Evaluation，prediction，and protection of water quality in Danjiangkou Reservoir，China

Xio-kng Xin*，Ke-feng Li，Brin Finlyson，Wei YinStte Key Lortory of Hydrulis nd Mountin River Engineering，Sihun University，Chengdu 610065，PRChinChngjingWter Resoures Protetion Institute，Chngjing Wter Resoures Commission，Wuhn 430051，PRChinDeprtment of Resoure Mngement nd Geogrphy，the University of Melourne，Vitori 3010，Austrli Reeived 13 July 2014；epted 24 Novemer 2014 Aville online 10 Ferury 2015



Evaluation，prediction，and protection of water quality in Danjiangkou Reservoir，China

Xiao-kang Xina，b，*，Ke-feng Lia，Brian Finlaysonc，Wei YinbaState Key Laboratory of Hydraulics and Mountain River Engineering，Sichuan University，Chengdu 610065，PRChinabChangjiangWater Resources Protection Institute，Changjiang Water Resources Commission，Wuhan 430051，PRChinacDepartment of Resource Management and Geography，the University of Melbourne，Victoria 3010，Australia Received 13 July 2014；accepted 24 November 2014 Available online 10 February 2015

Abstract

The water quality in the Danjiangkou Reservoir has attracted considerable attention from the Chinese public and government since the announcement of the M iddle Route of the South to North Water Diversion Project（SNWDP），w hich commenced transferring w ater in 2014. Integrated research on the evaluation，prediction，and protection of water quality in the Danjiangkou Reservoirw as carried out in this study in order to improve environmentalmanagement.Based on 120 water samples，wherein 17 water quality indicesweremeasured at 20monitoring sites，a single factorevaluationmethodwasused to evaluate the currentstatusofwaterquality.The resultsshow thatthemain indices influencing thewater quality in the Danjiangkou Reservoirare total phosphorus（TP），permanganate index（CODMn），dissolved oxygen（DO），and five-day biochemical oxygen demand（BOD5），and the concentrations of TP，BOD5，ammonia nitrogen（NH3-N），CODMn，DO，and anionic surfactant（Surfa）do not reach the specified standard levels in the tributaries.SeasonalM ann-Kendall tests indicated that the CODMnconcentration shows a highly significant increasing trend，and the TP concentration shows a significant increasing trend in the Danjiangkou Reservoir.The distribution of themain water quality indices in the Danjiangkou Reservoirwas predicted using a two-dimensionalwater quality numericalmodel，and showed that the sphere of influence from the tributaries can spread across half of the Han Reservoir if the pollutants are not controlled. Cluster analysis（CA）results suggest that the Shending River is heavily polluted，that the Jianghe，Sihe，and Jianhe rivers aremoderately polluted，and that they should be the focus of environmental remediation.

©2015 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/）.

Water quality；Single factor evaluation method；Mann-Kendall test；Numericalmodeling；Cluster analysis；Dangjiangkou Reservoir

1.In troduction

Water quality tends to degenerate gradually w ith human interventions，such as hydrological alterations（Booker and Woods，2014），land use change（Seeboonruang，2012），inputs of toxic chemicals and nutrients（Gevrey et al.，2010），and changes in other physicochemical properties of water（Paul and M eyer，2001；Vanlandeghem et al.，2012），which cause a series of environm ental problem s，for examp le，shortage of drinking water（Bao et al.，2012），deterioration of aquatic ecological systems（Hu and Cheng，2013），and emergence of endemic diseases（Zhao et al.，2012）.With environmental pollution becoming an increasingly serious problem，the issue ofwaterquality hasattracted seriousattention from the public and the government.The primary concern of the public is the current statusofwater quality，sowater quality evaluation has been extensively studied（Nives，1999；Simeonov etal.，2003；Crosa etal.，2006；Olsen etal.，2012）.However，governments are not only concerned about the current status，but also about the future trends（Crosa et al.，2006；Chang，2008）as well asm easures for the protection of water quality（Zhu et al.，2011）. Therefore，integrated research on water quality evaluation，prediction，and protection is one of the most important research issues.

