Guo-qing Wang,Jian-yun Zhang,*,Yue-ping Xu,Zhen-xin Bao,Xin-yue Yang

aState Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering,Nanjing Hydraulic Research Institute,Nanjing 210029,China

bResearch Center for Climate Change,M inistry ofWater Resources,Nanjing 210029,China

cInstitute of Hydrology and Water Resources,Civil Engineering,Zhejiang University,Hangzhou 310058,China

dCollege of Hydrology and Water Resources,Hohai University,Nanjing 210098,China

Estimation of futurewater resources of Xiangjiang River Basin w ith VICmodel undermultiple climate scenarios

Guo-qing Wanga,b,Jian-yun Zhanga,b,*,Yue-ping Xuc,Zhen-xin Baoa,b,Xin-yue Yangd

aState Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering,Nanjing Hydraulic Research Institute,Nanjing 210029,China

bResearch Center for Climate Change,M inistry ofWater Resources,Nanjing 210029,China

cInstitute of Hydrology and Water Resources,Civil Engineering,Zhejiang University,Hangzhou 310058,China

dCollege of Hydrology and Water Resources,Hohai University,Nanjing 210098,China

Abstract

Variation trends of water resources in the Xiangjiang River Basin over the com ing decades have been investigated using the variable infi ltration capacity(VIC)model and 14 general circulation models'(GCMs')projections under the representative concentration pathway (RCP4.5)scenario.Results show that the Xiangjiang River Basin w ill probably experience temperature rises during the period from 2021 to 2050,w ith precipitation decrease in the 2020sand increase in the 2030s.The VICmodelperformswell formonthly discharge simulationsw ith better performance for hydrometric stations on themain stream of the Xiangjiang River than for tributary catchments.The simulated annual dischargesare significantly correlated to the recorded annualdischarges forall theeightselected targetstations.The Xiangjiang River Basinmay experiencewater shortages induced by climate change.Annualwater resourcesof the Xiangjiang River Basin over theperiod from 2021 to 2050 are projected to decrease by 2.76%on averagew ithin the range from-7.81%to 7.40%.It isessential to consider the potential impactof climate change on water resources in future planning for sustainable utilization of water resources.

?2017 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:Water resources；Climate change；VICmodel；Xiangjiang River Basin；Climate scenarios；Hydrologicalmodeling

1.Introduction

Water resourcesplay a crucial role in ecological,social,and economic contexts(Zhang etal.,2013b).Globalwarmingwill alter the spatial and temporal distribution of water resources through accelerating hydrological cycle(IPCC,2008；Zhang and Wang,2007).A broad consensus indicates that global mean air temperature has risen by 0.85°C over the period from1880 to 2012with ahigher rateofwarming occurring in recent decades(IPCC,2013).About 20%of the world's large rivers have also experienced significant decreases in discharge(Dai et al.,2009)；these decreases in low-and middle-latitude areas could be linked to the recent drying and warm ing in West Africa,South Europe,and East and South Asia(Dai, 2013).The recorded runoff of the major rivers in northern China have been decreasing significantly over the past 50 years and this is consistent w ith the rising temperature, decreasing precipitation,and increasing water demands due to rapid agriculture and industry development(Wang et al., 2013a).Climate change has reduced renewable water resources in most sem i-arid and arid regions,and increased water stress to some extent(IPCC,2014).Understanding the role of climate change in water availability is essential for sustainable utilization of water resources.

Impacts of climate change on water resources have been w idely investigated in recent years.In terms of climate scenarios,numerous studies of hydrological responses have been divided into two types,hypothetical scenarios and projections from general circulation models(GCM s)(Yu et al.,1999；Jonesetal.,2006；Tavakoliand Smedt,2012；Xu etal.,2013).

Hypotheticalscenariosareused to analyze thesensitivity of hydrological variables to changes in climatic conditions,such as changes in temperature and precipitation.Bao et al.(2012) investigated the hydrological responses(stream flow,soil moisture,and actual evaporation)to climate change for the Haihe River Basin of China,and found that stream flow is muchmore sensitive than evapotranspiration and soilmoisture to climate change.Fu et al.(2007)indicated that a 30%increase in precipitation could result in a 50%increase in stream flow for a watershed in the Pacific Northwest,the United States；conversely,a 20%decrease in precipitation could lead to a 25%-30%of reduction in stream flow.Wang et al.(2013b)found that an increase in precipitation has greater impacton runoff than a decrease in precipitation does. In addition to the hydrologic impacts from changes in climatic variables,direct environmental responses due to increased carbon dioxide have been modeled,e.g.,increasing carbon dioxide and raising temperature by 6.4°C in a highly agriculturalwatershed decreased evapotranspiration by 37.5%and increased stream flow by 23.5%compared to present-day climate,while increased precipitation resulted in proportionally increased runoff(Ficklin etal.,2009).Long-term average monthly stream flow ismost sensitive to percentage change in precipitation,followed by the change in carbon dioxide concentration,and the change in temperature,for a forested watershed in the M ississippi River Basin in the United States (Parajuli,2010).The Intergovernmental Panel on Climate Change(IPCC)summarized the current studies of runoff response to changes in precipitation,show ing that changes in runoff are normally 1-3 times greater than equivalent percentage changes in precipitation(IPCC,2014).

