Comparative evaluation of impacts of climate change and droughts on river flow vulnerability in Iran

Zhr Noorismeleh ,Shhrir Khledi *,Alirez Shki Prviz Zeien Firouzdi ,Willim A.Gough ,M.Monirul Qder Mirz

a Faculty of Earth Sciences,Shahid Beheshti University,Tehran 1983969411,Iran

b Department of Physical and Environmental Sciences,University of Toronto Scarborough,Toronto M1C 1A4,Canada

c Faculty of Geographical Sciences,Kharazmi University,Tehran 15719-14911,Iran

Received 28 December 2019;accepted 30 May 2020 Available online 13 November 2020

Abstract Rivers in arid and semi-arid regions are threatened by droughts and climate change.This study focused on a comparative evaluation of the impacts of climate change and droughts on the vulnerability of river flows in three basins with diverse climates in Iran.The standardized precipitation-evapotranspiration index(SPEI)and precipitation effectiveness variables(PEVs)extracted from the conjunctive precipitation effectiveness index(CPEI)were used to analyze the drought severity.To investigate hydrological droughts in the basins,the normalized difference water index(NDWI)and the streamflow drought index(SDI)were calculated and compared.The effects of droughts were assessed under various representative concentration pathway(RCP)scenarios.Changes in the number of wet days and precipitation depth restricted hydrological droughts,whereas an increasing number of dry days amplified their severity.The projected increases in dry days and precipitation over short durations throughout a year under future climate scenarios would produce changes in drought and flood periods and ultimately impact the frequency and severity of hydrological droughts.Under RCP 4.5,an increase in the frequencies of moderate and severe meteorological/hydrological droughts would further affect the Central Desert Basin.Under RCPs 2.6 and 8.5,the frequencies of severe and extreme droughts would increase,but the drought area would be smaller than that under RCP 4.5,demonstrating less severe drought conditions.Due to the shallow depths of most rivers,SDI was found to be more feasible than NDWI in detecting hydrological droughts.? 2020 Hohai University.Production and hosting by Elsevier B.V.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords:Climate change;River flow;Precipitation;Drought index;Standardized precipitation-evapotranspiration index;Iran

1.Introduction

Sustainable development of water resources is a fundamental consideration for urban planners(Hurlimann and Wilson,2018).A sustainable water supply is one of the key requirements for the survival and development of human societies(Katko and Hukka,2015),and water supply sustainability requires research-informed economic,environmental,and social management policies(Rathnayaka et al.,2016).

In the coming decades,water management will face major social,environmental,and technical challenges associated with current global threats,such as changes in water resources,population increase,urbanization,resource degradation,and climate change(Butler et al.,2016).The effects of climate change on water accessibility,water demand,and weather conditions,in combination with the direct pressures created by population increase,land-use change,pollution,and inappropriate water management practices,are likely to reduce surface and underground water resources and enhance competition for water(Bates et al.,2008).

Drought hazards are caused by reduced precipitation and available water,and they can occur in all regions with varying frequency and intensity(Pathak and Dodamani,2016).Although droughts are random and unpredictable phenomena,they are persistent hazards when they occur(Alizadeh,2015).The most damaging droughts occur in arid and semi-arid regions where less water is accessible,even under normal climate conditions.A long-duration meteorological drought event(lack of precipitation)can lead to a hydrological drought event when the level of surface water and groundwater reservoirs drops(Guo et al.,2020).Meanwhile,the intensification of meteorological droughts and subsequent hydrological droughts makes the affected regions more susceptible to the increased frequency of dust,in particular,in arid and semi-arid regions such as Iran(Boroghani et al.,2019).

In recent years,there has been worldwide concern that climate change may increase the severity and duration of droughts(Wilhite et al.,2014).According to a scientific consensus,if global warming is limited to just 1.5°C,water stress will increase by approximately 50%,with certain regional variations(IPCC,2018).The risk of water scarcity induced by climate change is exacerbated in arid and semi-arid regions because of the compound effects of local droughts(Kahil et al.,2015).The projected climate change would make Iran very vulnerable to droughts(World Bank Group,2016).The importance of water resources accessibility has prompted researchers to investigate the likelihood that climate change impacts water resources(Shamir et al.,2015;Shrestha et al.,2017;Ma et al.,2018).

