A WEAP-assisted agent-based modeling approach for watershed planning in a coupled human and nature system

Effective management of water resources has always been crucial for human survival and economic stability. The United Nations has warned that diminishing surface and groundwater resources in water-scarce regions will hinder economic growth and elevate poverty levels. This emphasizes the urgent need to manage and preserve water resources to ensure sustainable development and economic stability (Beck & Villarroel Walker 2013).

Reservoirs are pivotal in agricultural water management, serving to balance downstream water requirements while safeguarding against adverse impacts on farming activities. Effective reservoir operation hinges on the precise identification of users and anticipation of their consumption patterns. Given the far-reaching consequences of water management choices, it becomes imperative to account for the intricate interplay within water systems, impacting all stakeholders involved.

Aligned with this, modern water resource management seeks to balance supply and demand while preventing the depletion of surface water and aquifers. Traditional water engineering has predominantly focused on hydrological factors, such as precipitation, runoff, and water storage. However, understanding the behaviors and interactions of individuals and organizations involved in water use is equally important. To address the complex challenges of water resource management effectively, it is essential to integrate engineering and management sciences through advanced modeling techniques. This integrated approach helps to capture both the physical aspects of water systems and the social dynamics that influence water consumption and conservation practices (Mollinga 2008). By doing so, water resource management can be more adaptive and resilient, ensuring sustainable water availability for various needs (Kuil et al. 2019).

Agent-based modeling (ABM) emerges as a promising technique for simulating intricate systems, encompassing interactions between human and water systems. ABM concentrates directly on individual entities, their behaviors, and interactions, offering a detailed portrayal of how individual actions amalgamate to shape larger patterns (Bonabeau 2002). This approach allows for a nuanced comprehension of modern social systems' complexity, capturing dynamic interactions between agents and providing valuable insights into emergent system properties. Consequently, ABM facilitates the development of more effective and adaptive management strategies (Akhbari & Grigg 2013).

In this regard, recent studies have demonstrated the potential of ABM to improve water management. For example, Cai & Xiong (2017) developed an ABM for organizing agricultural irrigation systems, highlighting the importance of government support, social learning, and cost management in promoting cooperation. Similarly, Huber et al. (2021) used ABM to investigate water balance in socio-ecological systems in the Italian Alps, demonstrating that water demand consistently outstrips supply under various scenarios, particularly with climate change considerations.

Bahrami et al. (2022) combined ABM with the standard operating policy (SOP) to manage water release and allocation from dams, revealing that dynamic interactions between farmers and reservoir operators can significantly reduce water stress and increase profits through adaptive decision-making. These studies underscore the necessity of integrating hydrological models with stakeholder interaction components to achieve effective water management.

Javan Salehi & Shourian (2023) developed a socio-hydrological model by integrating soil and water assessment tool (SWAT) and MODFLOW to analyze human–water systems. Using the value–belief–norm theory, they identified factors influencing farmers' water usage behaviors. Their findings indicate that climate change could lead economically disadvantaged farmers to favor lucrative yet water-intensive crops, potentially impacting Lake Urmia's water inflow. The study recommends policy adjustments to mitigate these effects, providing insights for future strategies and policy modifications aimed at restoring Lake Urmia amidst evolving economic and psychological factors for farmers.

Aghazadeh et al. (2024) investigate sustainable water management in Iran's Eshtehard plain, focusing on the consequences of excessive groundwater use and declining groundwater levels. Employing agent-based modeling, they analyze stakeholder interactions to improve groundwater management. Their framework models' farmers' reactions to evolving water resources, factoring in agricultural demand and socio-economic conditions. This proposed approach offers stakeholders a valuable tool for making well-informed decisions toward sustainable groundwater management.

