A scenario-based coupled SWAT-MODFLOW decision support system for advanced water resource management

Overexploitation of aquifers in recent decades has led to ever-growing depletion of groundwater resources over time, manifested in declining groundwater levels. Accompanied by climate changes, rapid population, and industrial growth, this issue has exerted pressure on available water resources (Joodavi et al. 2017). The use of numerical modeling can enhance an understanding of water resource conditions and sustainability of water resources systems (Izady et al. 2017; Joodavi et al. 2020). However, most decision-makers cannot directly use the results of the complex simulation and optimization models due to the lack of required knowledge and skills. In fact, the results of scientific research are not always available in the form required by stakeholders and decision-makers (Jacobs 2002; Giupponi et al. 2007; Van Kouwen et al. 2008). Therefore, developing decision support systems (DSSs) between the developed models and decision-makers is critically important to improve water resource decision-making (Walsh 1993; Parker et al. 2002; Liu et al. 2008).

The concept of DSSs emerged in the 1970s (e.g. Gorry & Morton 1989; Sprague & Carlson 1982), showing great potential for water resource management. The DSSs are designed to assist in the water resource decision-making process and to enhance usage and understandability of models' outputs. In addition, models incorporated within the DSS framework are often better adapted for decision-making than those models that are designed only for the technical specialist. Indeed, stakeholders in the decision-making process do not delve in the detailed theoretical background of the methodologies applied and prefer user-friendly features, allowing an integration of their judgments into the decision-making process (Walsh 1993).

In the literature, numerous DSSs exist for water resource management, in which increasingly sophisticated computerized systems integrate watershed processes operating at different spatial and temporal scales, simulation models, and decision-making approaches (e.g. Rizzoli & Young 1997; Koutsoyiannis et al. 2003; Mysiak et al. 2005; Giupponi 2007; Matthies et al. 2007; Makropoulos et al. 2008; Argent et al. 2009; Gastélum et al. 2009; Singh 2010; Volk et al. 2010; Heidari & Bozorgzadeh 2014; Babbar-Sebens et al. 2015; Kumar et al. 2015; Tian et al. 2016; Wang et al. 2016; Mohajeri & Horlemann 2017; Piemonti et al. 2017; Aliyari et al. 2018; Butchart-Kuhlmann et al. 2018; Goharian & Burian 2018; Nohara et al. 2018; Ruiz-Ortiz et al. 2019; Sarband et al. 2020). These DSSs have been developed for a variety of purposes, such as the following: Waterware (Fedra & Jamieson 1996; Jamieson & Fedra 1996), Aquatool for river basin management (Andreu et al. 1996), and Nelup, to provide economic and environmental impacts of rural land-use change at the river basin scale (Dunn et al. 1996), Floodss for flood management (Catelli et al. 1998), Dssipm for irrigation management (da Silva et al. 2001), Catchment Simulation Shell for supporting the participatory assessment and management of natural resources (Argent & Grayson 2003), WEAP for integrated water resource management and policy analysis (Yates et al. 2005), MULINO-DSS for sustainable use of water resources at catchment scale (Mysiak et al. 2005; Giupponi 2007), LADSS for land-use management at farm level (Rudner et al. 2007), E2 for water quality modeling (Argent et al. 2009), SWASAL to evaluate on-farm irrigation water management options (Singh 2010), WRESTORE (Watershed Restoration Using Spatio-Temporal Optimization of Resources), a web-based participatory planning tool which can involve a large community of stakeholders in using science-based, human-guided, interactive simulation-optimization methods for designing potential conservation practices on their landscapes (Babbar-Sebens et al. 2015), IHM3D for integrated SW–GW modeling in which virtual globe-based 3D environment was used to visualize the spatially distributed model inputs/outputs and georeferenced datasets (Tian et al. 2016), multi-criteria decision analysis framework for hydrological decision support system (Butchart-Kuhlmann et al. 2018), DSS-SMGW-01 to realize sustainable management of groundwater (Aliyari et al. 2018), AQUATOOL, with the aim of deepening the consideration of losses by evaporation of reservoirs for a better design of the basin management rules (Ruiz-Ortiz et al. 2019), and an interactive spatial DSS for the assessment of water resource allocation scenarios (Sarband et al. 2020).