Climate change impacts on the quantity and quality of water resources are generally investigated through hydrological modeling,which includes the use of long-term observational data and climate simulations under various scenarios(Mohammed et al.,2019).The effect of climate change on river flows around the world has been thoroughly studied(Palmer et al.,2008;Wang et al.,2019).Based on previous research,the effects of climate change on water resources on regional scales are significant(Blanco-G'omez et al.,2019;Nerantzaki and Nikolaidis,2020).

Based on the long-term predictions of consecutive dry days(CDD)and soil moisture anomaly(SMA)indices,Iran would undergo pronounced climate change impacts,particularly in terms of droughts(IPCC,2012).A study of the socio-economic impacts of droughts in Iran shows that the country has reached the state of water bankruptcy,in which water demands exceed the available natural water supply(Madani et al.,2016).

As indicated by long-term historical climate records,64.2%of the Iranian region has been affected by moderate,severe,and extreme droughts.In some years Iran experienced higher-thanaverage precipitation,but long-term drought conditions persisted(CRI,2020).The economic damage induced by droughts in Iran is substantial.A 1-mm reduction of precipitation potentially causes a loss of 98 billion Rials.As rural areas are largely dependent on agriculture,degradation of agricultural land caused by water shortage has resulted in an increase in rural-to-urban migration(Noubakht et al.,2018).A report from the agricultural insurance fund demonstrated that a drought event in 2015 in Kohgiluyeh and Boyer-Ahmad Province of Iran caused a reduction in the wheat yield of 50000 hm2of agricultural land,while the average precipitation in the province was only 10 mm below the average(AIF,2015).Cook et al.(2014)conducted climate-change impact simulations in Iran and found an increasing trend in drought intensity in the future,and the simulations indicated that the environmental,economic,and social impacts of droughts in Iran would likely increase.

Recent studies have shown that climate change,including changes in precipitation and temperature patterns,has caused hydrological changes in Iran(Heidari et al.,2018;Pourkarimi et al.,2018).A comparative assessment of the climate-change impact on meteorological droughts and the combined effects of climate change and droughts on river flows will help to better identify vulnerable regions and propose rational risk management strategies in adaption to these changes.Due to climatic and geographical diversity,the climate-change impact on river flows in Iran may vary in different basins.To examine these regional variabilities,this study conducted a comparative analysis on the differences of river flows in three basins with different climatic features.The standardized precipitationevapotranspiration index(SPEI)was used to identify and characterize the meteorological droughts.The conjunctive precipitation effectiveness index(CPEI)was used to determine precipitation variables that control the intensity of hydrological droughts in the basins.Given that the selected basins are located in the arid and semi-arid regions of Iran,the remote sensing-based normalized difference water index(NDWI)and the streamflow drought index(SDI)were used to identify the suitable hydrological drought index.To project future changes of river flows and identify highly vulnerable areas,future daily precipitation was simulated using the statistical downscaling model(SDSM),and the drought indices were calculated.

2.Methodology

2.1.Study area

Water resources in Iran have been divided into six regions with 30 basins in total(MEI,2018).In this study,the Haraz and Qarah Su Basin was selected as a study area.It is located in the Caspian Sea Watershed and has a humid climate.The Namak Lake Basin and the Central Desert Basin were also selected as study areas.They are located in the largest river basin in the country,namely,the Central Plateau Basin.These two basins are characterized by arid-to-semi-arid climates.The geographical locations of the three basins and the weather and hydrological stations are shown in Fig.1.

Fig.1.Geographical location of study area and weather and hydrological stations.