GCM s are arguably the best available tools for modeling future climate(IPCC,2014；Wang et al.,2012；Zhang and Wang,2007).Projections from GCMs are mainly used to investigate future scenariosof regionalwater resources(Wang et al.,2012；Meng and Mo,2012；Zhang et al.,2013a). However,themagnitude of anthropogenic climatic change on water resourcesw ill depend on the em issions scenario,global and local climatic response,and the geographical characteristicsof each watershed(Rasilla etal.,2013；Jung and Chang, 2011；Wang and Zhang,2015).Moreover,seasonal distribution patterns could be altered by climate change.Forexample, a reduction in annual runoff of the Iberian Basinwasprojected under theSpecial Report for Emissions Scenarios(SRES)of SRES-A2 and SRES-B2,w ith decreases in runoff principally occurring in spring and summer(IPCC,2007；Rasilla et al., 2013).Using projections from five GCMs under SRES, runoff of the Songhua River Basin was projected to decrease in the next decades w ith notable differences ranging from 5.8%to 11.5%decreases of mean annual runoff(Meng and Mo,2012).M cFarlane et al.(2012)simulated water yields in southwestern Australia over future decades and found that surface water yieldsmay decrease by about 24%but w ith a large range of decrease(4%-49%)due to the uncertainty of projected climate scenarios.In the study of M cFarlane et al. (2012),15 GCMs and three em ission scenarios used in the IPCC-AR4(IPCC,2007)have been employed.

Currently,many of the challenges in assessing the impacts of climate change on water resources are associated w ith uncertain em ission scenarios,imperfect GCMs,downscaling methods,and hydrological models(Khan and Coulibaly, 2010),among which,the largest sources of uncertainty are climate scenarios produced by GCM s for various em ission scenarios(Kay et al.,2009).Relatively,the choice of hydrologicalmodelsandmethods for their parameterization are less important than the other factors when estimating long-term annual runoff(Gadeke et al.,2013).As uncertainty is an unavoidable issue in climate scenarios,using multiple climate scenarios is considered a practical and effective approach to quantifying the uncertainty.SRES have been adopted inmost of the current studies.However,the updated representative concentration pathway(RCP)scenarios issued in the IPCCAR5(IPCC,2013)and multiple GCMs projections are encouraged to be applied to the future climate change assessment(Yu et al.,2015).

Issues of water resources shortage in dry regions have alwaysattracted attention notonly from the centralgovernment butalso from local communities(Li,2012).However,hum id regions also experience significant challenges due to higher inter-annual variability of water resources,particularly in the context of climate change.The Xiangjiang River,located in the central region of China,is the largest river in Hunan Province.The river has abundantwater resources and therefore is regarded as the“Mother River”by the local people in Hunan Province.In recent years,low flows in Xiangjiang River have frequently occurred due to less precipitation and higher temperatures,which has had significant impacts on communities throughout the region(Zhu,2009).It is critically important to understand the variability and trends of water resources for effective water resources management.Therefore,themajor objectivesof this study were to:(1)analyze the expected climate change of the Xiangjiang River Basin based onmultiple GCMs'projections；(2)calibrate and validate the variable infi ltration capacity(VIC)model and simulate hydrological processes for the whole basin；and(3)investigate the resulting trends in water resources.

2.Data and m ethodology

2.1.Study area and data sources

The Xiangjiang River Basin is located in the areaw ith longitudes from 109.5°E to 111.1°E and latitudes from 38.2°N to 39.8°N in the central region of China.It originates from the Haiyang Mountain in the Guangxi Autonomous Region and emptiesinto Dongting Lakeafter running northward across five administrative regions of Yongzhou,Hengyang,Zhuzhou, Xiangtan,and Changsha in Hunan Province.The XiangjiangRiver covers 85383 km2of drainage area with amainstream length of 842 km.The Xiangjiang River Basin is a highly populated areaw itha totalpopulationof37.74m illionby 2010. The GDP and industrial value added in 2010 w ithin the basin were approximately 1220.5 billion RMB and 484.2 billion RMB,respectively,accounting for about 76.7%and 82.2%of the totalGDPand industrialvalueadded in Hunan Province.