Different water quality evaluation methods have been developed.In the past，single factor evaluationmethodswere w idely used in rivers（Xu et al.，2012），reservoirs，and lakes（Xu，2005）.Comprehensive indexmethods，such as thewater quality index method，were developed by the National Sanitation Foundation（NSF）（Ott，1978），and several modified water quality indices have been developed based on this m ethod（Nasiri et al.，2007）.Some researchers then sw itched to the water quality evolution trend analysis，for which many m ethods have been successfully used，such as the rank correlationmethod（El-Shaaraw ietal.，1983），time seriesanalysis method（Long et al.，2009），and parametric test method（Kundzew icz and Robson，2004；Yenilmez et al.，2011），but there are still no good solutions to seasonal andmissing-value effectproblems，and these trend analysismethodscannotshow the distribution of water quality indices in water bodies.

The Danjiangkou Reservoir is the water source for the M iddle Route of the South to North Water Diversion Project（SNWDP），and the success of the SNWDP depends on the water quality in the Danjiangkou Reservoir.Therefore，protection of water quality in the Danjiangkou Reservoir is of extreme importance.Generally，water quality management in reservoirs and lakes ismore complicated than in rivers，since thewater in reservoirs and lakes resides for a long time，and sedimentation and transformations can occur.After the Danjiangkou Reservoir was selected as the water source for the m iddle route of the SNWDP，the water quality goal was determined to be Grade IIaccording to the Chinese Environmental Quality Standards for SurfaceWater（GB 3838-2002）. This is quite a strict requirement.As a result，many sources of pointsource pollution around the reservoir have been removed by the government and many environmental remediation projectshave been carried out in recentyears.So far，the good water quality of the reservoir has been wellmaintained，and almost all the indices satisfy the requirements.However，water quality in some tributaries is stillbelow the required standard，which is a potential threat to the water quality in the Danjiangkou Reservoir.

Little attention hasbeen paid to evaluation，prediction，and protection of thewater quality in the Danjiangkou Reservoir. Therefore，this study focused on the water quality in the Danjiangkou Reservoir，analyzed the current status of water quality w ith a single factor evaluation method（the recommended method for environmental protection standards in China），and investigated the evolution trend of water quality w ith the seasonalM ann-Kendallm ethod，which can dealwell w ith seasonal and m issing-value effect problems.Then，the distributions of water quality indices were simulated with a two-dimensional numerical model，which can identify the pollution areas conveniently.Based on these analyses，a hierarchical clustering method was used to classify the tributaries in order to identify themain pollution source.Finally，proper protectionmeasurementsare proposed for the polluted tributaries for better water quality management of the Danjiangkou Reservoir.A ll the analysis results can effectively support water quality protection management for the government.

Fig.1.Locations of Danjiangkou Reservoir and monitoring sites.

2.Study area

The Danjiangkou Reservoir is located in the upper reaches of the Hanjiang River（Fig.1），the largest tributary in the middle reaches of the Yangtze River，and the water source of the SNWDP，which sends water to Beijing and other northern cities in China.Ithasa surface area of1 050 km2，and a total storage capacity of 29.05 km3when the water reaches its norm al level of 170 m.The reservoir catchm ent area is 95 200 km2，and the averageannual inflow isabout38.80 km3.The average annual precipitation is 833 mm，and 75%of the annual precipitation falls in M ay to October.The Danjiangkou Dam is located at the junction of the Hanjiang River and Danjiang River，so the Danjiangkou Reservoir can be subdivided into the Han Reservoir and Dan Reservoir.There are six countiesin the reservoir catchment:Xixia，Xichuan，Yunxi，Yunxian，Danjiangkou，Shiyan counties，w ith a combined population of 3.14 million（Fig.1）.The gate for the water diversion canal is located at Taocha.There are approximately 200 tributaries in the Danjiangkou Reservoir Catchment，of which 16main tributariesmake up 90%of thewhole drainage area.Information for these 16 tributaries is described in Table 1.

Table 1 Information for tributaries in study area.