Each studied basin is comprised of several sub-basins.In this study,the Haraz and the Babolrood-Talar sub-basins in the Haraz and Qarah Su Basin,the Jajrood-Karaj and the Shoor sub-basins in the Namak Lake Basin,and the Hablehrood-Shoorab sub-basin in the Central Desert Basin were studied(Fig.2).Although the studied basins are not large,they are affected by different climates owing to the geographical diversity.Owing to climatic conditions influenced by the Caspian Sea and the Alborz Mountains in Northern Iran,perennial rivers appear in the Haraz and Qarah Su Basin.From west to east,the climatic conditions change from semi-arid(in Namak Lake)to arid(in the Central Desert),and river flows are low and seasonally variable.The Alborz Mountain Range,which is located in the northern part of the study area and has a cold and snowy climate,is the source region of many rivers.In terms of elevation,the Haraz sub-basin is the highest among all the studied sub-basins(Fig.1).The northern region(the Babolrood-Talar,Haraz,and northern Jajrood-Karaj sub-basins)has the highest coefficient of variation of monthly precipitation(Fig.2(a)),and the coefficient of variation of monthly river flow decreases from west to east over a longterm period(1989-2018)(Fig.2(b)).

Every year,drought threatens part of the study area with varying severity.Fig.3 demonstrates the annual average SPEI time series in the studied basins over a 30-year period(1989-2018).The drought identification based on SPEI in this period shows that the basins differ in drought severity owing to different climatic features and geographical locations(Fig.3).The drought severity in the three basins had a similar trend.In recent years,the drought severity in the Namak Lake Basin has increased more than that in the other two basins.

Fig.2.Coefficients of variations of monthly precipitation and river flows in selected basins(1989-2018).

2.2.Data

This study collected the daily precipitation and temperature data for the period of 1989-2018 at 49 weather stations in the selected basins from the Iran Meteorological Organization,and the river discharge data for the same period at 62 hydrological stations were obtained from the Iran Water Resources Management Company.The data were used to calculate drought indices and simulate the changes of river flows in drought events.To identify the regions affected by meteorological/hydrological droughts in the period of 1989-2018,SPEI and SDI were calculated on the point scale,and the average frequencies of droughts at different severity levels were computed at each point.Based on the SPEI and SDI values,spatial variability of droughts was obtained by interpolation.SDSM 5.2(http://www.sdsm.org.uk)was used to generate future daily temperature and precipitation data on the regional scale.SDSM simulated daily data based on the observation data and reanalyzed dataset from the National Center for Environmental Prediction(NCEP)(http://climate-scenarios.canada.ca).Using the general circulation model(GCM)variables as predictors and multiple regression analysis,SDSM provided downscaled future daily data at fine resolutions.With precipitation and temperature values from SDSM,SPEI was calculated under various representative concentration pathways(RCPs).Based on the relationships between the SPEI and SDI values obtained by linear regression during the statistical period(1989-2018),the areas with higher vulnerability regarding the river flow changes were simulated in the period from 2020 to 2070.The Arc-GIS software and the inverse distance weighting interpolation method were used to produce maps of the downscaled climate variables.For data quality assessment,the Wilcoxon(1945)and Mann-Kendall(Mann,1945;Kendall,1975)non-parametric tests were used for homogeneity and trend tests,respectively.In addition,the Landsat 5,Landsat 8,and Sentinel 2 satellite images were downloaded from the Google earth engine code website(https://code.earthengine.google.com).These satellite images were used to calculate NDWI for monitoring hydrological droughts in the study area from 1989 to 2018.

Fig.3.Annual average SPEI in three basins in period of 1989-2018.

2.3.Precipitation effectiveness variables in drought events

The precipitation effectiveness variables(PEVs)that describe precipitation features have the potential to characterize droughts.Otun(2005)introduced CPEI to arid and semi-arid regions and applied it in Nigeria and Iran.To calculate CPEI,the standardized value of PEVs should be computed:

whereSi，jis a standardized PEV value in thejth month of theith year,Ei，jis a PEV value in thejth month of theith year,andandQjare the mean and standard deviation of the PEV value for thejth month,respectively.Afterwards,CPEI is computed as follows:

whereICPEi,jis CPEI for thejth month of theith year,which is obtained by the optimal combination of PEVs;nis the number of PEVs in the combination;and the subscriptmmeans themth PEV.