Neishaboor plain is one of the important aquifers in Iran in which groundwater has significant socioeconomic importance, both as a factor of production in agriculture and as a source of drinking water (Nazarieh et al. 2018). During the past few decades, this aquifer has experienced severe groundwater depletion and overexploitation which has led to the general prohibition, since 1986, of any further development in this aquifer. Because of such emergent conditions in the Neishaboor watershed, an intensive groundwater and surface water management plan was established by the Khorasan Razavi Regional Water Authority. The plan was initiated by developing a conceptual model for groundwater and surface water in the watershed, as a first step (Izady et al. 2014), and was followed by numerical modeling of groundwater and surface water, as a second step, based on the developed conceptual models (Izady et al. 2015).

Therefore, the main objective of this study is to develop the integrated water resource management decision support system (IWRM-DSS), as a third step, to support water resource decision-makers to utilize the validated groundwater and surface water models, the outputs of the second step, to assess the long-term effects of different climate, recharge, and discharge scenarios in an attractive user-friendly environment on the Neishaboor watershed water resources. It should be noted that the existing DSSs in the literature are designed for the technical specialist. The decision-makers cannot directly use these DSSs by themselves, and they rely on the reports of the model results that are prepared by the experts. This knowledge gap is bridged by the development of the IWRM-DSS, as a first attempt to provide a scenario-based graphical user interface, in which decision-makers can assess different management scenarios using a coupled SWAT-MODFLOW model by themselves without having detailed modeling knowledge and computer skills. Moreover, the interactive scenario creation tool can engage decision-makers to investigate the status of the water resource systems and implement wise strategies to achieve water resource sustainability.

Figure 1 shows the undertaken steps and phases to develop the IWRM-DSS in a participatory approach. The yellow and blue highlighted colors present, respectively, the level of stakeholder and technical expert engagements in each step in which the ‘scoping’ and ‘IWRM-DSS scenario creation’ steps are mostly defined based on the decision-makers' concerns and viewpoints. Developing the IWRM-DSS was initiated by identifying main stakeholders and defining model purposes and objectives. The next step was developing a conceptual model for groundwater and surface water in the watershed (Izady et al. 2014) and numerical modeling of groundwater and surface water (Izady et al. 2015). As a final step, a user-friendly GUI was developed through numerous direct reciprocal discussions with stakeholders to get their feedback and concerns and to be implemented in the IWRM-DSS.

The developed IWRM-DSS consists of three different shells named modules, tools, and presentation (Figure 2). The IWRM-DSS framework was developed in Microsoft Visual Studio with C# programming language in which the tool shell has the Microsoft SQL server as the database, ArcGIS Engine DLLs for pre/postprocessing data and simulation results, and the calibrated SWAT-MODFLOW model for surface water and groundwater modeling. The modules and presentation shells are specifically designed for decision-makers/stakeholders to create/test different scenarios for the study area and view/print the scenario results in tabular or map forms. Further descriptions of the IWRM-DSS components are provided in the next section.

The Neishaboor watershed is located between 35°40′–36°39′ N latitudes and 58°17′–59°30′ E longitudes in the northeast of Iran (Figure 3). The total area is 9,158 km² and consists of 4,241 km² mountainous terrains and about 4,917 km² of plain. The area has a semi-arid to arid climate, with an average annual precipitation of 265 mm. The mean annual temperatures change from 13 °C at Bar station (in the mountainous area) to 13.8 °C at the Fedisheh station (in the plain area). The annual potential evapotranspiration is about 2,335 mm. Land use in the Neishaboor watershed is predominantly agricultural (47% of watershed), in which irrigated wheat and barley (70% from 47%), sugar beet, cotton, and alfalfa (30% from 47%) are the main crops grown in the watershed. The long-term annual groundwater abstraction through 4,003 agricultural wells over the watershed is 617 million m³. The aquifer budget shows a mean annual negative balance of 201 million m³ due to extensive extraction for agricultural purposes (Izady et al. 2015).