According to the reviewed studies, the current study represents an effort toward comprehensive socio-hydrological modeling. In this study, attempts have been made to offer a suitable and comprehensive solution to reduce the extraction of surface water resources with minimal impact on farmers' income. The agents serve as representatives of the farmers in Miandoab basin. Additionally, the environment where the agents interact in the agent-based environment downstream of Zarrineh-Roud Dam has been taken into consideration. The next section will present the step-by-step process of model development.

In this section, the environment simulation, the agent's behavior simulation and agent's decision-making method are explained.

WEAP software is a tool for integrated water resources planning that provides a comprehensive, flexible, and user-friendly framework for policy planning and analysis. Around the world, many regions are facing serious challenges in the field of water resources management. Traditional resource-oriented models (which focus on water resources rather than water consumption) are not always suitable for examining different management options. During the last decades, the integrated approach has increased in the development of water resources plans, and the needs, quality issues, economic issues, and other issues related to water resources have been taken into consideration. WEAP uses these issues together in a practical tool for water resource planning and policy analysis. Rivers, diversions, reservoirs, underground water sources, areas of need, drainage basins, runoff and infiltration, transmission lines, wastewater treatment plants, return flows from points of need, river power plants, required discharge, and flow measurement gauges (for comparison of calculated flow and data real) are elements that can be simulated in this model.

The catchment area downstream of the Zarrineh-Roud dam contains two regions of Saeen-Qala and Miandoab. Zarrineh River consists of Saqez, Khorkhore chai, and Sarouk chai branches and enters Zarrineh-Roud reservoir dam in Yamin Abad. Noruzlu diversion dam is also in operation with the aim of transferring water to the places of consumption. Between Zarrineh-Roud Reservoir Dam and Noruzlu Diversion Dam, other sub-branches, including Ajarlu, join this river. In the operation of the Zarrineh-Roud reservoir dam system and the Noruzlo diversion dam, the following water needs are considered, including the needs of agriculture, drinking, industry, and the environment (Table 1) based on the statistics provided in the agricultural studies report of Zarrineh-Roud watershed. Allocation of environmental water rights of rivers as a missing link in the sustainable management of water resources in Iran is an undeniable necessity to preserve the existing ecosystems in the watersheds of the country and also to prevent the death of wetlands and lakes leading to rivers.

An agent-based model (ABM) consists primarily of agents and the modeling environment. An agent serves as an autonomous entity that plays a role in the decision-making process and can take the form of an individual, an institution, or a group of stakeholders. These agents operate within a dynamic environment, engaging in interactive and mutually influential interactions with their surroundings.

In this study, Zarrineh-Roud Basin is identified as the environment and agricultural unions are identified as independent agents who interact with their environment through agricultural activities. Agents exhibit responses to the amount of available water and modify their consumption behavior accordingly. All of these agents have the power of recognition, reasoning, comparison, historical memory, and decision-making.

Agents make assumptions about the availability of water for the next year, then they decide on the pattern of cultivation for the upcoming year. In the first five years of modeling, the agents have the same and fixed cropping pattern (based on the information of the Agricultural Jihad Organization, ignoring partially cultivated crops) in order to create the necessary historical memory to make a decision in the first year of cropping pattern change. This initial model consists of the cultivation of wheat, barley, alfalfa, sugar beet, tomato, and potato crops, respectively, equal to 48, 13, 30, 4, 3, and 2% for Miandoab and Sain Dej agents. The farmers' behavioral decision-making model consists of four steps that are done in order and finally ends with the next year's cropping pattern. The decision-making behavior model of farmers consists of four steps performed in order and ultimately ends with the cultivation of the following year.

P4 assesses the combined utilization of available water by all agents relative to their long-term performance in the previous year. A P4 value exceeding 1 signifies that agents are collectively employing water more responsibly than in past years, mitigating concerns about increased withdrawals and motivating them to embrace new cultivation patterns.