2.4.SPEI

Vicente-Serrano et al.(2010)suggested using SPEI to identify drought processes based on precipitation and evapotranspiration data.This index combines the capacity concept of evapotranspiration demand changes in the Palmer drought severity index on different scales with the standardization scheme in the standardized precipitation index.Based on the categorized SPEI values,a SPEI value of greater than-0.5 indicates no drought;values ranging from-1 to-0.5,from-1.5 to-1,and from-2 to-1.5 refer to mild drought,moderate drought,and severe drought,respectively;and a SPEI value of less than-2 indicates extreme drought(Wang et al.,2018).To calculate SPEI,potential evapotranspiration(PET)is computed using Thornthwaite's equation(Thornthwaite,1948):

whereTis the monthly mean air temperature,Iis the heat index,Mis a coefficient depending onI,andKis the correction coefficient as a function of the latitude and month.

The water surplus or deficit in monthj,Dj,is defined as the difference between precipitation and PET:

wherepjandEpjare the monthly precipitation and PET in monthj,respectively.

Given the fact that theDseries may contain negative values,Vicente-Serrano et al.(2010)used log-logistic distribution to standardize theDseries.The probability density function of the three-parameter log-logistic distributionf（x）is as follows:

whereα，β,andγare the scale,shape,and location parameters,respectively,and they can be estimated with the Lmoment method(Hosking and Wallis,1987).The probability functionF（x）of the three-parameter log-logistic distribution is expressed as follows:

Using Eq.(6),the SPEI value is obtained as follows:

whereis the probability ofDexceeding a determined value,withP=1-F（x）;andc0，c1，c2，d1，d2,andd3are constants,withc0=2.515 517,c1=0.802 853,c2=0.010 328,d1=1.432 788,d2=0.189 269,andd3=0.001 308(Vicente-Serrano et al.,2010;Wang et al.,2018).

2.5.NDWI

NDWI was first introduced by McFeeters(1996)to delineate open water features.The sensitivity of NDWI to soil moisture has improved water resources identification and vegetation stress studies and facilitated leaf area index simulations and agricultural product modeling.In this study,three types of satellite images were used to extract NDWI in the study area.For the Landsat 5 images with a 30-m spatial resolution,NDWI is derived using the following equation:

whereP1is the reflectance of the visible blue band(band 1,with a wave length of 0.45-0.52μm),andP4is the reflectance of the near-infrared(NIR)band(band 4,with a wave length of 0.76-0.90μm).For the Landsat 8 images with a 30-m spatial resolution,NDWI is expressed as follows:

whereP2is the reflectance of the blue band(band 2,with a wave length of 0.45-0.51μm),andP5is the reflectance of the NIR band(band 5,with a wave length of 0.85-0.88μm).For the Sentinel 2 images with a 10-m spatial resolution,NDWI is calculated as follows:

whereP3is the reflectance of the green band(band 3,with a wave length of 0.56 nm),andP8is the reflectance of the NIR band(band 8,with a wave length of 0.842 nm).

Extraction of NDWI showed that the regions with the NDWI values greater than 0 in the Sentinel 2 images and greater than 0.1 in the Landsat 5 and 8 images were water bodies.

2.6.SDI

To characterize hydrological droughts,Nalbantis and Tsakiris(2009)developed SDI by considering monthly streamflow(Pathak and Dodamani,2016).SDI in monthjover a temporal scalek(k=1,2,3,or 4 months)is expressed as follows:

whereVj，kis the cumulative streamflow volume in monthjover the temporal scalek,andandDskare the mean cumulative streamflow volume and the standard deviation of cumulative streamflow volume over the temporal scalek,respectively.In this context,the truncation level is set asAccording to the values of SDI,hydrological droughts at five severity levels are defined.An SDI value of greater than 0 indicates non-drought conditions;values of-1.0-0,-1.5--1.0,and-2.0--1.5 demonstrate mild drought,moderate drought,and severe drought,respectively;and a value of less than-2.0 denotes extreme drought(Nalbantis and Tsakiris,2009).