Groundwater and surface water are not isolated components of a hydrologic system but interact in a variety of ways (Sophocleous et al. 1999). Therefore, developing a coupled groundwater and surface water model, such as a coupled SWAT-MODFLOW model, is a key step toward more intelligent water resource management. MODFLOW is a computer program that numerically solves the three-dimensional groundwater flow equation for a porous medium using a finite-difference method (McDonald & Harbaugh 1988). SWAT (Arnold et al. 1998) is a spatially distributed watershed scale model to predict the impact of land management practices on water, sediment, and agricultural chemical yields with varying soils, land use, and management conditions over long time periods (Neitsch et al. 2009). Spatial parameterization of the SWAT model is performed by dividing the watershed into sub-watersheds based on topography. These are further subdivided into a series of hydrologic response units (HRUs) based on unique soil, land use, and slope characteristics.

SWAT has limitations to deal with groundwater flow due to its lumped reflection of groundwater, and similarly, MODFLOW is not capable of estimating groundwater recharge from precipitation and irrigation return flow (Izady et al. 2015). Therefore, coupled SWAT-MODFLOW models have been developed to tackle the mentioned limitations associated with each model (e.g. Guzman et al. 2015; Izady et al. 2015; Bailey et al. 2016). Izady et al. (2015) developed a SWAT-MODFLOW model in which the models were iteratively executed to compute spatial and temporal distributions of hydrological and hydrogeological components. The basic process of the coupling SWAT and MODFLOW models is to pass HRU-calculated deep percolation as recharge to the grid cells of MODFLOW and then pass MODFLOW-calculated groundwater recharge to the SWAT. This process was repeated to achieve satisfactory results for both surface water and groundwater models. For more details on the development of the models, refer to Izady et al. (2015). This coupled SWAT-MODFLOW model is used in the IWRM-DSS to examine the impact of different management scenarios on the groundwater and surface water resources in the Neishaboor watershed.

The core of the IWRM-DSS is the Microsoft SQL server database, combined with ArcGIS Engine DLLs, which is designed to perform collection, storage, visualization, and processing of groundwater and surface water datasets. It consists of more than thousand relational tables and DLLs, which store measured time-series data and raster/vector maps. These data and maps are daily precipitation, temperature (minimum and maximum), solar radiation, monthly runoff, agriculture wells (AWs) discharge, observation wells (OWs), groundwater recharge, annual crop yield, hydrodynamic properties, crop management, map of basins, rivers, soil, land use, digital elevation model, plain boundary, bedrock, and numerous other types of data and information such as model simulation and scenario results. The integration of all these data and information in a coupled Microsoft SQL server and ArcGIS Engine DLLs facilitates the exchange of data and information between them and simulation models (Figure 4). In fact, SWAT and MODFLOW obtain the needed data from the different groundwater or surface water database with preprocessing in the Microsoft SQL server database and ArcGIS Engine DLLs to perform scenarios and write back the output from the modeling executions to the database. The results of the scenarios were displayed in tabular or map form using ArcGIS Engine DLLs. It is worth noting that all components of the IWRM-DSS are able to use the ArcGIS Engine DLLs for entering data and representing scenario results in a comprehensive way that facilitates the understanding of the water resources system. Also, it is flexible and easily extensible for new development in future applications.