Located in the East Asiamonsoon region,the Xiangjiang River Basin receives abundant precipitation(1500 mm per year),particularly in the flood season from April to August. Most of the basin area is covered by evergreen forest and deciduousbroad-leafandmixedevergreen trees,inwhich runoff is easily yielded(i.e.,more than 50%of precipitation becomes runoff).The mean annual runoff depth w ithin the basin is approximately 900mm w ith the highestmonthly runoff often occurring inMay.TheXiangjiang Riversystem consistsofmore than 70 rivers and streams,ofwhich five fi rst-level tributaries including the Chunlingshui,Leishui,M ishui,Lushui,and Liuyang riversare locatedon itsrightbankand threemajor fi rstlevel tributaries including the Xiangguang,Zhengshui,and Lianshui riversare located on its leftbank(Fig.1).

According to data availability,catchment size,and locations of hydrometric stations,six hydrometric stations on the tributaries and two key hydrometric stations,the Hengyang Station and Xiangtan Station on the main stream of the Xiangjiang River w ith a data length of over 20 years were selected to calibrate the hydrological model and test the model's performance.Recorded daily discharge data series from 1983 to 2010 at the eight hydrometric stations were collected from the Hydrology and Water Resources Bureau of Hunan Province,China.Meteorological data including daily precipitation and temperatures(m inimum,maximum,and mean)at20 nationalmeteorological stations,w ithin or nearby the Xiangjiang River Basin,were obtained from the China Meteorological Adm inistration(CMA).The basic information of the eight hydrometric stations is given in Table 1.

Fig.1.Xiangjiang Riversystem and locationsofhydrometric stations.

Table 1 shows the follow ing:(1)The annual precipitation for each catchment ranges from 1427.7 to 1635.1 mm w ith more precipitation occurring in the M ishui sub-basin located on the right bank of the m iddle reaches of the Xiangjiang River and relatively less precipitation occurring in the Lianshui sub-basin on the left bank of the lower reaches.(2)The runoff production efficiency ishigh,generating runoff ranging from 623.6 to 954.2mm.The Xiaoshuisub-basin in the upper Xiangjiang River Basin has the highest runoff efficiency, probably due to the extensive forest coverage in this area.(3) Annualmean temperature and annual pan evaporation range from 16.8°C to 17.4°C and 841.5 to 980.0 mm,respectively. Higher temperature and potential evaporation could affect water yield indirectly through increasing evapotranspiration.

2.2.GCMs and climate scenarios

IPCC-AR5 defines a number of climate scenarios from various GCMs.Based on these models'performance over China(Chen et al.,2014),14 GCM s'projections(Table 2) from the Coupled Model Intercomparison Project Phase 5 (CM IP5)were used to project future climate change for this study.The more details of these GCM s could be found in Taylor et al.(2012).

RCPs are greenhouse gas concentration(not em issions) trajectories adopted by the IPCC-AR5.The scenario numbers indicate the expected radiative forcing(e.g.,the lowest and highest scenarios,RCP2.6 and RCP8.5,w ill possibly have radiative forcing valuesof2.6 and 8.5W/m2by 2100).RCP4.5 is a moderate scenario which considers both econom ic developmentandm itigation actions；it is considered themost likely scenario and was therefore selected as themost appropriate scenario for this study.The downscaled RCP4.5 projections of the 14 GCMsw ith a resolution of 0.5°×0.5°and 1901-2100 data serieswere collected from CMA.Wang etal. (2015)selected the multi-year mean,Mann-Kendall coefficient,linear trend rate,aswell as seasonal and spatial distributions of precipitation and temperature series as indicators, and evaluated the performance of the 14GCM s in simulations over the period from 1961 to 2005 for the whole Xiangjiang River Basin by comparing indicators of the observed precipitation and temperature and the GCMs'simulations.Results show that the BNU-ESM model ranks fi rst among the 14 GCMs forhistorical simulationwhile no GCM could simulate historical variationsof precipitation and temperature perfectly (Wang et al.,2015).Therefore,it is very essential to project future climate change by usingmultiple GCMs'projections.

2.3.Description of VICmodel

The VIC model is a physically-based hydrologicalmodel, whichwasdeveloped by Liang etal.(1994)and later improved by Lohmann et al.(1998).Themodel considers two types of runoff yield mechanisms,infi ltration excess and saturation excess.The total runoff estimates of the VICmodel consistofsurface flow and base flow(Habetsetal.,1999).TheVICmodel hasbeen w idely applied to aw ide variety of sub-basins(Liang and Xie,2001；Bao et al.,2012；Wang et al.,2012).In comparison w ith other hydrologicalmodels,the VIC model has advantages of physically-based interpretation,w ide suitability fordifferentclimaticzones,andgood performance fordischarge simulation(Zhang andWang,2014).Estimatesof surface flow and base flow aremathematically described as follows:

whereQdis the surface flow,Qbis the base flow,Pis precipitation,W0is the initialsoilmoisture of the upper soil layer,Wmax0is themaximum soilmoistureof theuppersoil layer,I0is the initial infi ltration rate,Imis themaximum infi ltration rate,Bisa variable infi ltration curve parameter,Dmis themaximum daily base flow,Dsis the fraction ofDmin which non-linear base flow occurs,Wc2is the maximum soil moisture of the lowersoil layer,W-2is the initialsoilmoistureof the lowersoil layer,andWsis the fraction ofWc2in which non-linear base flow occurs.Themaximum soilmoisture of each soil layer is estimated based on soil texture and soil layer thickness.