2.7.SDSM

SDSM facilitates the rapid development of multiple,lowcost,single-site scenarios of daily weather variables under current and future regional climate forcing(Wilby et al.,2002).SDSM was developed by Wilby et al.(2002)as a useful tool for statistical downscaling to reproduce present climatic conditions and to project future climatic conditions(Abbasnia and Toros,2016).Many researchers have used this software to generate climate data for climate-change impact studies(Singh and Goyal,2016;Bessah et al.,2020).In this study,the correlation coefficient,root mean square error,and mean bias error indices were used to evaluate the performance of statistical downscaling of daily precipitation.

3.Results and discussion

3.1.Historical meteorological and hydrological droughts

Fig.4.Distributions of meteorological droughts at different severity levels according to SPEI in study area in period of 1989-2018.

Fig.5.Distributions of hydrological droughts at different severity levels according to SDI in study area in period of 1989-2018.

In this study,the regions suffering from meteorological and hydrological droughts at different severity levels were identified using historical climate data(Figs.4 and 5).The meteorological/hydrological drought frequencies based on SPEI and SDI values from the sub-basin stations were compared,and the drought events at different severity levels were divided into high and low-frequency groups.If the number of drought events at different severity levels at a station was lower than the average of the whole sub-basin,the corresponding region was marked as low frequency.Otherwise,it was marked as high frequency.The comparison of SPEI and SDI showed that severe and extreme hydrological droughts occurred over an expanded region,indicating that river flows are dependent on precipitation and drought conditions.Figs.4 and 5 demonstrate that the regions affected by meteorological and hydrological droughts are different owing to the spatial variation of climatic and geographic conditions that result in a high variability of precipitation over the study area.These two figures show that the study area experienced frequent mild meteorological droughts and severe hydrological droughts.In the northern region,mild and moderate meteorological droughts as well as moderate and severe hydrological droughts were frequent.In the Hablehrood-Shoorab sub-basin,the driest subbasin,severe hydrological droughts occurred more frequently than severe meteorological droughts.High evaporation in this region enhanced the intensity of droughts,and river flows responded strongly.For instance,mild hydrological droughts had a more immediate response to meteorological droughts in arid basins,demonstrating the fragile conditions of rivers in these regions.

Comparison of Figs.4 and 5 shows that the cumulative effect of meteorological droughts led to more severe hydrological droughts.There are some differences with regard to the occurrence of hydrological droughts and meteorological droughts.According to Fig.6,although mild meteorological droughts were more frequent,the frequencies of hydrological droughts at moderate,severe,and extreme levels exceeded those of meteorological droughts in most sub-basins.For example,moderate hydrological droughts were more frequent than meteorological droughts at the same level in all subbasins except the Hablehrood-Shoorab sub-basin(Fig.6).

Fig.6.Frequencies of meteorological droughts(SPEI)and hydrological droughts(SDI)at different severity levels in different subbasins.

3.2.Impact of precipitation variables on hydrological droughts

The low precipitation amount and fluctuations of precipitation on daily,seasonal,and annual scales represent the inherent climatic features in Iran(Mafakheri et al.,2017).A few precipitation indices can feasibly describe the frequency of precipitation events within a given period.For example,in the situation in which annual precipitation is higher than the long-term mean precipitation,the increasing number of dry days and,consequently,the reduction of the number of precipitation days in the year likely combine to worsen the overall drought conditions.

Zolfaghari and Noorisameleh(2016)used CPEI to conduct drought analysis in Iran and concluded that,in addition to the mean annual rainfall depth(MAR),two other variables play an important role in controlling the severity of droughts in most parts of the country.They are the total number of wet(rainy)days on the seasonal scale(TWD)and the total number of dry days within a year(TDY).

