Impact of climate change on droughts: a case study of the Zard River Basin in Iran

Climate change increases the likelihood of worse droughts occurring in many parts of the world in the next decades. There are plenty of ways climate change may contribute to aridity. Changing weather can also alter atmospheric rivers (narrow streams of moisture transported in the atmosphere), which can especially disrupt precipitation patterns in the world (https://www.c2es.org/). A combination of shifting atmospheric rivers and warmer temperatures can also affect snowpacks and melt, potentially decimating the water supplies (Jenkins & Warren 2015; Lee et al. 2017). The estimates of future changes in periodic or yearly rainfall in a specific location are less certain than the estimates of future warming. However, at the global scale, researchers are confident that quite wet locations, such as the tropics and higher latitudes will get wetter, while relatively dry regions in the subtropics will become drier (Parry et al. 2004; Nassiri et al. 2006; Moradi et al. 2013; Ngo et al. 2015; Samuel et al. 2016).

Of all extreme weather types, droughts have one of the greatest influences on civilization and are an economically vital hazard to the agriculture and water sectors in many countries (Jenkins & Warren 2015). Therefore, any changes to hydrological processes will pose an important risk to society. Evidence shows that droughts have been growing in frequency and intensity in many regions over the 20th century in response to global climate change (Jenkins & Warren 2015; Lee et al. 2017). Projections also show that droughts will likely be further exacerbated in certain regions under future climate change, including Iran (Nassiri et al. 2006; Moradi et al. 2013; Pirttioja et al. 2019).

Many statistical indices have been developed in the areas of climatology, hydrology, and agricultural assessments to characterize drought. Drought researchers often focus on changes in just one or two characteristics of drought such as their frequency or duration (Jenkins & Warren 2015; Lee et al. 2017; Eini et al. 2023). However, the most popular drought index to evaluate and quantify droughts is the Standardized Precipitation Index (SPI), developed by McKee et al. (1993). The most important advantage of the SPI is that it can be determined for different time periods to allow the dynamics of different types of droughts to be assessed. The SPI is relatively simple to calculate, as it only requires precipitation data, and it has been shown that simple indexes such as the SPI can perform better than other, often more complex indexes.

Since the impacts of climate change can be diverse across a region, different spatial and temporal distributions are created for water resource components. Furthermore, studies show that the variation in precipitation patterns plays a vital role in streamflow in various regions across the world (Nassiri et al. 2006; Moradi et al. 2013; Ngo et al. 2015; Serur & Sarma 2017; Zhang et al. 2017; Pirttioja et al. 2019).

To estimate climate conditions in the future, the General Circulation Models (GCMs) and Regional Climate Models (RCMs) which consider various emission scenarios have been usually employed (Parry et al. 2004; Vaighan et al. 2017; Wang et al. 2017; Reshmidevi et al. 2018; Pirttioja et al. 2019). Different researchers have focused on the Representative Concentration Pathway (RCP) scenarios described in the Fifth Assessment Report (AR5). Comparing CMIP5 (Coupled Model Intercomparing Project Phase 5) with CMIP3 reveals that the CMIP5 encompasses comprehensive models not only based on greenhouse gas concentrations and emissions pathways but also capable of assessing impacts caused by land-use changes. Samavati et al. (2023) studied the effects of climate change on future hydrological drought in mountainous basins in Iran, utilizing SWAT (Soil and Water Assessment Tool) and CMIP5. They found that the basin's future climate conditions, such as an increase in temperature and decrease in rainfall were appropriate. According to Miroc5 (RCP8.5), the annual runoff in the future period (2020–2040) would decrease by 8.36% compared to the past. Ougahi et al. (2022) investigated the application of the SWAT model to assess climate and land use/land cover (LULC) change impacts on water balance components of the Kabul River Basin in Afghanistan. The results demonstrated that all GCMs projected an increase in temperature over the Kabul River basin, and an average annual water yield was projected to increase under the agriculture-dominant scenario. Gupta et al. (2021) employed CMIP5 and SWAT to investigate the climate change impacts on the hydro-climatology of the Subansiri river basin in India. The results showed an increase in the annual average maximum temperature (T_max), annual average minimum temperature (T_min), annual precipitation of the river basin, and also, an increase in the discharge for a particular percent of the dependable flow in the case of all the RCP scenarios. Delavar et al. (2022) concluded that SWAT is able to represent surface runoff processes in the semi-dry basin (Karkheh River Basin) in Iran and that the model is suitable for runoff, land, and water management studies. Similarly, Eini et al. (2020) showed that the SWAT model was reliable to simulate runoff in the Dagu River basin located in China with R² > 0.7 and NSE > 0.6.