In Equations (2)–(5), the index i represents the number of the agricultural agent, the index y represents the year, represents the annual profit of agent i, represents the cultivated area of agent i, N represents the total number of agents, B represents the number of years in the long-term period (assumed to be five), Y represents the total number of years in the period, represents the allocated water in terms of (MCM) to agent i, and represents the amount of water requirement of agent i.

Now, in the final step, based on the calculated FarmerSub for each farmer and considering the water requirements of six agricultural products in the region, namely alfalfa, tomato, sugar beet, potatoes, wheat, and barley, which are 6,850, 6,660, 6,500, 5,750, 2,730, and 1,990 cubic meters per hectare, respectively, one can choose a corresponding cropping pattern. The method of selecting the cropping pattern based on the FarmerSub is as follows: the numerical range of possible values is divided into seven intervals according to the water requirements of the crops. Depending on the interval in which the FarmerSub falls, the percentage of cultivation for each agricultural product is determined. The lower the estimated value of FarmerSub, the pattern tends toward using more barley and less alfalfa. Conversely, the higher the estimated value of FarmerSub, the pattern tends toward using more alfalfa and less barley. In average FarmerSub values, the cropping pattern leans toward using more wheat.

This research is segmented into two parts. The initial section concentrates on the calibration and validation analysis of the proposed ABM and WEAP model, with the goal of adjusting parameters and validating the model. The subsequent section encompasses the creation of five distinct scenarios to assess the effects of socio-economic changes on water resources.

The initial step in validating the proposed ABM-WEAP model was to ensure its effective performance in simulating the hydrological conditions of the Zarrineh-Roud Basin for the period from 1983 to 2021. To achieve this, the risk-taking levels of agricultural agents (L1, L2, and L3) in the ABM were adjusted to assist their decision-making regarding changes in cropping patterns for the upcoming irrigation season (Table 2).

This verification process employed a trial-and-error method (Pouladi et al. 2019) and involved calculating the Nash–Sutcliffe (NS) and root mean square error (RMSE) coefficients for annual flow at the Nezam Abad hydrometric station. Runoff calibration and validation were carried out at the Nezam Abad station, which serves as the basin's exit point, using monthly time steps for the entire period.

Comparable NS and R² values have been reported in similar studies assessing hydrological model performance (e.g., Du et al. 2017; Aghazadeh et al. 2024; Javan Salehi & Shourian 2024), indicating that the ABM-WEAP model accurately represents observed streamflow. This high level of accuracy supports the model's acceptability for further analysis and decision-making processes related to water resource management in the Zarrineh-Roud Basin.

A government entity is responsible for overseeing and managing water resource utilization, with the goal of guiding farmers toward cultivating crops with lower water demands. This involves employing various strategies across different scenarios. One effective strategy is ‘education and increasing farmers’ awareness’ about authorized water usage and groundwater preservation. By conducting educational programs, the government can inform farmers about the consequences of depleting water resources and encourage adherence to water extraction limits in the Zarrineh-Roud Plain. Mathematically, this approach helps align farmers' water usage with permissible limits.

Another strategy is ‘alternative employment,’ which involves creating platforms and providing training for farmers to start home-based businesses like handicrafts. This reduces their reliance on agricultural income, allowing them to shift from high-profit, high-water demand crops such as cotton to less profitable, less water-intensive crops like wheat and barley. For this study, it is assumed that a 20% reduction in farmers' economic dependence on agriculture will result in a corresponding 20% reduction in water usage, making them more willing to adjust their cropping patterns without significantly impacting overall profitability due to supplementary non-agricultural income.

The ‘water pricing’ approach aims to discourage the cultivation of high-water demand crops by reducing their profitability. By implementing water purchase costs based on guidelines from the Water Resource Pricing Council of the Ministry of Energy and relevant time-series data, this strategy decreases the attractiveness of high-water-consuming crops, encouraging farmers to switch to less water-intensive alternatives.

In this study, the proposed ABM-WEAP model is used to create and analyze five scenarios (SC i), which combine the aforementioned strategies. These scenarios, detailed in Table 3, are designed to comprehensively assess the socio-economic impacts on farmers' activities and the condition of water resources within the studied area.

