Water quantity–quality assessment in the transboundary river basin under climate change: a case study

CRU

Climatic Research Unit

DDB

Doosti Dam Basin

DEM

digital elevation model

GCM

general circulation models

HRU

Hydrological Response Units

MCM

million cubic meters

RCPs

representative concentration pathways

SCS

soil conservation service

SUFI-2

Sequential Uncertainty Fitting

SWAT

soil and water assessment tool

TN

total nitrogen

TP

total phosphorous

WEAP

water evaluation and planning

95PPU

95% prediction uncertainty

There is a critical need for a cooperative legal framework for 263 transboundary river basins of the world (Wolf 2002). Nonetheless, according to Article 21 of the United Nations 1997 Convention, combined monitoring programs and planning should be expanded to decrease non-point and point pollution sources. However, pollution source management is time-consuming and highly complicated, especially for non-point resources in border basins (Moridi 2019). Resolving international conflicts while finding sustainable solutions to conflicts over transboundary water bodies, such as rivers (Salman & Boisson de Chazournes 1998), remains a major water resource management issue. Due to the importance of transboundary river basins for riparian countries, different studies have evaluated the effect of climate change and upstream development on water demands, water quality, and potential conflicts (e.g., Al-Faraj et al. 2016; Ahmed et al. 2019; Han et al. 2019; Yang et al. 2019; Khoi et al. 2020; Shamseddin & Chaibi 2020; Badrzadeh et al. 2022; Yousefi & Moridi 2022; Aggarwal et al. 2023; Singh et al. 2023).

In the Hirmand River Basin, although Afghanistan has claimed that the river flow has decreased due to precipitation reduction, Iran has accused it of depriving its water rights (Mianabadi et al. 2020). The researchers investigated precipitation data through remote sensing and found no reduction. In addition to precipitation data, irrigation water consumption, and other components were considered in the analysis. The results represented that sustainable cooperation between the two countries could lead to environmental and economic benefits in the region. Al-Faraj et al. (2016) and Avarideh et al. (2017) addressed the irrigation efficiency of the Diyala Basin, located between Iran and Iraq. They studied irrigation demand sensitivity to improvements in sustainable agriculture in the changing climate. A rainfall-runoff model was utilized to assess irrigation water under climate change. Tarebari et al. (2018) expanded a multiple objective allocation model considering parameters associated with managing sustainable water resources to resolve stakeholder conflicts by applying non-symmetric Nash bargaining. Although the average flow rate supplied 100% of the environmental demand related to Lake Urmia, the flow rate influenced by climate changes could satisfy 97.5% of the demand. Likewise, Kaini et al. (2021) evaluated the effect of climate change on the hydrological regime of the Himalayan River Basin and available water for irrigation. The projected data for water availability were simulated by employing the soil and water assessment tool (SWAT) model, precipitation, and temperature data. Moreover, Eamen et al. (2020) applied an input–output model to examine the direct and indirect economic effects of water supply restrictions on two varied water supply restriction scenarios on the whole river basin and its sub-basins separately. Prioritizing water allocation, replacing alternative water resources, and employing technologies for water reuse could reduce 50% of the economic loss.

The Harirud River is an invaluable water body between Iran, Afghanistan, and Turkmenistan. This river plays a vital role in the residential population, economy, food security, and drinking water safety and provides about half of the domestic water demand in Mashhad City (Moridi 2019). In 2004, Iran and Turkmenistan built a joint dam called the Doosti Dam on the Harirud River to improve water management and allocation. However, Afghanistan inaugurated the Salma Dam officially in the river's upper reaches in 2016. Dam construction without any agreements or cooperation among Afghanistan, Turkmenistan, and Iran has become a source of disputes (Peterson 2013). Some studies have been performed on the water quality or quantity of the Harirud Transboundary River. Nagheeby et al. (2019) studied Afghanistan's current actions upstream of the Harirud River and its negative impact on water supplies in other countries. Moridi (2019) optimized the withdrawal from the Doosti Dam Reservoir and assessed its water quality impacts. The results showed that the dissolved oxygen can be improved if a proportion of downstream need is discharged from the bottom outlet of the reservoir. In a comprehensive study, Loodin & Warner (2022) reviewed the hydro-hegemony concept among the riparian countries (i.e., Afghanistan, Iran, and Turkmenistan) of the Harirud River Basin. They found that Iran's hegemonic power increased after the US forces' withdrawal and the re-establishment of the Taliban regime.