The VICmodelneeds three typesof forcing inputdata,i.e., soil data,vegetation data,and hydro-meteorological data. Vegetation parameters include architectural resistance,theminimum stomata resistance,the leaf-area index,albedo, roughness length,and zero-plane displacement.The values of vegetation parameters in the VICmodel are given in Table 3, and the values of soil parameters are listed in Table 4.

Table 1Basic information of eight hydrometric stations.

Table 2CM IP5 GCM s used in this study.

The VIC model divides a target catchment into many grid cells w ith a flexible cell size(normally 0.25°×0.25°or 0.5°×0.5°for climate change studies),and runs ateach grid cell(Wang etal.,2012；Zhang andWang,2014).In this study, we divided the whole Xiangjiang River Basin into 54 cells w ith a resolution of 0.5°×0.5°(Fig.1).The station meteorological data was interpolated to each grid cell using the linear distance weighted interpolation method.Daily discharges at the Hengyang and Xiangtan stations on themain stream and the outlet stations of the six tributary sub-basins were used to calibrate and validate the VIC model.

3.Results and discussion

3.1.Future climate change in Xiangjiang River Basin

Basedonprojectionsw ith the14GCMs,trendsofprecipitation and temperatureovertheperiod from 2021 to2050wereanalyzed throughquartileanalysis.Fig.2 showsthebox-and-whiskerplots for changes in annual precipitation and temperature under 14 scenarios relative to the reference(from 1961 to 2010).

Fig.2 shows the follow ing results:(1)Uncertainty is still a major challenge in climate scenarios,w ith various GCMs producing different climatic changes.For example,scenario mean precipitation in the 2020s is expected to decrease by 1.73%,although eight GCMs'projections decrease and six projectionsincrease.(2)Precipitationovertheperiod from 2021 to 2050 seems to be roughly equivalent to the baseline(from 1961 to 2010),w ith amean decrease of 1.73%(w ithin a range from-10.96%to 7.17%)in the 2020s and amean increase of 2.65%(within a range from-7.59%to 14.88%)in the 2030s. The range of precipitation change for the period from 2021 to 2050(-4.64%-7.68%)ismuch smaller than those for other decades.(3)In contrast,the temperature w ill be consistently warm ing.Theannual temperatureover the period from 2021 to 2050 w ill likely increase by 1.36°C(ranging from 0.84°C to 2.14°C),w ith 1.0°C(ranging from 0.61°C to 1.65°C),1.39°C (ranging from 0.79°C to 2.05°C),and 1.69°C(ranging from 0.86°C to 2.73°C)of increase in the 2020s,2030s,and 2040s, respectively.In general,the Xiangjiang River Basinw ill likely becomewarmer in thenextdecades,w ith thepossibility ofdrier conditions in the 2020s and wetter conditions in the 2030s relative to the period from 1961 to 2010.

3.2.Model calibration and hydrologicalmodeling for Xiangjiang River Basin

In this study,we used the Nash and Sutcliffe efficiency criterion(NSE)and the relative error of volumetric fi t(RE)as objective functions to calibrate the hydrologicalmodel(Nashand Sutcliffe,1970).A good simulation resultw ill haveNSEclose to 1 andREapproaching 0.

Table 3Vegetation parameters in VICmodel.

Table 4Soil parameters in VICmodel.

Fig.2.Box-and-whisker plots for changes in precipitation and temperature over Xiangjiang River Basin during 2021-2050 relative to 1961-2010.

In order to calibrate and validate the VIC model,the data series from 1983 to 2010 were divided into two periods:a calibration period from 1983 to 2000,and a verification period from 2001 to 2010.The VICmodelwas run ata daily timestep and the resultswereaggregated tomonthly dischargevolumesat thehydrometricstations.Theperformanceof theVICmodel for discharge simulation is summarized in Table 5.Themonthly recorded and simulated discharges from 1983 to 2010 at the Laobutou Station on the tributary and the Xiangtan Station on themain stream of theXiangjiangRiverare takenastwo typical examplesand shown in Fig.3.Simulated annual runoffagainst recorded annual runoff for theeightstationsareplotted in Fig.4.