3.M aterials and m ethods

3.1.Sample data collection

There are 20 water quality monitoring sites in the Danjiangkou Reservoir Catchment.Four of thesemonitoring sites，Danjiangkou Dam，Taocha，Taizishan，and Langhekou，numbered R1，R2，R3，and R4，respectively（Fig.1），and distributed near the Danjiangkou Dam，were used to m easure thewater quality in the reservoir.Another 16 of themonitoring sites，located near themouths of the 16 tributaries，numbered S1 to S16（Fig.1），were used tomeasure thewater quality in the 16 tributaries.

Water quality samples were collected monthly at four monitoring sites in the reservoir from 2005 to 2012，while sampleswere collected tw iceamonth at16monitoring sites in the tributaries from January 2012 to December 2012.Seventeen water quality indices were selected for analysis.The analytical techniques for these indices can be found in the Chinese water quality analytical standards，such as Determination of34 Elements（Pb，Cd，V，Petc.）（SL 394.1-2007），and the limit valuesof differentwater quality grades can be found in the EnvironmentalQuality Standards for SurfaceWater（GB 3838-2002）.The range of observed values for each index is shown in Table 2，where DO，NH3-N，TN，CODMn，BOD5，CN-，Vola，TP，Surfa，and Coliare the abbreviations for dissolved oxygen，ammonia nitrogen，total nitrogen，permanganate index，five-day biochemical oxygen demand，cyanide，volatile phenol，totalphosphorus，anionic surfactant，and fecal coliform.

3.2.Single factor evaluationmethod

Compared w ith the comprehensive index method，the single factor evaluation method is more conservative and considered to be more suitable for serious water pollution situations in China.In addition，the single factor evaluation m ethod can expediently identify the primary water quality indices.Thus，the single factor evaluation method was used to calculate eachwaterquality index.Through comparison of the calculated value of each index w ith the standard value，the grade for each index was determ ined.After the grades were determined for all 17 indices for each sample at each measuring site，theworstonewasconsidered thegrade for that site.

3.3.SeasonalMann-Kendall test

The seasonal Mann-Kendall test is a kind of non-parametric test method，which can deal well w ith the missing values and seasonal changes in the data.Itwas put forward by Hirsch et al.（1982）.It has been w idely used in water quality trend testing and analysis（Bouza-Dea~no et al.，2008；Yenilmez et al.，2011；Naddeo et al.，2013）.The related equations for calculating the Mann-Kendall test statistic S and the standardized Mann-Kendall test statistic Z are as follows:

where Xijis thewater quality index of the i thmonth of the j th year，Var（S）is thevariance of S，and n is the numberof years. Since thewater quality condition isa random event，Z follows a standard normal distribution.Under no trend assumptions，Z is equal to zero.If Z is not equal to zero，the water quality index has an increasing trend（Z>0）or a decreasing trend（Z<0）.Hirsch et al.（1982）suggested that the significance level be set asα=0.1 orα=0.01，indicating that the waterqualityindexhasasignificanttrendorahighlysignificant trend，respectively.Accordingtothestandardnormaldistributionfunction，theabsolutevalueofZis1.645whenα=0.1 and2.575whenα=0.01.Thatistosay，ifthecalculatedvalue ofZisgreaterthan2.575，thewaterqualityindexhasahighly significantincreasingtrend；if1.645

3.4.Two-dimensionalwaterqualitynumericalmodel

Thetwo-dimensionalwaterqualitynumericalmodelMIKE 21AD（DHI，2001）issuitableforsimulatingandpredicting thewaterqualityindexdistributionsforrivers，lakes，and reservoirs，andithasbeenwidelyusedinthisfield（Lindim etal.，2011；Zhouetal.，2011；Sokolovaetal.，2013；Zhang andXin，2013）.Thus，itwasusedtocomputethewaterquality indexdistributionsintheDanjiangkouReservoirbysolving theshallowwaterequationsandtheconvection-diffusion equation.Themodelwasvalidatedinthefloodseason（July）andthedryseason（December）usingaManningroughness valueof0.03anddegradationcoefficientsof0.04d-1and 0.0004d-1forCODMnandTP，respectively.Thevalidation resultsarelistedinTable3，whereVfisthemeasuredvalue，Vsisthesimulatedvalue，andErristherelativeerror.Asshownin Table3，therangeoferrorisbetween-10.0%and10.0%，whichisanacceptablelevelofagreement.