Table 1 shows the importance of different precipitation variables in controlling the severity of hydrological droughts that vary with different climatic and geographical conditions.Table 1 was obtained by calculating the correlation between the intensity of hydrological droughts and different combinations of the precipitation variables(MAR,TWD,and TDY).Considering the different distribution patterns of precipitation variables in the Jajrood-Karaj sub-basin,the sub-basin was divided into two groups,with a semi-arid climate in the north and an arid climate in the south.In the northern part of the Jajrood-Karaj and the Haraz sub-basins,TWD was predominant in controlling severe hydrological droughts.In the case of extreme droughts,TDY played a more important role than MAR and TWD in all sub-basins except in the Hablehrood-Shoorab sub-basin.For mild droughts in the Babolrood-Talar,the southern Jajrood-Karaj,and the Hablehrood-Shoorab sub-basins,moderate droughts in the Haraz and the Shoor sub-basins,severe droughts in the southern Jajrood-Karaj sub-basin,and extreme droughts in the Hablehrood-Shoorab sub-basin,MAR was the critical variable in controlling drought severity.In general,according to the relationship between these variables and mild and moderate droughts in the sub-basins,MAR and TWD played an important role in controlling drought severity.However,the increased TDY triggered river flows in a critically threatened state and consequently led to severe and extreme hydrological droughts.

Table 1 Rank of importance of precipitation variables in controlling severity of hydrological droughts in study area.

3.3.River flow changes detected by NDWI

In this study,satellite images were used to compute NDWI,and the derived NDWI values were adopted to monitor the variation in river flows.Subsequently,the NDWI-detected river flow changes were compared with those from SDI.Evaluations showed that NDWI was only useful for monitoring river flow changes in the humid Babolrood-Talar subbasin.Therefore,SDI was selected to uniformize the calculations.Vicario et al.(2019)and Zhao et al.(2020)concluded that several factors,such as river depth,basin location,deep valleys,and seasonality of rivers,largely affect river flows,and thus,streamflow data and SDI are more useful than NDWI for river flow detection.However,this study still used the Landsat 5,Landsat 8,and Sentinel 2 satellite images to derive NDWI and tried to evaluate the feasibility of monitoring river flow changes via the satellite-based techniques.Fig.7 shows that NDWI was only useful in the study of large reservoirs,dams,and lakes in the basins,as well as the most humid Babolrood-Talar sub-basin.

Fig.7.Water body detected by Sentinel 2-image-based NDWI in study area.

3.4.Projection of changes in drought conditions

RCPs refer to the possible pathways of greenhouse gas(GHG)emissions,atmospheric concentrations,air pollutant emissions,and land use in the 21st century.RCPs include a few future scenarios,such as a stringent mitigation scenario(RCP 2.6),two intermediate scenarios(RCP 4.5 and RCP 6.0),and a scenario with very high GHG emissions(RCP 8.5)(IPCC,2014).

Using the precipitation and temperature data statistically downscaled by SDSM,SPEI was calculated under different RCP scenarios(2020-2070).Under RCPs 2.6,4.5,and 8.5,the changes in drought events at different severity levels in the study area were projected,as shown in Fig.8,where HQS,NL,and CD denote the Haraz and Qarah Su Basin,Namak Lake Basin,and Central Desert Basin,respectively.Fig.8 shows that mild drought events in the historical period(1989-2018)and under future scenarios(2020-2070)had much higher occurrence frequencies than droughts at other severity levels,and there were no distinct differences between the historical and future periods.Under all future scenarios,the moderateto-extreme drought events would be significantly affected by future climate change.For example,the frequencies of moderate droughts in the Central Desert Basin would increase by 3% under RCPs 2.6 and 4.5.Meanwhile,extreme droughts would increase slightly across all basins,particularly in theHaraz and Qarah Su Basin.Based on the linear regression coefficients between meteorological and hydrological droughts(the values of SPEI and SDI)in the historical period(1989-2018),the vulnerability of river flows at each point in the study area affected by future droughts was simulated under RCPs 2.6,4.5,and 8.5(Fig.9).The average decreases in river flows less than 0.2% and in the range of 0.2%-0.5% from 2020 to 2070 were identified as insignificant and significant changes,respectively(Fig.9).