In this paper, an array of water quality and quantity simulation and planning models are integrated and used to support integrated water resources management and climate change adaptation measures in transboundary river basins. The suggested methodology provides a valuable decision support system tool for policy-makers, decision-makers, actors within the civil society, and anyone interested in transboundary river basin management. Although some studies have addressed the water quality or quantity of the Harirud Transboundary River, the water quality and quantity in the Doosti Dam Basin (DDB) under upstream developments, dam construction, and climate change conditions have remained debatable.

In this study, the agricultural, domestic, and environmental water demand and supply of three riparian countries are assessed using the SWAT and the water evaluation and planning (WEAP) model under climate change scenarios. Thus, the novelty of the current study is the development of a water quality and quantity simulation model in a poorly gauged basin and the projection of water quantity and quality based on water demand regarding the development of Harirud's upstream in Afghanistan and climate change.

The Doosti Dam lies on the border of Iran with 1,250 million cubic meters (MCM) of reservoir capacity, which is designed to control the water inflow of the Harirud River and provide the Mashhad and Sarakhs plain water demands. The Salma Dam is located upstream of the Harirud River with a capacity of 633 MCM in western Afghanistan. This dam generates 42-megawatt hours of electricity to supply 40,000 households and the irrigation water for 35,000–80,000 hectares of agricultural land (Moridi 2019). The water allocation system is also illustrated in Figure 1(b). This figure shows the location of the two dams as well as the different users and flow requirements downstream of these dams.

The SWAT is a continuous-time, semi-distributed, and process-based hydrological model that has been increasingly utilized to study the impact of climate, land use, and crop pattern changes on large basins (Arnold et al. 1998; Narsimlu et al. 2013; Kumar et al. 2018). This model divides the basin into several sub-basins called Hydrological Response Units (HRU) regarding soil, land use, and slope characteristics. It uses the soil conservation service (SCS) curve number and the Green-Ampt diffusion equation to calculate the rainfall-runoff volume of basins. The runoff water quality components of QUAL2E govern the nutrient routings, transformations, and kinetics in the runoff of the SWAT (Neitsch et al. 2011). In the present study, the runoff and sediment load of the Doosti Gauge Station are simulated by the SWAT. The measured data of the suspended load at the Khatun Station are presented in Table S1 in the Supplementary Material. This table includes the concentration of suspended load and the average measured flow. The Doosti Dam catchment area, most of which is in Afghanistan, lacks statistics and basic information suitable for hydrological analysis. In this research, the Harirud watershed was simulated using rainfall satellite data and hydrological simulation in the form of the SWAT model.

The calibration, validation, and uncertainty analysis of the SWAT are performed by river discharge and the Sequential Uncertainty Fitting (SUFI-2) algorithm using the SWAT-CUP software package (Abbaspour 2008). The mentioned algorithm maps all uncertainties (conceptual model, parameter, and input) on parameters and attempts to capture most of the measured data within the model with 95% prediction uncertainty (95PPU). The 95PPU is computed at 2.5 and 97.5% of the cumulative distribution of the output obtained via Latin hypercube sampling. Two bands (i.e., the 95PPU for model simulation and the band demonstrating measured data plus the error), the P-factor and R-factor, are compared to assess the goodness of fit. The P-factor varies between 0 and 1, where 1 represents a flawless model simulation. The R-factor denotes the ratio of the average width of the 95PPU band and the standard deviation of the measured variable. Eleven objective functions, including Nash–Sutcliff efficiency (NSE), R², root mean square error, and mean square error, can be considered in the model using SUFI-2 (Abbaspour et al. 2004, 2007).

The WEAP model was first proposed in 1988 and has been extensively used for climate change adaptation investigations (Yates et al. 2005). WEAP solves a water mass balance equation for each node and connection of the system in time steps. Water is allocated to meet demands based on resource preference, mass balance equation, and inflow constraints. This study employed WEAP as a water supply demand and allocation model for each process simulation, including runoff generation, groundwater recharge, or crop-water calculation. The input data of the WEAP model are derived from the SWAT while ignoring the hydropower purpose of dams.