Table 5 shows the follow ing results:(1)The VIC model performs well for each sub-basin in general.TheNSEvalue varies from 89.4%to 98.6%and 73.2%to 93.0%w ith theREvalue ranging from-0.32%to 9.20%and-5.10%to 0.30%in the calibration and verification period,respectively,indicating that the model has high skill and small bias for discharge simulation.(2)Relatively,the VICmodel performs better for monthly dischargesimulationatthehydrometric stationson the main stream of the Xiangjiang River.TheNSEvalues in both calibration and verification periods at the Hengyang and Xiangtan stations are higher than 90.0%with theREvalues being very small.(3)There is a slight negative bias in runoffduring the verification period(w ith the exception of the Ganxi and Xiangtanstations),probably dueto the rapidurbanization in the region,thus increasing the imperviousarea and stream flow.

Table 5Performance of VIC model for discharge simulation for eight hydrometric stations.

Fig.3.Recorded and simulatedmonthly discharge from 1983 to 2010 at Laobutou and Xiangtan stations.

Fig.4.Recorded and simulated annual runoff for eight hydrometric stations in Xiangjiang River Basin.

Results in Figs.3 and 4 are in accordance w ith those in Table 5.The observed and simulated monthly discharges match well in general w ith a slight overestimates for peak discharges in early years and underestimates in more recent years.The simulated annual runoff is highly correlated to the observed annual runoff for all the eight hydrometric stations w ith correlation coefficients exceeding 0.85(Fig.4).The simulated annual runoffapproaches the recorded annual runoff in general.Higher values of runoff at the Laobutou,Xiangxiang,and Hengyang stations approach butare located below the 1:1 line,indicating good simulation and underestimation for higher runoff at these stations.In general,the VIC model performs well not only for tributaries,but also for the main stream of the Xiangjiang River.

3.3.Hydrological responses to climate change

Various runoff simulations from 1960 to 2050 were performedw ith thecalibrated VICmodeland14climateprojections undertheem ission scenarioofRCP4.5.Fortheperiod from 1961 to 2010,the spatialdistributionsof the runoff simulation driven by the recordedmeteorological forcing and the ensemblemean of the 14GCMs'projectionsare compared in Fig.5.

The spatial distributions of runoff simulations based on recorded meteorological forcing and multiple GCM s'projectionsarehighly in agreement.This implies thattheensemble mean of multiple GCMs'scenarios has good performance in simulating the spatial pattern of historical runoff.Both simulationsof runoff tend to decrease from theupper reaches to the lower reaches in general,w ith an exception of a high runoff regionoccurring in the threegrid cells(in lightblue in Fig.5)in the lower Xiangjiang River Basin,which is in accordancew ith thespatialpatternsofprecipitationandwater resourcesover the Xiangjiang River Basin(Xiao etal.,2013；Zhu,2009).

A box-and-whisker plot for changes in future annual runoff under 14 scenarios relative to the reference(from 1961 to 2010)is shown in Fig.6.The figure shows the follow ing:(1) The Xiangjiang River w ill probably experience decreased runoff relative to the baseline for the period from 1961 to 2010.The ensemble mean of annual runoff over the period from 2021 to 2050 w ill probably decrease by 2.76%(w ithin a range from-7.81%to 7.40%).The 75th percentile is -1.94%,indicating that most of the GCMs'projections decrease in runoff for the Xiangjiang River Basin exceptMPIESM-LR and CCSM 4.(2)The ensemblemean runoff of the 2020s and 2040sw ill probably decrease by 4.85%(w ithin a range from-17.63%to 8.75%)and 3.63%(w ithin a range from-14.26%to 8.11%)relative to the period from 1961 to 2010,which implies that the Xiangjiang River may havehigher water resource stress due to climate change.(3)In comparison to the baseline,there is no significant change in water resources over the period from 2031 to 2040,although all GCM s projected relatively a more significant change in runoff ranging from-11.11%to 16.43%.

Fig.5.Spatial distributions of simulated runoff obtained from recorded meteorological forcing and ensemblemean of 14 GCMs'projections.

Fig.6.Box-and-whiskerplot for change in runoffof Xiangjiang River Basin relative to 1961-2010.

W ith runoff simulation based on the GCM s'projections for the period from 1961 to 2010 asabaseline,theensemblemean based spatial distributions of change in water resources for each decade over the next20-50 yearsare presented in Fig.7.

Fig.7.Ensemblemean based spatialdistribution of percentage change inwater resourcesover Xiangjiang River Basin for2021-2050 relative to 1961-2010.

Fig.7 indicates the follow ing:(1)In the next20-40 years, the Xiangjiang Riverw ill probably presenta general decrease in annual runoff for the whole basin,particularly for the periods from 2021 to 2030 and from 2041 to 2050.(2)From 2021 to 2030,the annual runoff w ill probably decrease by 3.57%-6.36%for the whole basin w ith less significant decrease occurring in the upper reaches and more significantdecrease in the other areas.(3)The annual runoff for the period from 2031 to 2040 variesw ithin a narrow range from -0.97%to 1.10%w ith a decrease occurring in themiddle part of the Xiangjiang River Basin and an increase probably occurring in the other areas.(4)The spatial pattern of change in water resources for the period from 2041 to 2050 is sim ilar to that for the period from 2021 to 2030,w ithmore significant decrease occurring in the left part of the lower reaches.