As mentioned before, a calibrated and validated coupled SWAT-MODFLOW model is used in the IWRM-DSS to predict the groundwater and surface water behaviors in the Neishaboor watershed by decision-makers under different management scenarios. In the IWRM-DSS, after defining the scenario name and prediction period by users (decision-makers), the scenario type should be selected among four scenario creation approaches, i.e. ‘climatic’, ‘recharge’, ‘discharge’, and ‘coupled’ (Figure 5).

The ‘climatic’ scenario creation tool was designed to predict the effect of changes in precipitation, temperature (minimum and maximum), and solar radiation on surface water balance components using the SWAT model (Figure 6). The change in the value of these parameters is provided in both percent and absolute modes. Also, these three parameters can be considered simultaneously or individually for the scenario period. The results of the different climate change models can also be used in the IWRM-DSS as an input.

The aim of the ‘recharge’ scenario creation tool was to evaluate the impact of artificial recharge projects to understand the impact of artificial recharge projects on groundwater resources in the study area using the MODFLOW model (Figure 7). In addition to the change in recharge value, groundwater abstraction can be optionally changed based on the decision-maker's perception.

The ‘discharge’ scenario creation tool provides a simple way to assess the influence of changes in groundwater abstraction on water resources using the coupled SWAT-MODFLOW model. Given the relation between groundwater abstraction and irrigation depth in agricultural-based watersheds, the amount of irrigation depth is automatically adjusted based on changes in abstraction (Figure 8).

The ‘coupled’ scenario creation tool is the most sophisticated scenario tool in the IWRM-DSS in which the climatic parameters, the irrigation depth, and the groundwater abstraction can be simultaneously changed. The irrigation depth and climatic parameters can be changed together or independently to execute the SWAT model. The SWAT-based groundwater recharge is automatically fed into the MODFLOW, given the MODFLOW execution after the SWAT run. At the end, MODFLOW is executed, and both groundwater and surface water results are presented in different forms (Figure 9).

One of the fascinating and practical capabilities of the developed IWRM-DSS is the presentation shell to display and extract the results of different scenarios in various ways. Figure 10 shows the tool designed to access the scenario results in different formats (raster map, table, and graph).

As mentioned earlier, the IWRM-DSS was developed in Microsoft Visual Studio with C# programming language. The written code consists of 57,822 lines, and it has 840 ‘class forms’ (Supplementary Material, Figure S1). Figure 11 shows the main class forms that are designed for the setting, database, creating scenarios, and extracting the models' results. Supplementary Material, Figure S2 displays the scripts and the class forms in the Microsoft Visual Studio related to the first page of the IWRM-DSS (Figure 4) in which the forms are exhibited in the right side of Supplementary Material, Figure S2 and the corresponding codes are shown in the left side. Moreover, the view of the entire written codes is presented in the middle of Supplementary Material, Figure S2.