Fig.8.Comparison of drought frequencies at different severity levels based on SPEI in historical period(1989-2018)and under different RCP scenarios(2020-2070).

Based on the drought projection in the study area,the Hablehrood-Shoorab sub-basin in the Central Desert Basin and Jajrood-Karaj sub-basin in the Namak Lake Basin would undergo a significant increase in the severity of hydrological droughts.According to Fig.5,these regions experienced severe-to-extreme droughts with a high frequency in the historical period.Under RCP 8.5,the mountainous region would be vulnerable to intensified droughts,thereby producing critical threats to the rivers that originate in these mountains(Fig.9).The increased drought frequency induced by climate change would drastically reduce river flows in the study area.Also,in the areas identified as having significant changes in river flows in Fig.9,the coefficient of dispersion of precipitation variables would change.For example,under RCPs 4.5 and 8.5,the average variation coefficient of the three PEVs(MAR,TWD,and TDY)that control the severity of hydrological droughts would increase by 5% in the Babolrood-Talar and Jajrood-Karaj sub-basins.Changes in distribution pattern and coefficient of variation of precipitation variables will change the severity of meteorological/hydrological droughts.

Fig.9.Projected changes in river flows induced by future climate change and droughts in study area under different RCP scenarios(2020-2070).

4.Conclusions

This study was conducted to evaluate the impacts of climate change and drought variations on river flow vulnerability.The main conclusions are as follows:

Hydrological droughts were more severe than meteorological droughts at moderate,severe,and extreme levels in most sub-basins,and river flows demonstrated high vulnerability to climate change.In arid and semi-arid regions,the response of river flows to droughts was stronger than in humid regions.Severe droughts will be more frequent in the Haraz and Qarah Su Basin than in other basins,with an increase of frequency under RCPs 2.6 and 8.5.The changes in extreme droughts were projected to be slight.

The importance of precipitation variables in controlling mild-to-moderate droughts varied across different regions.However,as droughts intensified and became severe or extreme,the importance of TDY increased.In general,TWD and MAR were critical factors in controlling mild and moderate droughts.

Streamflow data and statistical hydrological drought indices are more accurate than remote-sensing-based indices in detecting river flow changes because the geographical location of rivers,river depth,and river seasonality greatly influence river flow changes.

Climate change in the semi-arid basins increases aridity and further intensifies hydrological droughts in these basins.Under RCP 4.5,river flows were projected to significantly change in drier basins because these regions will likely respond more to slight changes in climate conditions.Under RCP 8.5,the northern region of the Jajrood-Karaj sub-basin,which coincides with the central Alborz Mountains and is a valuable source area for many important rivers,was projected to be threatened by intensified droughts(Fig.9).The increased air temperature and intensified drought severity,combined with decreased precipitation days,would likely change the length of seasons.

Further concentrations of annual precipitation in a limited number of wet days and increased dry days will likely cause more severe floods and droughts,as occurred in 2019.The effects of climate change and increased meteorological and hydrological droughts in the study area will likely intensify the seasonality of rivers flows;reduce the average water volume of perennial rivers,dam capacity,and surface water supply;and increase withdrawal pressure on groundwater.To combat these issues,all sectors depending on water resources in Iran should conduct rational water resources planning and adjust the supply and demand patterns of water resources in adaption to future climate change and geographical differences in Iran.Owing to different impacts of climate change and droughts in various basins,planning for sustainable development of water resources should be conducted and adaptation policies should be designed regionally.

Declaration of competing interest

The authors declare no conflicts of interest.