In keeping with other water-limited areas of the Middle East, water security is of foremost concern in Iran. The results of this study are intended to provide reliable guidelines for policymakers and water resource authorities to ensure sustained water accessibility under a changing climate and increased droughts. The Zard River Basin is one of those areas that provide the drinking water for over 1 million people in Khuzestan province and also, the water for agriculture and animal husbandry that affects the whole country (Khuzestan province produces 20% of the country's food resources) which shows the crucial role of rivers (including Zard River Basin) in water security of Iran. As climate change tends to affect arid and semi-arid areas more dramatically, Zard River as a semi-arid basin is a good choice for investigating the impact of climate change in the mentioned areas. The objectives of this study include the following:

• Comparing past and near-future temperatures and precipitation using different GCMs.

• Calibrating a hydrological model and investigating runoff change in the future compared to the past.

• Meteorologic and hydrological drought assessment in the past and near-future.

The river constitutes a major branch of the Allah River situated within the Baghmalek (Janaki) region in Izeh (a town) with a highly dense network of rivers. Bulavan or Abul Abbas is known as the initial and main branch of the river, which emanates from the slopes to the east of Mongasht Mountain and Sefid Kooh, and then flows through narrow valleys within a mountainous region toward the northwest. After crossing a valley called by the same title located in Tang-e-Kure, the flow orientation of the river is changed toward the southwest. The river also acts as the main resource of water supply for the villages located on its route, such as Robat-e-Abul Abbas, Mal Agha, Sang, and Zolab. Afterwards, the river enters Baghmalek and joins Ab Galal, Ab Mangian, Dom Dali, and Al-e-Khorshid rivers which form the Zard River. Then, the Zard River streams toward the southwest, where it becomes the main source of the water supply for the Rud Zard Sadat, Rud Zard Kafi, Jare, and Karim villages. By joining the Aala River in the vicinity of Rud Zard village, it forms the Allah River. The studied area is mainly composed of woodlands and forests (14%), urban areas and agricultural lands (32%), and rangelands (48%). Given the strategic importance of the Zard River Basin, a new dam called the Jare dam was founded on the river in 2012. We employed the data pertaining to two periods, i.e., the future period (2022–2041) and the past period (1979–2009) in order to monitor the drought. At the same time, we analyzed the data obtained from 13 rain gauge stations. The stations and their specifications are presented in Table 1 in more detail, in which the historical period data (i.e., runoff, precipitation, evaporation, and temperature data) have been acquired from the water utility company of Khuzestan (a state-owned enterprise) and the Iranian organization of meteorology (also a state-owned enterprise). In addition, land use-land coverage, the maps of digital elevation, and the soil map were acquired from Khuzestan's water utility company.

The annual precipitation within the drainage basin under the study varies between 402 and 792 mm for various stations with an approximate estimated average of 585 mm. Table 2 shows the climate classification of the Zard River Basin based on the De Martonne method.

Drying lakes and rivers, declining groundwater resources, land-use change, water supply scarcity due to poor infrastructure and disruptions, forced migration, agricultural losses, sand storms, ecosystem damages by constructing dams, improper management of water resources, old farming methods, and more pressing matter of temperature rising due to climate change and its impacts on the rivers are the water-related issues of Khuzestan province and also Zard River Basin that needs to get more attention.