4.Resultsanddiscussion

4.1.Waterqualitystatusandprimarywaterquality indices

4.1.1.WaterqualitystatusofDanjiangkouReservoir

Thereare12samplesforthefourmeasuringsitesinthe DanjiangkouReservoirin2012.Theresultsforfourmain waterqualityindices（TP，CODMn，DO，andBOD5）arelisted inTable4，andtheremainingwaterqualityindicesareall belowthethresholdofGradeI.Thisindicatesthatthewater qualityinthereservoirisquitegood，andcansatisfythegoal inalmosteverymonthateachsite.Thus，thefourmainindices arethecontrolindicatorsfordeterminationofthewater qualitygradeoftheDanjiangkouReservoir.

Theboxplotsofthemainwaterqualityindicesareshownin Fig.2.AsshowninFig.2，themeanvaluesofTPconcentration areequaltoorlessthan0.02mg/Latthefourmeasuringsites（theupperlimitofGradeIIforthelakeorreservoiris 0.025mg/L），andthemeanvaluesofDOconcentrationare greaterthan8.0mg/Latthefoursites（thelowerlimitofGrade Iis7.5mg/L）.ThemeanvaluesofCODMnandBOD5concentrationsarelessthan2.5mg/Land1.5mg/L，respectively，bothofwhicharelessthantheupperlimitofGradeII.

Table 3 Comparison betw een simulated and observed w ater quality indices.

Table 4 Percentages of different grades of water quality and main water quality indices in Danjiangkou Reservoir.

Table 5 Percentages of different grades of water quality and water quality indices exceeding standards in tributaries.

4.1.2.Water quality status of tributaries

There are 24 samples from the tributaries in 2012.Table 5 shows that the grades for TP，BOD5，NH3-N，CODMn，DO，and Surfa exceed the goal grades for tributaries，The water quality in the Hanjiang（S1），Qihe（S11），Taohe（S12），and Jiangjun（S15）rivers is relatively good，while thewaterquality in the Shending（S4），Jianghe（S5），Sihe（S6），Guanshan（S7），Jianhe（S8），and Langhe（S9）rivers is poor.The box plots of the water quality indices exceeding the standard levels are shown in Fig.3.It can be seen from Fig.3 that the mean values of the indices are relatively high at sites 4，5，6 and 8.

Fig.2.Box plots of main water quality indices in Danjiangkou Reservoir（°denotes theoutlier that liesoutside the rangeof1.5 times the interquartile，and*denotes the outlier that lies outside the range of three times the interquartile）.

4.2.Temporal trends of primary indices

Based onmonthlymonitoring data from the fourmeasuring sites in the Danjiangkou Reservoir from 2005 to 2012，calculated valuesof Z are listed in Table 6.The concentrations of BOD5at the four sites are lower than the detection limit from 2005 to 2009，so trend analysis isunavailable.As can be seen in Table 6，at all four sites，the valuesof Z for CODMnare greater than 2.575，indicating that the CODMnconcentration show s a highly significant increasing trend.This results from increasing loads from the tributaries，particularly the Shending，Sihe，Guanshan，Jianhe，and Langhe rivers.These rivers run through countiesand townswhere serious pollution exists and the pollution load is from domestic sewage and agricultural runoff.At the Danjiangkou Dam（R1）and Taocha（R2）sites，thevaluesof Z for TPare greater than 1.645，which implies that the TP concentration has a significant increasing trend.Agricultural runoff isone of themost important sources of TP，especially in the upper reaches of the Hanjiang and Duhe rivers，where there are irrigated areas and m ountainfarm lands.At the Taocha site（R2），the Z value for DO is less than-1.645，indicating that the DO concentration has a significant deceasing trend，which imp lies that the organic pollutants that consum e oxygen are increasing.