4.Conclusions

Variation trends in water resources of the Xiangjiang River Basin for the next decades were simulated using the VIC model and 14 GCMs'projections under the RCP4.5 scenario. The results show that,in the next decades,temperature over the Xiangjiang River Basin w ill probably continue to rise while precipitation is highly uncertain w ith indications of possible decreases in the 2020sand 2040sand increases in the 2030s relative to the reference period from 1961 to 2010.By themiddle of the 21st century,temperaturew ill probably rise by 1.69°C,which variesw ithin a range from 0.86°C to 2.73°C. Over the period from 2021 to 2050,the annual precipitation is roughly equivalent to the baseline for the period from 1961 to 2010,w ith a large range of change from-4.64%to 7.68% relative to that of the reference period from 1961 to 2010.

The VIC model performs well for monthly discharge simulation of the eight typical hydrometric stations w ith theNSEvalues exceeding 70%and theREvalues falling w ithin the range between-5.1%and 9.2%,for both calibration and validation periods.The simulated annual runoff is highly correlated to the observed annual runoff for all the eight hydrometric stations.Relatively,the VIC model presents better simulation of discharge for the hydrometric stations on the main stream of the Xiangjiang River than it does for the stations on tributaries.The VICmodelnotonly can simulate the inter-annual variability of runoff,but also can simulate the spatial distribution of water resources over the Xiangjiang River Basin reasonably.The annual runoff depth over the Xiangjiang River Basin tends to decrease northward w ith a high value occurring in the lower reaches.

The Xiangjiang River Basin may undergo climate-induced decreases in runoff in the next decades,particularly for the periods from 2021 to 2030 and from 2041 to 2050.Over the period from 2021 to2050,theensemblemeanofannualrunoffis expected to decreaseby 2.76%,w ithin a range from-7.81%to 7.40%.A lthough the projected water resources are still very uncertain,the Xiangjiang River Basinw illprobably confronta new stress in water shortages induced by climate change.It is therefore of significance to consider the potential effects of climate change in the plan ofwater resourcesutilization.

Bao,Z.X.,Zhang,J.Y.,Liu,J.F.,Wang,G.Q.,Yan,X.L.,Wang,X.J., Zhang,L.R.,2012.Sensitivity of hydrological variables to climate change in the Haihe River basin,China.Hydrol.Process.26(15),2294-2306. http://dx.doi.org/10.1002/hyp.8348.

Chen,X.C.,Xu,Y.,Xu,C.H.,Yao,Y.,2014.Assessment of precipitation simulations in China by CM IP5 multiplemodels.Progress.Inquisitiones De.Mutat.Clim.10(3),217-225.http://dx.doi.org/10.3969/j.issn.1673-1719.2014.03.011(in Chinese).

Dai,A.G.,Qian,T.T.,Trenberth,K.E.,M illiman,J.D.,2009.Changes in continental freshwater discharge from 1948 to 2004.J.Clim.22, 2773-2792.http://dx.doi.org/10.1175/2008JCLI2592.1.

Dai,A.G.,2013.Increasing drought under global warm ing in observations and models.Nat.Clim.Change 3,52-58.http://dx.doi.org/10.1038/ nclimate1633.

Ficklin,D.L.,Luo,Y.Z.,Luedeling,E.,Zhang,M.H.,2009.Climate change sensitivity assessment of a highly agricultural watershed using SWAT.J. Hydrol.374,16-29.http://dx.doi.org/10.1016/j.jhydrol.2009.05.016.

Fu,G.B.,Barber,M.E.,Chen,S.L.,2007.Impacts of climate change on regionalhydrological regimes in the Spokane RiverWatershed.J.Hydrol. Eng.12(5),452-461.http://dx.doi.org/10.1061/(ASCE)1084-0699(2007) 12:5(452).

Gadeke,A.,Holzel,H.,Koch,H.,Pohle,I.,Grunewald,U.,2013.Analysisof uncertaintiesin thehydrological responseofamodel-based climatechange impact assessment in a sub-catchment of the Spree River,Germany. Hydrol.Process.28(12),3978-3998.http://dx.doi.org/10.1002/hyp.9933.

Habets,F.,Noilhan,J.,Golaz,C.,Goutorbe,J.P.,Lacarr`ere,P.,Leblois,E., Ledoux,E.,Martin,E.,Ottl■e,C.,Vidal-Madjar,D.,1999.The ISBA surface scheme in a macroscale hydrologicalmodel applied to the Hapex-Mobilhy area,Part I:Model and database.J.Hydrol.217(1-2),75-96. http://dx.doi.org/10.1016/S0022-1694(99)00019-0.