The effect of climate change on hydrological processes relies on the climate model's predicted future climatic scenarios. The present study utilized 11 GCM CMIP5 for the RCP85 and RCP45 emission scenarios as input to assess future climate. In several studies, the present environment and future changes in the context of various greenhouse gas and aerosol scenarios were simulated by climate modeling. With many stronger GCMs for climate change predictions, uncertainty about future climate projections persists. This challenge can be addressed by using a multi-model ensemble of GCMs to get a range of potential future outcomes. In this study, 22 climatic scenarios from 11 CMIP5 GCMs were generated. Being defined in almost a 150–300 km coarse grid, these climate forecasts cannot be applied to the hydrological model as input to evaluate the effects of climate change. Hence, researchers use various dynamical or statistical downscaling methods to achieve a higher spatiotemporal resolution needed for hydrological uses. LARS-WG is a stochastic weather generator developed by Barrow and Semenov to downscale and generate daily time series of climatic variables, including temperature, solar radiation, and precipitation.

By utilizing the observed climatic data for a baseline period, this model calculates the variations in climatic parameters. The output of the weather generator was used in this analysis as a reference to the SWAT model for calibration to determine changes in monthly runoff in the foreseeable future (2022–2041). The CMIP5 GCM data set were then retrieved from the 11 climate simulations, including the weather data(precipitation, minimal and maximum temperature) for the baseline period (1979–2009) and foreseeable future (2022–2041), covering the field of research (the CMIP5 models were obtained from ‘http://climate-scenarios.canada.ca/index.php?page=gridded-data’). These 11 models (Table 3) were selected and ranked based on the correlation between baseline data of GCMs (monthly precipitation and temperatures) and historical data from weather stations. This selection and ranking were also confirmed by another study (Zamani & Berndtsson 2019).

⁠, ⁠, and denote changes in rainfall and minimum/maximum temperature for the month i (January to December). (⁠⁠, and ⁠), (⁠⁠, and ⁠)/(⁠ and ⁠) represent the long-term average (LTA) of the month i for rainfall and minimum/maximum temperature for a historical (present) and a future period, respectively.

In the above equation, SW₀ and SW_t represent the initial and final soil moisture contents for the ith day, respectively. R_day indicates the rainfall reaching the soil surface. Q_surf, E_a, w_seep, and Q_gw denote the surface runoff, evapotranspiration, interflow, and base flow, respectively.

Arc SWAT interface 2012 was employed as a tool for extension in Arc GIS 10.4 to set up and determine the parameters of the hydrological model. To do so, first, the watershed delineation tool was employed to divide the basin into several sub-basins and as hydrological response units (HRUs) to the unique characteristics of the slope, soil, and land use. This was done with the addition of layers of soil and land-use maps with a definite range of slope values. DEM-derived stream network maps were implemented in the SWAT model along with river discharge sites. Calibration was facilitated by applying some additional adjustments, including the observed outlet and the river network. Moreover, soil and land-use maps attached to the SWAT database were reclassified using lookup tables.

The source used to derive drought indices was short-term time series of runoff, precipitation, and river flows to provide a perceptible sizable sample. We have presented the above-cited indicators only numerically in order to use the raw data better so that they become perceptible and enhance the planners' and designers' decision-making capabilities. The acquired data of such indicators can be used to determine the drought characteristics, such as its severity, frequency, and duration. Although none of such indicators are preferable over the others, a number of them are preferable for specific applications (Barua et al. 2011). This investigation employed the Streamflow Drought Index (SDI) and SPI in order to interpret the hydrological and meteorological droughts.

If the value of standard precipitation maintains its negative amounts and equals −1 or less, a drought takes place, while positive values are rendered as drought termination (Table 4). The total negative values can be employed for the analysis of the properties of a standard drought (magnitude, intensity, and duration).

and stand for the standard deviation of the cumulative volume of flow and the mean total volume flow rate, respectively, for the base period k for a long duration of time. A number of drought states in the SDI approach have been presented in Table 5 (Nalbantis 2008).