Fig.3.Box plotsofwaterquality indicesexceeding standard levels in tributaries（°denotes the outlier that liesoutside the rangeof 1.5 times the interquartile，and*denotes the outlier that lies outside the range of three times the interquartile）.

Table 6 Valuesof Z for differentw ater quality indices at fourmeasuring sites.

Fig.4.Distributions of predicted TP concentration in Danjiangkou Reservoir and tributaries in two cases.

4.3.Prediction of concentration distributions

Point and non-point sources in the 16main tributaries are the main contributors of pollutants to the Danjiangkou Reservoir.Runoff in the Hanjiang Basin is concentrated in the flood season（≥75%），but the stored water volume in the Danjiangkou Dam is kept low to provide the floodm itigation. Therefore，water quality is worse in the flood season than in the dry season.The distributions of TP and CODMnconcentrations in two cases were predicted using the M IKE 21 AD model.The first case assumes that the TP and CODMnconcentrations in 16 tributaries reach their highest measured valuessynchronously，and thesecond case assumes thatthe TP and CODMnconcentrations of 16 tributaries reach the upper limitsof theirgoalgrades.Fig.4 shows the distributionsof the predicted TP concentration in the two cases，and Fig.5 shows the distributions of the predicted CODMnconcentration in the two cases.

As can be seen in Table 6，the TP concentration in the Danjiangkou Reservoir showsa significantly increasing trend，especially at the Danjiangkou Dam and Taocha sites，and this isunfavorable to the SNWDP.Fig.4（a）implies that if the TP concentrations in the tributaries are not controlled，especially in the Hangjiang and Duhe rivers，which have the largestand second largest discharges，thewater quality in almost half thearea of the Han Reservoir w ill notmeet the required goal，since the currentconcentration of TPisclose to theupper lim it for Grade II.Thus，control of TP concentration w ill be the focus ofwater pollutionmanagement in this zone.

Fig.5（a）shows that，although the water quality in the tributaries does not meet the goal，the CODMnpollution p lumes remain restricted to the tributaries，because the background concentration of CODMnin the Danjiangkou Reservoir is low and the discharges of the tributariesare relatively small. However，the increasing trend of CODMnconcentration at the fourmeasuring sites in the reservoir implies that there exists a potential threat to water quality in the Danjiangkou Reservoir.Fig.5（b）shows that，even if water quality in the tributaries satisfies their goals，improvement of the water quality in the tributaries is still necessary in order to ensure that the water quality in the reservoir reaches its goal，because there are different water quality goal requirements for the tributaries（Grade IV，IIIor II）and the Danjiangkou Reservoir（Grade II）.

Fig.5.Distributions of predicted CODMnconcentration in Danjiangkou Reservoir and tributaries in two cases.

4.4.Water quality protection strategy

4.4.1.Tributary classification

Tributariesare the key points forwater quality protection in theDanjiangkou Reservoir，and the question arisesas towhich tributaries'water quality should be improved first.We approach this problem by classifying the 16 main tributaries using a hierarchical clustering method.Based on six main water quality indices，TP，BOD5，NH3-N，CODMn，DO，and Surfa，a dendrogram of tributariesmade using Ward'smethod is shown in Fig.6，where Dlinkis the linkage distance，and Dmaxis the maximum linkage distance.As can be seen in Fig.6，the 16 tributaries fall into three distinct clusters:the light-pollution cluster（cluster 1），moderate-pollution cluster（cluster 2），and heavy-pollution cluster（cluster 3）.Cluster 1 consists of the Qihe，Jiangjun，Hanjiang，Taohe，Tianhe，Danjiang，Duhe，Laoguan，Quyuan，Guanshan，Langhe，and Taogou rivers.Cluster 2 consists of the Jianghe，Sihe，and Jianhe rivers，and cluster 3 is formed by the Shending River.This implies that the Shending River is heavily polluted，because this tributary runs though the Maojian and Zhanwan counties and receivesa lotof domestic and industrial sewage.

Fig.6.Dendrogram of tributariesmadeusingWard'smethod based on six main water quality indices.