Table 4 provides a comparison of water deficit and agents' profits across various scenarios. Based on the results obtained from Table 4, scenarios SC2, SC4, and SC5 have demonstrated the most efficient water consumption performance. However, scenario SC5 exhibited suboptimal performance in terms of profitability due to the expenses incurred by farmers for water purchases.

Regarding profitability, although the profitability of factors in different years has been more influenced by weather conditions, especially precipitation levels, scenarios SC1, SC3, and SC4 have exhibited favorable performance. Notably, SC4 has demonstrated suitability not only in water consumption efficiency but also in cost-effective water utilization. It is essential to mention that SC4 is financially demanding for governmental entities and requires investment.

Another noteworthy result is the substantial influence of constant parameters, particularly the risk-taking levels of factors, on the performance of agents. The risk-taking levels, influenced by age and the degree of economic dependence on agriculture, determine the response of agents and their coping mechanisms in various circumstances.

Considering the cumulative results, the following summarizations can be made:

• 1. Scenario SC4, which integrates ‘Education and Environmental Awareness’ and ‘Alternative Employment,’ demonstrates the most significant improvements in hydrological–social conditions when the governmental entity has the necessary financial budget for investment. This scenario ensures that farmers are well-informed and financially supported to make sustainable choices, leading to reduced water usage and improved environmental conditions.

• 2. Alternatively, Scenario SC5, which incorporates ‘Water Pricing,’ ‘Alternative Employment,’ and ‘Education and Environmental Awareness,’ is effective when the governmental entity lacks the financial budget for extensive investments. This scenario leverages pricing mechanisms alongside educational and alternative employment strategies to achieve sustainable water usage.

So, both SC4 and SC5 show potential to meet sustainable inflow targets by combining education, economic incentives, and regulatory measures. SC4 is optimal with sufficient financial resources, while SC5 provides a viable solution under budget constraints. These scenarios highlight the importance of a multifaceted approach to water resource management, ensuring that it improves environmental conditions while supporting the socio-economic well-being of farmers.

This study models socio-hydrological systems of water resources by combining surface water simulation models with behavioral simulations of water resource users. A primary objective is to assess the effectiveness of governmental strategies in controlling water extraction by farmers while preserving their economic interests. The watershed was simulated using the WEAP model, which was validated for accuracy. An ABM was then developed for stakeholders in the water system, including a governmental entity and 12 farmer agents capable of independent decision-making and altering their cropping patterns.

By integrating the ABM with the hydraulic model and creating the ABM-WEAP model, three strategies – ‘Water Pricing,’ ‘Alternative Employment,’ and ‘Education and Environmental Awareness’ – were evaluated across five different scenarios.

The results indicate that if the governmental entity has the necessary financial budget, the simultaneous implementation of ‘Education and Environmental Awareness’ and ‘Alternative Employment’ (scenario SC4) can significantly improve the region's hydrological-social conditions. Conversely, if the budget is lacking, the concurrent execution of all three approaches (scenario SC5) is the most effective.

This analysis highlights the importance of financial capability and the concurrent implementation of multiple strategies to address the complex dynamics of socio-hydrological systems. The findings offer valuable insights for policymakers and stakeholders in water resource management, presenting nuanced approaches for enhancing sustainability and socio-economic well-being in the studied region.

No funds, grants, or other support were received.

This research does not contain any studies with human participants or animals performed by any of the authors.

Data collection, models preparation, execution, and analysis were performed by Amirreza Mosavi, checked by Maryam Javan Salehi and supervised by Mojtaba Shourian. Saeed Lotfi helped in data preparation. The first draft of the manuscript was written by Amirreza Mosavi and edited by Maryam Javan Salehi. Mojtaba Shourian read and approved the final version of the manuscript.

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

The authors declare there is no conflict.