The Zard River Basin was split into 29 sub-basins by precise flow routing using DEM data. These sub-basins were then subdivided into 319 HRUs, according to slope, soil, and land-use maps in Arc SWAT 2012. Meteorological station data such as wind speed, minimum/maximum temperature, and daily precipitation, were then fed into the model. The Penman–Monteith combination method was utilized to calculate potential evapotranspiration (Me et al. 2015). The SWAT model was calibrated on a monthly time step with the selection of 1979–81 as the warm-up period. The 1982–2001 and 2002–2009 were regarded for the calibration and validation periods, respectively.

Several effective, highly spatiotemporally variable parameters are central to hydrological models when converting input data into output variables. The hydrological cycle and the performance of the model in representing natural conditions as accurately as possible can be controlled by the above parameters.

Sensitivity analysis is performed to reduce the number of parameters during calibration, which can be attributed to a lack of observed data. To do this, insensitive parameters were removed from the model, which allowed the sensitive parameters to be fitted efficiently (Srinivas & Gopal 2017). To this end, uncertainty and sensitivity analysis, validation and calibration, and parameterization of hydrological variables were performed using the SWAT Calibration Uncertainty Procedure (SWAT-CUP) (Abbaspour 2015). The sequential Uncertainty Fitting 2 (SUFI-2) technique was used in this study efficiently for calibrating the model. Therefore, this method was utilized to achieve the above objectives, namely calibration and validation and sensitivity analysis of parameters (Goyal et al. 2018). For this purpose, daily data of discharge for the period 1979–2009 at the Mashin station, which is situated at the Zard River Basin outlet, were used for warmup, calibration, and validation process, respectively.

Tables 8 and 9 and Figures 7 and 8 indicate that the longest drought events may not change much in the future compared to the historical period but the severity of droughts will increase especially under the RCP85 scenario. Also, the severity of droughts decreases moving from SPI/SDI-3 to SPI/SDI-12. Furthermore, severe and extreme droughts happen in the future that have never happened in the past. The longest drought events in 6- and 9-month periods are similar because the precipitation is very low in the summer. Although the inputs of SPI and SDI indexes are different (precipitation and runoff) we can see a correlation between them. Based on Tables 8 and 9, there is a good correlation between SPI and SDI indexes, for example, SPI-12 and SDI-12 show the same longest drought duration in the past and future. When comparing the hydrological drought in future and past periods, the severity of droughts is shifting from mild drought to moderate, severe, and extreme droughts, which may be an indication of more change in the behavior of future droughts compared to the past.

In this study, SWAT and the new LARS-WG6 stochastic weather generator were used to generate climate scenarios to assess the impacts of climate change. Uncertainty is a pivotal component of GCMs such that natural variability and coarse resolutions can affect them. Thus, the spatial downscaling for the conformity between the GCM output and the conditions of the study area is of particular importance. Hence, this study can be a footstep for other researchers to evaluate and compare the outputs of this generator to other weather generators. The results indicated that the average temperature will increase in the future period especially under the RCP85 scenario (more than 2° in summer and fall) and rainfall will decrease with a maximum of 6 and 10% under the RCP45 and RCP85 scenarios, respectively. These results are also confirmed by other studies (Vaghefi et al. 2019; Zakizadeh et al. 2021). Our results are inconsistent with Lotfirad et al. (2021). In this study, they used 23 GCMs as input for the LARS-WG and weather generators to project climate change scenarios. The results of our study are inconsistent with Dongwoo's study (2018), in which the author concluded that the average rainfall would not change in the future under the RCP85 scenario (this scenario was only investigated in the study).

In this research, SPI and SDI indices were used to evaluate the meteorological and hydrological droughts, respectively. The results show that severe and extreme droughts increase in the future. The results of this study are in agreement with the study by Noorisameleh et al. (2020). They concluded that extreme droughts would increase especially under the RCP85 scenario. Meanwhile, severe and moderate droughts will increase in the future period compared to the past period, and it is in agreement with our research. The hydrological drought assessment in our study shows that mild droughts will occur more than other droughts. The future projections show more intense changes in drought conditions. Also, there is a strong relation between SPI and SDI indices especially on a 12-month scale which is confirmed by other researchers (Tabari et al. 2012; Boudad et al. 2018).