4.4.2.Seasonal classification

The pollution load from pollution sources may vary seasonally，especially for non-point source pollution.Thus，identification of the primary pollution period is equally important.Hierarchical clusteringwasalso used to classify the months using the CODMnconcentrations at the Danjiangkou Dam site.Three clusters can be seen in Fig.7.March，April，and November cluster together as light-pollution months；January，December，and February are moderate-pollution months；and the remaining months（May to October），which constitute thewet season，have heavy pollution.Thus，we can infer that non-point source pollution is the primary pollution source in the Danjiangkou Reservoir zone.

4.4.3.Water protection strategy

Since industrial point sources have been removed，and the heavymetal，oil，and other inorganic contaminantsare not the principal pollutants in the Danjiangkou Reservoir area，attention should be paid to domestic pollution and agricultural non-point source pollution.Considering the characteristics of these two sources，water quality rem ediation should treat each tributary basin as a single unit.

The water protection strategy is as follow s:

It is first necessary to determine the water environmental capacity of the Danjiangkou Reservoir and the tributaries. Based on this，it is possible to determ ine allowable pollutant loads thatcan be discharged into the tributaries.Theallowable pollutant loads should be allocated among the industrial point sources，domestic point sources，and non-point sources.

Second，emergencymeasuresshould be taken to control the pollutant load of the Shending Basin.Sewage pipelines should be constructed to deliver sewage to treatmentplants.Industrial p lants should be required to treat their effluent on-site prior to discharging it.

Third，priority shouldbegiven to theconstructionof township sewage treatmentplantsand pipelinesin the Shending，Jianghe，Sihe，and Jianhe basins.Vegetation projectsshould bearranged in the upper reachesof theHanjiang and Duhe rivers.

Finally，thereshould beplanning forwater pollution control for all the 16 main tributaries and the other small tributaries，based on theirwater environmental capacity，and the pollution sources should bemonitored to ensure that the load reduction is achieved.

Fig.7.Dendrogram ofmonthsmade using Ward'smethod based on CODMnconcentrations.

5.Conclusions

Integrated research on the evaluation，prediction，and protection of the Danjiangkou Reservoirwas carried out in this study.The follow ing conclusions can be drawn:

（1）The overallwater quality in the Danjiangkou Reservoir isquite good，as itmeets thewater quality requirementof the SNWDP（Grade II），andmostof thewater quality indices can reach the Grade I standard.The water quality in the 16 m ain tributaries is relatively poor，especially that of the Shending，Jianghe，Sihe，and Jianhe rivers.The main water quality indices that exceed the standard in these tributaries are TP，BOD5，NH3-N，CODMn，DO，and Surfa，which is a serious threat to thewater quality in the Danjiangkou Reservoir.

（2）Of the main water quality indices in the Danjiangkou Reservoir，the CODMnconcentration shows a highly significant increasing trend，and the TP concentration shows a significant increasing trend.CODMnoriginates mainly from domestic and industrial point sources in the tributary basins，and TP comesmainly from agricultural non-point sources.

（3）Predictions of the distributions of main water quality indices imp ly that，w ithout rem ediation measures，the water quality in half the area of the Han Reservoirw ill fail to meet the required standard，endangering the quality of the source water for the SNWDP.

（4）Tributary classification results implied that the 16 tributaries can be divided into three clusters；the Shending River isheavily polluted；the Jianghe，Sihe，and Jianhe rivers are moderately polluted；and they should be the focus of environmental remediation.According to the pollution degree，some management，p lanning，and engineering measures should be taken to control the pollution loads from the tributary basins.

（5）Long-term water qualitymonitoring in the Danjiangkou Reservoir area should be carried out to illustrate the water quality change trends.Water quality restoration technologies should be further researched.

References

Bao，L.J.，M aruya，K.A.，Snyder，S.A.，Zeng，E.Y.，2012.China's water pollution by persistentorganic pollutants.Environ.Pollut.163，100-108. http://dx.doi.org/10.1016/j.envpol.2011.12.022.