Intergovernmental Panel on Climate Change(IPCC),2007.Climate Change 2007:The Physical Science Basis.Cambridge University Press, Cambridge.

Intergovernmental Panel on Climate Change(IPCC),2008.Climate Change and Water.Cambridge University Press,Cambridge.

Intergovernmental Panel on Climate Change(IPCC),2013.Climate Change 2013:The PhysicalScienceBasis.CambridgeUniversity Press,Cambridge.

Intergovernmental Panel on Climate Change(IPCC),2014.Climate Change 2014:Impacts,Adaptation,and Vulnerability.CambridgeUniversity Press, Cambridge.

Jones,R.N.,Chiew,F.H.S.,Boughton,W.C.,Zhang,L.,2006.Estimating the sensitivity of mean annual runoff to climate change using selected hydrological models.Adv.Water Resour.29(10),1419-1429.http:// dx.doi.org/10.1016/j.advwatres.2005.11.001.

Jung,W.,Chang,H.,2011.Assessment of future runoff trendsundermultiple climate change scenarios in the Willamette River Basin,Oregon,USA. Hydrol.Process.25(2),258-277.http://dx.doi.org/10.1002/hyp.7842.

Kay,A.L.,Davis,H.N.,Bell,V.A.,Jones,R.G.,2009.Comparisonofuncertainty sources for climate change impacts:Flood frequency in England.Clim. Change92(1),41-63.http://dx.doi.org/10.1007/s10584-008-9471-4.

Khan,M.S.,Coulibaly,P.,2010.Assessing hydrologic impact of climate change w ith uncertainty estimates:Bayesian neural network approach.J. Hydrometeorol.11,482-495.http://dx.doi.org/10.1175/2009JHM 1160.1.

Li,P.,2012.The present situation of water resources in Baicheng City and countermeasures to solve the problemsof droughtand water shortage.Jilin Water Resour.45(4),27-28.http://dx.doi.org/10.15920/j.cnki.22-1179/ tv.2012.04.006(in Chinese).

Liang,X.,Lettenmaier,D.P.,Wood,E.F.,Burges,S.J.,1994.A simple hydrologically based model of land surface water and energy fluxes for generalcirculationmodels.J.Geophys.Res.99(D7),14415-14428.http:// dx.doi.org/10.1029/94JD00483.

Liang,X.,Xie,Z.H.,2001.A new surface runoffparameterizationwith subgridscale soil heterogeneity for land surface model.Adv.Water Resour. 24(9-10),1173-1193.http://dx.doi.org/10.1016/S0309-1708(01)00032-X.

Lohmann,D.,Raschke,E.,Nijssen,B.,Lettenmaier,D.P.,1998.Regional scale hydrology:I.Formulation of the VIC-2Lmodel coupled to a routing model.Hydrol.Sci.J.43(1),131-141.http://dx.doi.org/10.1080/ 02626669809492107.

M cFarlane,D.,Stone,R.,Martens,S.,Thomas,J.,T.,Silberstein,R.,S.,A li,R., Hodgson,G.,2012.Climate change impactson water yields and demands in south-western Australia.J.Hydrol.475,488-498.http://dx.doi.org/ 10.1016/j.jhydrol.2012.05.038.

Meng,D.J.,Mo,X.G.,2012.Assessing the effect of climate change onmean annual runoff in the Songhua River Basin,China.Hydrol.Process.26(7), 1050-1061.http://dx.doi.org/10.1002/hyp.8180.

Nash,J.E.,Sutcliffe,J.V.,1970.River flow forecasting through conceptual models,Part 1:A discussion of principles.J.Hydrol.10(3),282-290. http://dx.doi.org/10.1016/0022-1694(70)90255-6.

Parajuli,P.B.,2010.Assessing sensitivity of hydrologic response to climate change from forested watershed in M ississippi.Hydrol.Process.24(26), 3785-3797.http://dx.doi.org/10.1002/hyp.7793.

Rasilla,D.F.,Garmendia,C.,García-Codron,J.C.,2013.Climate change projections of stream flow in the Iberian Peninsula.Int.J.Water Resour. Dev.29(2),184-200.http://dx.doi.org/10.1080/07900627.2012.721716.

Tavakoli,M.,Smedt,F.D.,2012.Impactof climate changeon stream fl ow and soil moisture in the Vermilion Basin,Illinois.J.Hydrol.Eng.17(10), 1059-1070.http://dx.doi.org/10.1061/(ASCE)HE.1943-5584.0000546.

Taylor,K.E.,Stouffer,R.J.,Meehl,G.A.,2012.An overview of CM IP5 and the experiment design.Bull.Am.Meteorol.Soc.93,485-498.http:// dx.doi.org/10.1175/BAMS-D-11-00094.1.