Risk assessment is recognized as a critical parameter of sustainability in water resource management (Simonovic et al. 1997). However, evaluating the mean value and standard deviation of variables is insufficient to demonstrate risky behavior. Hence, system reliability, vulnerability, and sustainability have been employed in this study to estimate returns, duration, intention, and the other effects of system performance under adverse conditions (Zbigniew & Kindler 1995).

Eutrophication is an advanced process of enriching a reservoir or lake with nutrients, reducing water quality. The symptoms of eutrophication include the uncontrolled growth of algae, plants, and grasses next to the water body, reduced water transparency, color change to green, red, or brown, and reduced dissolved oxygen. The lakes and reservoirs are classified into oligotrophic, mesotrophic, eutrophic, and hypereutrophic categories based on nutrient concentrations. According to the Vollenweider index, eutrophication happens when TP and TN concentrations exceed 16 and 0.39 mg/l (Vollenweider & Kerekes 1980).

General circulation models (GCMs) are employed to simulate the global climate at various spatial resolutions (∼100–250 km²) and to present future climate change projections in 2100 (Iqbal et al. 2018). However, some researchers typically apply various scenarios to demonstrate future climate changes under specific assumptions (Iqbal et al. 2018). Overall, the IPCC Fifth Assessment Report presents four representative concentration pathways (RCPs) regarding climate change projections and modeling. In the current study, the climatic data of RCPs, including RCP 2.6 (S1), RCP 4.5 (S2), RCP 6.0 (S3), and RCP 8.5 (S4), were downloaded from http://2w2e.com. It should be noted that S1, S2, and S3 represent low, medium, and high concentration stabilization scenarios in which greenhouse gas emissions peak around 2040 and then decrease, respectively. In addition, S4 denotes an extremely high baseline scenario in which emissions continuously increase throughout this century. Table 1 presents climate change models.

The bias correction and spatial downscaling of GCM data are required before regional impact analyses (Chen et al. 2011). Statistical downscaling and dynamic downscaling using a regional climate model are regarded as two distinct techniques to downscale GCM data. The advantage of the statistical downscaling method is producing the data for a point instead of a grid. The point can be utilized directly in a hydrological model that relies on point data such as the SWAT. In the mentioned method, a grid cell is selected based on one of the following conditions:

• (1)

  The grid cell on top of the local meteorological station can be applied.

• (2)

  The grid cells around various stations can be interpolated to center the meteorological station in the grid.

Based on the nearest climate station to each grid point, the correction factor of each month is applied to the CRU or GCM daily data if necessary (Ashraf Vaghefi et al. 2017). In this study, five GCM models and four RCP scenarios of climate change modeling, including precipitation and temperature data, were extracted and applied to the SWAT model (a spatial resolution of 0.5° × 0.5°).

The DDB was delineated using a 90-m resolution DEM from the Shuttle Radar Topography Mission: http://www2.jpl.nasa.gov/srtm/. The soil and land cover data were obtained from FAO-UNESCO global soil map (https://www.fao.org/soils-portal/soil-survey/soil-maps-and-databases/harmonized-world-soil-database-v12/ru/ and https://www.fao.org/soils-portal/soil-survey/soil-maps-and-databases/harmonized-world-soil-database-v12/land-cover-datadescription/ru/). The slope of the DDB varied from 0° to 67° and was divided into five parts in the simulation model. The SWAT model divided the DDB into 35 sub-basins and 401 HRU.

There are several hydro-climatic stations in the Harirude River in Iran and Afghanistan. However, Afghanistan has missing and inaccurate data in this regard. Khorasan Razavi Meteorological Department has recorded climate data from 1970 to 2016, including daily precipitation, wind speed, relative humidity, and daily minimum and maximum temperature. For this reason, the CRU data station in Afghanistan (available from CRU East Anglia http://www.cru.uea.ac.uk/data) was used in the current study. Figure 1 illustrates the location of the CRU network of DDB. The intended meteorological stations were employed to provide the required weather input files for the SWAT model. The daily inflow of the Doosti Dam Station was recorded from January 1955 to December 2016. Khorasan Razavi Regional Water Authority measured TN and TP concentrations (2012–2013) in the Doosti Gauge Station. The data from the Pol-e-Khatoun Hydrometric Station were utilized to rebuild the data of the dam's pre-construction period. The observed inflow data of the Doosti Dam were used to calibrate and validate the SWAT model.