Booker，D.J.，Woods，R.A.，2014.Comparing and combining physically-based and empirically-based approaches for establishing the hydrology of ungauged catchments.J.Hydrol.508，227-239.http://dx.doi.org/10.1016/ j.jhydrol.2013.11.007.

Bouza-Dea~no，R.，Ternero-Rodríguez，M.，Fernndez-Espinosa，A.J.，2008. Trend study and assessment of surface water quality in the Ebro River（Spain）.J.Hydrol.361（3-4），227-239.http://dx.doi.org/10.1016/ j.jhydrol.2008.07.048.

Chang，H.，2008.Spatial analysis of water quality trends in the Han River basin，South Korea.Water Res.42（13），3285-3304.http://dx.doi.org/ 10.1016/j.watres.2008.04.006.

Crosa，G.，Froebrich，J.，Nikolayenko，V.，Stefani，F.，Galli，P.，Calamari，D.，2006.Spatial and seasonal variations in the water quality of the Amu Darya River（Central Asia）.Water Res.40（11），2237-2245.http:// dx.doi.org/10.1016/j.watres.2006.04.004.

Danish Hydraulic Institute（DHI），2001.MIKE 21 User Guide and Reference Manual.DHI，Horsholm.

El-Shaaraw i，A.H.，Esterby，S.R.，Kuntz，K.W.，1983.A statistical evaluation of trends in thewater quality of the Niagara River.J.Gt.Lakes Res.9（2），234-240.http://dx.doi.org/10.1016/S0380-1330（83）71892-7.

Gevrey，M.，Com te，L.，de Zwart，D.，Lek，S.，2010.Modeling the chemicaland toxicwater status of the Scheldtbasin（Belgium），using aquatic invertebrate assemblages and an advancedmodelingmethod.Environ.Pollut.158（10），3209-3218.http://dx.doi.org/10.1016/j.envpol.2010.07.006.

Hirsch，R.M.，Slack，J.R.，Smith，R.A.，1982.Nonparametric test for trend in water quality.Water Resour.Res.18，107-121.http://dx.doi.org/10.1029/ WR020i001p00127.

Hu，Y.N.，Cheng，H.F.，2013.WaterpollutionduringChina'sindustrialtransition. Environ.Dev.8，57-73.http://dx.doi.org/10.1016/j.envdev.2013.06.001.

Kundzew icz，Z.W.，Robson，A.J.，2004.Change detection in hydrological records:a review of themethodology.Hydrol.Sci.J.49（1），7-19.

Lindim，C.，Pinho，J.L.，Vieira，J.M.P.，2011.Analysisof spatialand temporal patterns in a large reservoir using water quality and hydrodynamic modeling.Ecol.Model.222（14），2485-2494.http://dx.doi.org/10.1016/ j.ecolmodel.2010.07.019.

Long，J.，Ji，H.F.，Huang，Z.Y.，2009.Application of time series analysis in water quality prediction.Chin.J.Sci.Instrum.30（6），350-352（in Chinese）.

Naddeo，V.，Scannapieco，D.，Zarra，T.，Belgiorno，V.，2013.River water quality assessment:implementation of non-parametric tests for sampling frequency optimization.Land Use Policy 30（1），197-205.http:// dx.doi.org/10.1016/j.landusepol.2012.03.013.

Nasiri，F.，M aqsood，I.，Huang，G.，Fuller，N.，2007.Water quality index:a fuzzy river-pollution decision support expert system.J.Water Resour. Plan.Manag.133（2），95-105.http://dx.doi.org/10.1061/（ASCE）0733-9496（2007）133:2（95）.

Nives，S.G.，1999.Water quality evaluation by index in Dalmatia.Water Res. 33（16），3423-3440.http://dx.doi.org/10.1016/S0043-1354（99）00063-9.

Olsen，R.L.，Chappell，R.W.，Loftis，J.C.，2012.Water quality sample collection，data treatment and results presentation for principal components analysis:literature review and Illinois River watershed case study.Water Res.46（9），3110-3112.http://dx.doi.org/10.1016/j.watres.2012.03.028.

Ott，W.R.，1978.WaterQuality Indices:a Survey of IndicesUsed in the United States.USEnvironmental Protection Agency，Washington，D.C.

