).

Karkheh basin with an area of about 50,000 km², is one of the important river basins in Iran, especially in terms of agricultural and electricity production. The Karkheh River consists of two main branches called Seymareh and Kashkan. Figure 2 shows the average inflow to the Karkheh reservoir and monthly evaporation from the reservoir.

The Karkheh Dam is the largest embankment dam in Iran and the Middle East, which is built on the Karkheh River, 22 km northwest of Andimeshk city in Khuzestan province. The most important goals of the Karkheh Dam are to supply and regulate water for irrigation of downstream lands, generating hydropower, and flood control. The specifications of the Karkheh Dam are presented in Table 1 and the volume–elevation curve of the Karkheh reservoir is presented in Figure 3.

Hoor al-Azim is located in the west of Khuzestan province at the end of the Karkheh River in the border area of Azadegan plain between Iran and Iraq. The depth of this wetland is low, but gradually it reaches several meters in the middle. The whole of Hoor al-Azim is covered with reeds. The area of the wetland is about 118,000 ha. Hoor al-Azim is a very rich wetland in terms of animal and plant resources, which are now facing extinction. The destruction of the wetland can cause irreparable economic, social, and environmental damage. Figure 4 shows the trend of changes in the wetland area.

The drying up of the wetland has created many problems for the people of the region. Some of the most important problems include the destruction of plant and animal habitats, unemployment and flood migration from the region, and climate change and the most important issue is the conversion of wetland into a source of dust storms and causes damage to human health, livestock, and agricultural products. Therefore, long-term planning in the basin is necessary to achieve various goals, including the development of agriculture on which the country's food security depends, as well as the provision of environmental water requirements to achieve sustainable development in the basin. The minimum environmental requirement of Hoor al-Azim wetland (Iran Water and Power Resources Development Company 2015) is presented in Figure 5.

The simulation model for long-term planning and water allocation is developed in WEAP software. The simulation period is considered to be 59 years with monthly time steps. Figure 6 demonstrates the schematic of the Karkheh basin in WEAP.

The MOPSO–WEAP model is used to solve the problem of optimal development and water allocation in the basin. The decision variables are the area under the cultivation of agricultural development projects downstream of the Karkheh Dam, so there are 17 decision variables. The minimum and maximum values of the decision variables are shown in Table 2.

As mentioned before, the objective functions in this study are (1) maximizing agricultural development sectors (cultivated area) and (2) maximizing the supply reliability of Hoor al-Azim flow requirement, which are conflicting. Using the MOPSO algorithm with 50 particles, the optimal Pareto curve between objectives after 200 iterations is displayed in Figure 7. This curve represents the non-dominated solutions to the problem.

According to Figure 7, it is clear that the supply reliability of the environmental demand obtained from the optimization–simulation model is about 76–80%, which is a suitable range to meet the requirements of the wetland. In the following, two optimal solution points from the Pareto curve are selected for further study, which are presented in Table 3.

In solution 2, due to the increase of the area under cultivation (about 3.5%) compared to solution 1, the supply reliability of the wetland demand has decreased by about 3%. The solutions obtained from the Pareto curve are close to each other and the best and worst solutions obtained based on objective function 1 (area under cultivation) differ by only about 5%. Another important point is the small amount of penalty resulting from the reliability violation (Equation (1)), which for solutions 1 and 2 are 17 and 23%, respectively.

When we consider the environmental flow as a hard constraint in the optimization problem, which is not possible in practice because economic development has a higher priority in the basin, and in case of water shortage the first sector that suffers is the environment, but it provides a good understanding of the condition of the basin. In this case, the coverage and supply reliability of wetland are 100%. By solving the problem again as a single objective optimization, the area under cultivation of agricultural sectors is calculated according to Table 3. As can be seen, the full supply of the wetland requirement is possible only with a significant decrease in the agricultural sector. The amount of reduction in area under cultivation is 44.5% compared to the previous study and 21.3% compared to the optimal result obtained in this study (solution 2).

For comparison and further discussion, the last column of Table 3 shows the results of studies conducted by Iran Water & Power Resources Development Company (2015). According to the results, it is clear that supplying the environmental requirement is significantly improved and reliability is increased from 55 to 76.7%. Figure 8 shows the average monthly amount of wetland requirement coverage (% of requirement met). As shown in Figure 8, the supply of wetland requirements in some months of the year, especially from April to September is reduced, which is understandable given the amount of rainfall in these months. However, the coverage in the present study did not fall below 74% in any of the months of the year. In the previous study, the coverage in June is about 56%.

As stated in Table 3, the nominal area under cultivation has been estimated more in previous studies. However, it should be noted that in addition to significantly reducing the reliability of supplying Hoor al-Azim, the value of objective function 1 is calculated less than the results of the current study. Due to the lack of adequate supply of agricultural development demands and failure to satisfy the minimum desired reliability, objective function 1 is severely penalized. The amount of nominal area under cultivation in the study is 334,425 ha, which has reached 179,130 ha after the penalty and is decreased by 46%. This matter is further clarified in Table 4, which shows the supply reliability values of agricultural development units. According to Table 4, the average supply reliability of agricultural demands according to previous studies is 51%. But in this study, the reliability is reached to 72%. This indicates a lack of proper planning in the basin, which not only environmental requirement does not supply well but also the agricultural developments have not been properly planned and designed according to the situation of the basin resources.

The agricultural sector is the main engine of economic growth in developing countries. Developing countries need to consider the agricultural sector as one of the main pillars of economic development to get out of economic crises. Besides, increasing irrigation and drainage has caused major damage to the environment, which necessitates achieving a balance between the two sectors. Therefore, it seems that the most important way to achieve sustainable development is optimal development planning at the basin scale, which in addition to showing how to properly develop agriculture sectors, environmental protection can be done well. Hoor al-Azim Wetland is one of the largest wetlands in southwestern Iran, which has caused many social, economic and environmental problems due to the lack of proper supply of environmental requirements. Therefore, a long-term multi-objective optimization–simulation approach was used for development planning. The decision variables were considered the area under the cultivation of agricultural units. The objective functions were the total cultivated area of the agricultural units and the reliability of supplying the Hoor al-Azim requirement. Given the conflict between these goals and the limited water resources in the basin, achieving an optimal trade-off between these sectors is necessary. Therefore, the simulation–optimization approach was employed. The WEAP model was used to simulate and manage the water resource allocation in the basin and the MOPSO algorithm was employed as an optimizer. According to the Pareto curve obtained from the MOPSO–WEAP model, the supply reliability of the Hoor al-Azim wetland has reached about 76–80%, which is acceptable. For comparison, in previous studies, the supply of Hoor al-Azim demand was met with 55% reliability. Also, the average supply reliability of agricultural demands has increased from 52 to 72%, which shows the good performance of the model and proper planning in the basin. Long-term modeling of the basin shows that excessive increase in the area under cultivation of agricultural projects leads to unsustainable development. This will cause many problems, especially in drought periods. On the one hand, the lack of proper supply of agriculture demands will create competition and tension among stakeholders, and on the other hand will lead to increased use of environmental water, which causes many economic, social, and environmental issues.

The authors declare that there are no conflicts of interest.

Data cannot be made publicly available; readers should contact the corresponding author for details.