The configuration of any project is set through three class forms. In detail, the paths of ‘the calibrated SWAT and MODFLOW native text files’ and ‘the GIS layers and shapefiles’ are, respectively, entered in ‘dbo.pjConfig’ and ‘pjLayer’ class forms (Supplementary Material, Figures S3 and S4). Moreover, the information related to the SWAT and MODFLOW models and their inputs is entered in the ‘dbo.Config’ class form (Supplementary Material, Figure S5). It should be highlighted that these three class forms need to be modified for the application of the IWRM-DSS to the other regions. In fact, the required files for any new project are native text files of the calibrated SWAT and MODFLOW models along with GIS layers and shapefiles for that specific region. Figure 11 illustrates the main class forms that are designed to store the surface water and groundwater database (see Figure 4 for the database). Microsoft SQL server and ArcGIS Engine DLLs are also employed to design these class forms to visualize the location of the spatial-based datasets. Supplementary Material, Figure S6 shows the scripts and the designed class forms for the database. The core of the IWRM-DSS is class forms related to the create and examine scenarios. According to Figure 11, several class forms are designed to conduct different management scenarios using the IWRM-DSS. For example, the ‘frmFunction’ is designed to acquire the required data in suitable formats from the database by SWAT and MODFLOW models to conduct the scenarios. Moreover, SWAT-based recharge is fed into the MODFLOW model using the ‘SWATToMODFLOWRCH’ class form. The scripts of this class form are shown in Supplementary Material, Figures S7–S9 and C# script in the Supplementary Material, material. In detail, data are exchanged between the SWAT and MODFLOW models using ‘SWATToMODFLOWRCH’ class form that relates SWAT subbasins to MODFLOW grid cells. Then, the SWAT-based recharge for subbasins is mapped to MODFLOW cells. It should be noted that the SWAT subbasin maps and MODFLOW cells are overlaid to determine the corresponding cells to each subbasin. More details about the exchange of data between two models are given in Izady et al. (2015). After conducting scenarios, the results are visualized in the map and tabular formats using several designed class forms (see Figure 10 for result types). Figure 11 shows the main class forms to extract data from the tested scenarios. The hydrological and hydrogeological water balance components are visualized and reported in suitable formats through these class forms. Moreover, the tested scenarios are saved in the database of the IWRM-DSS to compare different scenarios together. Supplementary Material, Figure S10 and C# script in the Supplementary material show the details of these class forms in which the hydrological water balance components such as precipitation, snowfall, snowmelt, surface runoff, and SWAT-based recharge can be visualized in the map and tabular formats.

To demonstrate the capabilities of the IWRM-DSS, the results of four tested scenarios for the study area are presented here.

The first scenario, developed using the ‘discharge’ scenario creation tool, represents the ‘no change’ case in which current groundwater abstraction is continued through 2025. Calculated hydraulic head for each time step can be visualized by groundwater-level raster or groundwater-level drawdown maps (Figure 12). Moreover, time-series plots can be used to evaluate and compare temporal trends for the calculated head at the selected OWs, or for the regional groundwater level calculated from all OWs (the composite groundwater-level hydrograph), as shown in Figure 13. The simulation results of this scenario showed remarkable groundwater-level drawdown, more than 2.5 m/year, in the eastern part of the Neishaboor aquifer (Figure 12). A total groundwater-level decline of 11 m was predicted between 2012 and 2025, equivalent to 0.92 m/year (Figure 13).

The second scenario was based on reduced groundwater extraction in order to achieve equilibrium condition. The result showed that groundwater extraction should be decreased by almost 40% to accomplish this objective (Figures 13 and 14). It was assumed that the recharge rates were constant during the prediction years. Also, discharge of all AWs was equally reduced all over the plain. The overall water table rises approximately 0.68 m at the end of the scenario period (Figure 12). Despite achieving overall balance with a slight decline in the composite groundwater levels, the groundwater level in the east and southeast part of plain continued to decline; however, the drawdown trend was milder than that of scenario 1 (Figure 13). Figure 14 shows the groundwater balance components for a selected water year, simulation period, and scenario period.

Climate change is anticipated to cause negative impacts on water resource systems through direct and long-term impacts such as intense precipitation, flooding, sea-level rise, and droughts (Valipour et al. 2021). Because of the utmost importance of climate change on water resource systems, scenario three was developed using the ‘coupled’ scenario creation tool. It includes combinations of simultaneous temperature increment and precipitation decrement to assess the impact of climate change on recharge, groundwater level, and surface runoff in the Neishaboor watershed. This scenario includes minimum and maximum temperature increments by 37 and 11%, respectively, whereas solar radiation and rainfall decrements by 2 and 21% are included under the A1B climate scenario (Taei Semiromi et al. 2015).

After conducting the scenario, the IWRM-DSS provides the scenario results in tables and map forms (Figures 15 and 16). The results indicate that the average annual potential evapotranspiration (Alizadeh et al. 2013) is increased 8.2%, while runoff and recharge rates are, respectively, reduced 35 and 63% during the scenario period (2012–2025) compared with the simulation period (2002–2012) (Figure 15). The significant decrease in groundwater recharge (Ahmadi et al. 2012, 2015) led to more declines (approximately 2.5 m) in average groundwater level at the end of the scenario period (September 2025) compared with scenario 1.