Paul，M.J.，Meyer，J.L.，2001.Streamsin theurban landscape.Annu.Rev.Ecol.Syst. 32，333-365.http://dx.doi.org/10.1146/annurev.ecolsys.32.081501.114040.

Seeboonruang，U.，2012.A statisticalassessmentof the impactof land useson surface water quality indexes.J.Environ.M anag.101，134-142.http:// dx.doi.org/10.1016/j.jenvman.2011.10.019.

Simeonov，V.，Stratis，J.A.，Samara，C.，Zachariadis，G.，Voutsa，D.，Anthemidis，A.，Sofoniou，M.，Kouim tzis，T.，2003.Assessment of the surfacewater quality in Northern Greece.Water Res.37（17），4119-4124. http://dx.doi.org/10.1016/S0043-1354（03）00398-1.

Sokolova，E.，Pettersson，T.J.R.，Bergstedt，O.，Hermansson，M.，2013.Hydrodynamicmodeling of themicrobialwater quality in a drinking water source as input for risk reduction management.J.Hydrol.497，15-23. http://dx.doi.org/10.1016/j.jhydrol.2013.05.044.

Vanlandeghem，M.M.，Meyer，M.D.，Cox，S.B.，Sharma，B.，Pati~no，R.，2012. Spatial and temporal patternsof surfacewaterquality and ichthyotoxicity in urban and rural riverbasins in Texas.Water Res.46（20），6638-6651. http://dx.doi.org/10.1016/j.watres.2012.05.002.

Xin，X.K.，Yin，W.，Yang，F.，2012.Research on waterquality evolution trend of Danjiangkou Reservoir by Kendall Method.Yangtze River 43（13），91-94（in Chinese）.

Xu，H.S.，Xu，Z.X.，Wu，W.，Tang，F.F.，2012.Assessmentand spatiotemporal variation analysisofwaterquality in the Zhangweinan River Basin，China. Procedia Environ.Sci.13，1641-1652.http://dx.doi.org/10.1016/ j.proenv.2012.01.157.

Xu，Z.X.，2005.Single factor water quality identification index for environmental quality assessmentof surfacewater.J.TongjiUniv.（Nat.Sci.）33（3），321-325（in Chinese）.

Yenilmez，F.，Keskin，F.，Aksoya，A.，2011.Water quality trend analysis in Eym ir Lake，Ankara.Phys.Chem.Earth Parts A/B/C 36（5-6），135-140. http://dx.doi.org/10.1016/j.pce.2010.05.005.

Zhang，S.P.，Xin，X.K.，2013.Application of M IKE 21 Module in waste drainage outletarrangement for enterprise sewage treatment plant.Water Resour.Power 31（9），101-104（in Chinese）.

Zhao，Y.，Xia，X.H.，Yang，Z.F.，Wang，F.，2012.Assessmentofwaterquality in Baiyangdian Lake using multivariate statistical techniques.Procedia Environ.Sci.13，1213-1226.http://dx.doi.org/10.1016/j.proenv.2012.01.115.

Zhou，N.Q.，Westrich，B.，Jiang，S.M.，Wang，Y.，2011.A coupling simulation based on a hydrodynam ics and water quality model of the Pearl River Delta，China.J.Hydrol.396（3-4），267-276.http://dx.doi.org/10.1016/ j.jhydrol.2010.11.019.

Zhu，D.S.，Zhang，J.Y.，Shi，X.X.，Yin，W.，2011.M odern water resources protection planning technology system.Water Resour.Prot.27（5），28-31（in Chinese）.

Thiswork was supported by the National Natural Science Foundation of China（Grants No.41101250 and 51309031）and the Chinese 12th Five-Year Science and Technology Support Program（Grant No.2012BAC06B00）.

*Corresponding author.

E-mail address:xin.xiaokang@163.com（Xiao-kang Xin）.

Peer review under responsibility of HohaiUniversity.

http://dx.doi.org/10.1016/j.wse.2014.11.001

1674-2370/©2015 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/）.