The agricultural water demand in Afghanistan is 560 MCM, while Iran and Turkmenistan need 278 MCM. The domestic demand in Iran and Turkmenistan is about 150 MCM, and the environmental demand of the basin is as follows:

• The downstream of the Salma Dam and the upstream of the Doosti Dam is 211 MCM.

• The downstream of the Doosti Dam to the Sarakhs Plain is 128 MCM (Moridi 2019).

The historical period's average cumulative annual precipitation variation varies from 210 to 393 mm/year (Figure 5). In the Hindu Kush Mountains, the rainfall is about 350–390 mm/year. It reaches its lowest level in the heart areas, about 210–247 mm/year upstream of the Doosti Dam. Under the RCP 2.6 scenario, precipitation in the eastern and upstream regions of the Salma Dam faces a 26% decrease. Under the scenarios from 2020 to 2100, the average annual cumulative precipitation trend shows that higher carbon emissions will decrease the precipitation (approximately 35%). The same pattern is observed in the western and inland areas of the Doosti Dam, having the lowest rainfall under the RCP 2.6 to RCP 8.5 scenarios. RCP 2.6 and RCP 8.5 increase precipitation by 16 and 2%, respectively. Snowmelt, precipitation reduction, and temperature increase upstream of DDB result in population growth in DDB and adverse effects on water availability in the future.

Overall, 21 parameters were identified for calibration according to the sensitivity analysis of parameters in SUFI-2. All parameters, along with their descriptions, were arranged according to their sensitivity level in Table 2. The initial and final intervals of parameters are presented in this table. The basin curve number (CN2) and baseflow parameter were the most and least sensitive model parameters (Table 2). The model was calibrated and validated by NS, R², P-factor, and R-factor (Table 3). Based on these results, the SWAT model simulated the water characteristics of the Doosti Gauge Station to acceptable levels. The performance criteria of the model demonstrate the acceptability and authenticity of the SWAT model. The NSE value of the model is more than 50% for four items. The R² is 0.65, 0.58, 0.60, and 0.70 for discharge, sediment, TN, and TP, respectively. Moreover, a comparison of the simulated and observed data for discharge, sediment, and TN are shown in Figures S1, S2, and S3, respectively, in the Supplementary Material.

The Doosti Dam was full in wet seasons (February–June) before the Salma Dam construction. However, after the construction, the stored water of the Doosti Dam decreased to 40% under RCP scenarios. As a result, climate change will have the worst effect upstream of the Salma Dam in dry seasons (July–December).

As shown in Figure 6, the Salma Dam upstream experiences a 60% decrease in precipitation under the RCP 8.5 scenario due to increased temperature and reduced snow storage in the Pashtun Mountains. The area temperature and average precipitation between dams increased, and the inflow discharge increased by about 16%.

Figure 6(a)–6(d) illustrates the vulnerability, volumetric reliability, resilience, and sustainability index based on the presence and absence of the Salma Dam under climate change scenarios. According to the agreement between Iran and Turkmenistan, both countries equally contributed to constructing and exploiting the Doosti Dam. They use the water and energy of the dam equally (Thomas & Warner 2015). The high reliability and low vulnerability of the water supply downstream of the Doosti Dam (Figure 6(a) and 6(c)) indicate its excellent design. After the Salma Dam construction, the vulnerability increased from 16.9 to 64.7%. However, resiliency, volumetric reliability, and sustainability decreased by 40–49%. Under RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5 scenarios, the volumetric reliability will be 5, 4.4, 4.1, and 2.4%, while sustainability will decrease by 4.8, 4.5, 3.9, and 2.7%, respectively. Although agriculture in Iran and Turkmenistan lacks water (Figure 6(d)), Afghanistan's agricultural water demand is supplied with a sustainability of 51.8%. The Salma Dam construction has no impact on Iran and Turkmenistan's drinking water supply, but its sustainability will decrease by 40–50% under climate change scenarios. The dam construction has reduced the inflow of the Doosti Dam from 68.5 to 27.5% for environmental demand. The environmental demand sustainability of the Salma Dam was 74.3% before dam construction, but it will reduce by about 31.7–36.2% under climate change scenarios. Climate change and upstream development of the DDB will decrease the total share of Iran and Turkmenistan from 850 to 300 MCMs, and the water supply sustainability of the inter-basin will fall to 31.1–35%. Nagheeby et al. (2019) predicted a 50 and 34% reduction in environmental flow and irrigation water for Iran and Turkmenistan after the Salma Dam construction.