These intense changes in metero-hydrological conditions can have profound effects on water resources. Hence, adaption strategies are a way to face these effects. To address the challenges of climate change, Iran acceded to the United Nations Framework Convention on Climate Change (UNFCCC 1992) in 1996, the United Nations Convention to Combat Desertification (UNCCD 1994) in 1997, and the Kyoto Protocol (1997) in 2995. It has also integrated climate change issues into its economic and development plans, including Iran's 2025 Development Vision, National Action Programs (NAPs), and the Fifth National Development Plan under UNCCD. The government has also adopted several measures and policies regarding climate change. These include compliance programs, active regional and international participation, the adoption of technologies needed to adapt to climate change, the development of greenhouse gas emission reduction policies, and capacity building to deploy climate change monitoring systems. Unfortunately, none of these policies and programs has been implemented yet (von Spannenberg & Hennum 2012). In addition, Iran is very vulnerable to climate change. However, agricultural adaptation programs have been neglected. Climate change has significant effects on the agricultural sector. Therefore, several adaptation strategies should be designed in this regard. Of these strategies and programs designed, those that ensure improved production under drought conditions can withstand the devastating effects of climate change. For farmers to cope with these devastating effects, the government intends to devise a national action plan and strategy to prepare, manage, and reduce drought. The following is a description of Iran's macro-level drought management strategies.

Most popular strategies:

• Conserving the Earth's water resources (Karimi et al. 2018; Morid et al. 2012; Keshavarz & Karami 2013; Moradi et al. 2013; Mahdavi et al. 2021b).

• Certain drought-tolerant varieties must be introduced (Keshavarz & Karami 2013; Ministry of Jihad-Keshavarzi 2013).

• The systems that enable us to use water efficiently must be promoted (NRIDR 2005; NCCO 2010; Morid et al. 2012; WAVA 2015).

• Some emergency response systems must be developed (Keshavarz & Karami 2013; Ministry of Jihad-Keshavarzi 2013).

• Sufficient drought assistance funds must be allocated to those stricken by drought (Karimi et al. 2018; Keshavarz & Karami 2013; Ministry of Jihad-Keshavarzi 2013).

• The available irrigation schemes must be provided with technical support (Keshavarz & Karami 2013; Moradi et al. 2013).

• Farmers should be encouraged to consume less water (Keshavarz & Karami 2013; Zarrineh & Azari Najaf Abad 2014).

• The old irrigation facilities and canals must be improved and renovated (WAVA 2015).

• Some programs must be established for the improvement of crops (Moradi et al. 2013; WAVA 2015).

• Conserving the Earth's existing soil resources (Keshavarz & Karami 2013; WAVA 2015).

This study showed only some changes in droughts based on the 22 climate scenarios. Different climate scenarios may generate different outputs. Also, there are some limitations to the SPI and SDI, such as the only input for these indices are rain and runoff.

Zard River Basin drenches the plains of Ramhormoz and Baqhmalek, which is highly significant from an agricultural aspect. In this study, the output of 11 GCMs was utilized to simulate future climate change scenarios. It is very important to use compatible models to simulate these future climate scenarios and that is why these 11 models were selected and ranked based on the correlation between the baseline data of GCMs (monthly precipitation and temperatures) and historical data from weather stations. The analysis of the SPI showed that near normal droughts have higher proportions among different types of meteorological droughts. The occurrence of severe and extreme droughts in the future period is more than in the past period. Hydrological droughts in the future have similar behavior to the meteorological droughts and the intensity of the drought was predicted to be higher in the future period compared to the past period. Overall, RCP85 scenarios cause higher meteorological change. It may be the reason for more intense changes in drought conditions in the future period. Also, severe and extreme droughts will occur in the future period which has not happened in the past period. Based on the intensity of the occurrence of droughts in the future period, we recommended adaptation strategy guide for water supply and food production to meet severe droughts.

This research received no external funding.

We would like to thank the Iran Meteorological Organization and the Water Utility Company of Khuzestan province for providing the necessary data for this study.

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

The authors declare there is no conflict.