Because of the groundwater crisis in the Neishaboor plain, some artificial recharge projects have been implemented to enhance the aquifer condition. Faroub-Roman and Darroud projects are the most important projects implemented at the northeast of the study area (Figure 17). The results show that the groundwater level rises between 1 and 2 m in the OWs located at the northeast of the study area due to a significant increase in the mean annual groundwater recharge of about 6.0 million m³ (Figure 17).

The main added value of the IWRM-DSS is to empower the stakeholders/decision-makers to understand the behavior of water resource systems in the study area and provide a scientific framework for examining various scenarios to conceptualize possible future realities. The interactive scenario creation tool gives the stakeholders an opportunity to engage in the water resources management process and investigate the effects of different scenarios on the water resource system. With the help of the scenario creation tool in the IWRM-DSS, the decision-makers can obtain a shared vision for the future of the water resources system, plan for various contingencies, and agree on what scenarios to work toward the sustainable future. Although the decision-makers posit that the use of the IWRM-DSS helped to frame the watershed's water resources challenges and possible solutions, some limitations in the IWRM-DSS (e.g. the impact of land-use changes and socioeconomic factors) should be included in the coming updated versions.

The developed IWRM-DSS has been successfully employed by the decision-makers to evaluate different management scenarios. As real practices, the decision-makers found, based on scenario 2, that groundwater extraction should be decreased by almost 40% to accomplish the equilibrium condition. They examined several scenarios to reduce the AW's discharge in different parts of the watershed and negotiate with the farmers to decrease the discharge of AWs through either incentive programs or changing the crop patterns with less water requirement crops. The latter is tested using the ‘coupled’ scenario in which the irrigation depth can be easily changed based on considered crop patterns. Moreover, they obtained a better vision of managed aquifer recharge through scenario 4 in which they examined the potential regions over the watershed for the artificial aquifer recharge and secure the sustainability of the water resources. The important point to mention here is that these findings are achieved by decision-makers through testing different scenarios merely by themselves. Indeed, the variety of management scenarios can help managers to know the watershed better and to make wise decisions.

The successful experience of the IWRM-DSS in Iran is a good indication of its potential for sustainable planning and management of water resources for areas with stressed aquifers. In fact, the developed IWRM-DSS is watershed-independent, and therefore, native SWAT and MODFLOW text files of other watersheds can be incorporated into the IWRM-DSS.

As global populations and economies continue to increase and environmental changes threatening the quantity and quality of water resources, significant pressure is placed on water managers to make major decisions to keep the sustainability of water resources. Modeling allows the prediction of an expected future state and, therefore, is an important part of informed decision-making in basin-scale management of water resources systems. However, most stakeholders and decision-makers cannot directly use the results of complex simulation and optimization models due to the lack of required knowledge and skills. Therefore, an IWRM-DSS was developed in an attractive user-friendly environment for the Neishaboor watershed, Iran. The IWRM-DSS is a new interface for the coupled SWAT-MODFLOW models to perform four climatic, recharge, discharge, and coupled scenarios to understand the future condition of the water resources in the Neishaboor watershed. The current IWRM-DSS is designed as desktop software that could be installed on normal PCs and laptops. With the ongoing advances in World Wide Web technologies, developing web-based DSS is one of the main tasks of future developments.

The authors acknowledge the HydroTech Toos Consulting Engineers Company, Mashhad, Iran for the financial support under grant number #40202005 and the Khorasan Razavi Regional Water Authority (KRRWA), Mashhad, Iran for providing the data and information. The contributions of Azizallah Izady were supported by the Sultan Qaboos University under grant number #IG/DVC/WRC/20/01. The authors also extend appreciation to the research group DR/RG/17.

All relevant data are included in the paper or its Supplementary Information.