In this research, the drinking water demand of Iran and Turkmenistan is proportional to the growing population. The water demand of Mashhad in 2041 will increase by 1.5 times more than the present (Thomas & Warner 2015). The findings of this study indicate that with 1.5 times increase in water demand, 60% of Mashhad drinking water (85% reliability) can be supplied until 2036. Then, the supply demand amount must be considered 238 MCMs per year with the same reliability.

Figure 7 displays the mean TN and TP load changes of the Doosti Dam Reservoir after the Salma Dam construction and under the scenarios. During dry months, the runoff is zero, no pollutant enters the reservoir, and high amounts of TN and TP are measured in the reservoir in wet months. The water storage of the Doosti Dam was reduced to zero 15 times in the last 43 years, and five times happened before the Salma Dam construction (Nagheeby et al. 2019). The TN concentration of cultivated areas and the Salma Dam has increased due to the growing fertilizers and runoff levels. The TN entering the Doosti Dam Reservoir will increase by 20 and 80% under RCP 2.6 (S1) and RCP 8.5 (S4) scenarios. However, the average TP concentration will reduce by about 12% under the RCP 8.5 (S4). After constructing the Salma Dam, the Doosti Reservoir water status will vary from mesotrophic to hypereutrophic under the RCP 8.5 scenario during the wet months. This result is consistent with the findings of Moridi (2019).

Climate change, land use changes due to population growth, and over-abstraction of water impact the stability of the transboundary river basins. Therefore, transboundary river basin management has heightened in recent years, and the effective governance and resolution of conflicts over water are only possible when it can help bridge the gap between hydrologists and policy-makers.

The Harirud River is a transboundary river that originated in Afghanistan and is socially, economically, and environmentally significant to three riparian countries. The present study evaluated the impacts of climate change and upstream development on the Harirud River inflow, water supplies, and the Doosti Dam Reservoir quality. Due to the poor-gauged area of the basin in Afghanistan, the satellite products of P-CDR, CHIRPS, CRU, PERSIANN, and ERA5 were applied as precipitation and temperature data sources. In addition, a SWAT-WEAP model was used to simulate and predict the RCPs, land use change, and Afghanistan development influence on water demands, supplies, and dam storage. The findings are as follows:

• Before the Salma Dam construction, the Doosti Dam supplied Iran and Turkmenistan demand with 90% reliability.

• Despite climate change, the construction of the Salma Dam has changed the water allocation so that 70% of river water is allocated to Afghanistan. The Doosti Dam can provide domestic water to Mashhad by 2036 if Afghanistan stops the Harirud River Development Plan.

• Under climate change scenarios, the Salma Dam construction will reduce the reliability of Iran–Turkmenistan agricultural and environmental water supplies. Climate change will increase the average temperature by 1.5–3.8°C and decrease the downstream precipitation.

• After the Salma Dam construction, the Doosti Reservoir status will change from mesotrophic to hypereutrophic under the RCP 8.5 scenario in the wet months.

Considering there is no agreement among Afghanistan, Iran, and Turkmenistan, this research and its suggested method can be utilized as an integrated modeling framework to support water resources management and climate change adaptation measures in transboundary river basins and can be used for the preliminary water allocation model for the Harirud River Basin. The proposed integrated modeling can help bridge the gap between hydrologists and policy-makers during negotiation and also shows the conditions for supplying downstream water demands. The results show that the Doosti Dam is not a reliable source to supply the domestic water demand of Mashhad City (the second most important city in Iran). Iran's government must improve the reliability of supplying demand by alternate resources and political negotiation with neighboring countries.

The primary objective of this paper was to project the future condition of the river, considering forthcoming development in Afghanistan and climate change scenarios. The results reveal the possibility of future conflicts when water bankruptcy happens in the basin. The negotiation can promote cooperation among the three countries to meet their environmental and drinking water needs. Cooperative game theory approaches, such as bankruptcy rules, can be adopted and developed to achieve this goal. Future research should find optimal strategies to obtain these goals with the highest reliability and sustainability.

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

The authors declare there is no conflict.