Wang,G.Q.,Zhang,J.Y.,Jin,J.L.,Pagano,T.C.,Calow,R.,Bao,Z.X., Liu,C.S.,Liu,Y.L.,Yan,X.L.,2012.Assessing water resources in China using PRECISand a VICmodel.Hydrol.Earth Syst.Sci.16(1),231-240. http://dx.doi.org/10.5194/hess-16-231-2012.

Wang,G.Q.,Yan,X.L.,Zhang,J.Y.,Liu,C.S.,Jin,J.L.,Liu,Y.L.,Bao,Z.X., 2013a.Detecting evolution trends in the recorded runoff from themajor rivers in China during 1950-2010.J.Water Clim.Change 4(3),252-264. http://dx.doi.org/10.2166/w cc.2013.021.

Wang,G.Q.,Zhang,J.Y.,Xuan,Y.Q.,Liu,J.F.,Jin,J.L.,Bao,Z.X.,He,R.M., Liu,C.S.,Liu,Y.L.,Yan,X.L.,2013b.Simulating theimpactofclimatechange on runoff in a typical river catchmentof the Loess Plateau,China.J.Hydrometeorol.14(5),1553-1561.http://dx.doi.org/10.1175/JHM-D-12-081.1.

Wang,G.Q.,Wan,S.C.,Bao,Z.X.,Liu,C.,S.,2015.Evaluating the performance of GCMs in historical simulations for the Xiang River Basin.In: Proceedings of CHES 2015.Hohai University Press,Nanjing, pp.1189-1193.

Wang,G.Q.,Zhang,J.Y.,2015.Variation of water resources in the Huang-Huai-Hai areas and adaptive strategies to climate change.Quat.Int. 380-381,180-186.http://dx.doi.org/10.1016/j.quaint.2015.02.005.

Xiao,Y.,Tang,S.H.,Chen,H.,Hu,L.J.,2013.Temporaland spatial trendsof precipitation and temperature from 1960 to 2008 in Xiangjiang River Basin.Yangtze River 44(3),10-12,32.http://dx.doi.org/10.3969/ j.issn.1001-4179.2013.03.003(in Chinese).

Xu,Y.P.,Zhang,X.J.,Ran,Q.H.,Tian,Y.,2013.Impactof climate change on hydrology of upper reaches of Qiantang River Basin,East China.J. Hydrol.483,51-60.http://dx.doi.org/10.1016/j.jhydrol.2013.01.004.

Yu,Z.,Lakhtakia,M.N.,Yarnal,B.,White,R.A.,M iller,D.A.,Frakes,B., Barron,E.J.,Duffy,C.,Schwartz,F.W.,1999.Simulating the river-basin response to atmospheric forcing by linking a mesoscale meteorological model and hydrologicmodel system.J.Hydrol.218(1-2),72-91.http:// dx.doi.org/10.1016/S0022-1694(99)00022-0.

Yu,Z.B.,Jiang,P.,Gautam,M.R.,Zhang,Y.,Acharya,K.,2015.Changes of seasonal storm properties in California and Nevada from an ensemble of climate projections.J.Geophys.Res.Atmos.120(7),2676-2688.http:// dx.doi.org/10.1002/2014JD022414.

Zhang,J.Y.,Wang,G.Q.,2007.Impactsof Climate Changeon Hydrology and Water Resources.Science Press,Beijing(in Chinese).

Zhang,J.Y.,Wang,G.Q.,Pagano,T.C.,Jin,J.L.,Liu,C.S.,He,R.M.,Liu,Y.L., 2013a.Usinghydrologic simulation toexplore theimpactsofclimatechange on runoff in the Huaihe River basin of China.J.Hydrol.Eng.18(11), 1393-1399.http://dx.doi.org/10.1061/(ASCE)HE.1943-5584.0000581.

Zhang,J.Y.,Wang,G.Q.,2014.Variation of Stream Flow and Quantitative Identifi cation for Attribution.Science Press,Beijing(in Chinese).

Zhang,Q.,Singh,V.P.,Li,J.F.,2013b.Eco-hydrological requirements in arid and sem i-arid regions:The Yellow River in China as a case study.J. Hydraulic Eng.18(16),689-697.http://dx.doi.org/10.1061/(ASCE) HE.1943-5584.0000653.

Zhu,Y.,2009.Analysis on water resources safety issues of Xiangjiang River Basin.Hunan Hydro Power 43(4),35-37.http://dx.doi.org/10.16052/ j.cnki.hnslsd.2009.04.025(in Chinese).

Received 2 January 2017；accepted 23March 2017 Available online 24 June 2017

Thiswork was supported by the National Natural Science Foundation of China(Grants No.41330854 and 41371063)and the National Key Research and Development Programs of China(Grants No.2016YFA0601601 and 2016YFA0601501).

*Corresponding author.

E-mail address:jyzhang@nhri.cn(Jian-yun Zhang).

Peer review under responsibility of Hohai University.

http://dx.doi.org/10.1016/j.wse.2017.06.003

1674-2370/?2017 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/).