Journal of Hydrology 494 (2013) 83–95

Contents lists available at SciVerse ScienceDirect

Journal of Hydrology

journal homepage: www.elsevier .com/ locate / jhydrol

Distinguishing the relative impacts of climate change and humanactivities on variation of streamflow in the Poyang Lake catchment,China

0022-1694/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.jhydrol.2013.04.036

⇑ Corresponding author. Tel.: +86 25 86882102; fax: +86 25 57714759.E-mail address: qzhang@niglas.ac.cn (Q. Zhang).

Xuchun Ye a,b, Qi Zhang b,⇑, Jian Liu c, Xianghu Li b, Chong-yu Xu d,e

a School of Geographical Sciences, Southwest University, Chongqing 400715, Chinab Nanjing Institute of Geography and Limnology, State Key Laboratory of Lake Science and Environment, Nanjing 210008, Chinac Key Laboratory of Water Resources and Environment, Water Research Institute of Shandong Province, Jinan 250013, Chinad Department of Geosciences, University of Oslo, Norwaye Department of Earth Sciences, Uppsala University, Sweden

a r t i c l e i n f o

Article history:Received 18 January 2013Received in revised form 8 April 2013Accepted 20 April 2013Available online 3 May 2013This manuscript was handled byKonstantine P. Georgakakos, Editor-in-Chief,with the assistance of Ashish Sharma,Associate Editor

Keywords:Climate changeHuman activitiesHydrological responseMK testPoyang Lake catchment

s u m m a r y

Under the background of global climate change and local anthropogenic stresses, many regions of theworld have suffered from frequent droughts and floods in recent decades. Assessing the relative effectof climate change and human activities is essential not only for understanding the mechanism of hydro-logical response in the catchment, but also for local water resources management as well as floods anddroughts protection. The Poyang Lake catchment in the middle reaches of the Yangtze River has experi-enced significant changes in hydro-climatic variables and human activities during the past decades andtherefore provides an excellent site for studying the hydrological impact of climate change and humanactivities. In this study, the characteristics of hydro-climatic changes of the Poyang Lake catchment wereanalyzed based on the observed data for the period 1960–2007. The relative effect of climate change andhuman activities was first empirically distinguished by a coupled water and energy budgets analysis, andthen the result was further confirmed by a quantitative assessment. A major finding of this study is thatthe relative effects of climate change and human activities varied among sub-catchments as well as thewhole catchment under different decades. For the whole Poyang Lake catchment, the variations of meanannual streamflow in 1970–2007 were primarily affected by climate change with reference to 1960s,while human activities played a complementary role. However, due to the intensified water utilization,the decrease of streamflow in the Fuhe River sub-catchment in 2000s was primarily affected by humanactivities, rather than climate change. For the catchment average water balance, quantitative assessmentrevealed that climate change resulted in an increased annual runoff of 75.3–261.7 mm in 1970s–2000sfor the Poyang Lake catchment, accounting for 105.0–212.1% of runoff changes relative to 1960s. How-ever, human activities should be responsible for the decreased annual runoff of 5.4–56.3 mm in the otherdecades, accounting for �5.0% to �112.1% of runoff changes. It is noted that the effects of human activ-ities including soil conservation, water conservancy projects and changes in land cover might accumulateor counteract each other simultaneously, and attempts were not made in this paper to further distinguishthem.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Climate change and human activities are the two factors that af-fect the change of catchment hydrology. According to the IPCC(2007), the average global surface temperature increased by0.74 �C over the last 100 years. One of the most significant poten-tial consequences of climate change may be alterations in regionalhydrological cycles (e.g., Huntington, 2006). General consensus

have revealed that global warming and related changes to thehydrological cycle are likely to enhance the frequency and severityof extreme climate events, causing more severe floods anddroughts (e.g., Milliman et al., 2008; Bates et al., 2008; Déryet al., 2009; Jung et al., 2012; Thompson, 2012; Li et al., in press;Xiong et al., in press). In addition to global climate change, in-creases in human activities such as cultivation, irrigation, affores-tation, deforestation and urban construction have also introducedchanges to flow regime, especially large scale changes of land coveror its management (e.g., Yang et al., 2004; Brown et al., 2005; Jianget al., 2012; Yang et al., 2012a,b). Depending on the study region,

84 X. Ye et al. / Journal of Hydrology 494 (2013) 83–95

impacts of human activities on streamflow may be different.Reduction in streamflow has been shown in several arid andsemi-arid catchments in north China due to implementation ofconservation practices and increased water utilization (e.g., Liet al., 2007; Wang and Meng, 2008; Liu et al., 2009a). However,studies in Iowa’s rivers noted that increasing agricultural intensitymay increase stream discharge but reduce its variance due to thedecreasing surface runoff and increasing baseflow (e.g., Tomeret al., 2005; Schilling, 2004). Moreover, water utilization for agri-cultural and industrial development can also lead to significantchange in the water cycle and affect the variation of surface orsub-surface runoff (e.g. Du et al., 2012).

Within the last decades, water quantity and quality have be-come increasingly serious issues for water resources managementat catchment and/or regional scale (e.g., Kizza et al., in press; Liet al., in press; Ren et al., 2002; IPCC, 2007; Tomer and Schilliing,2009; Lakshmi et al., 2012). Therefore, understanding the influenceand relative importance of climate change and human-inducedchange on hydrology and water resources has recently drawn con-siderable concerns (e.g. Siriwardena et al., 2006; Ma et al., 2008; St.Jacques et al., 2010; Jin et al., 2012; Carless and Whitehead, inpress; Zhan et al., in press). Lahmer et al. (2001) indicated that cli-mate change is the dominant factor that affects the change ofstreamflow in wet regions, while human activities such as someextreme land-use change only resulted in comparatively small im-pacts on regional water balance. Similar result was also revealed byLegesse et al. (2003) for tropical Africa. However, the study by Ray-mond et al. (2008) suggested that land use change and manage-ment were more important than climate change for explainingthe increasing water export from the Mississippi River. In northernChina, increasing water shortage is very common in recent yearsdue to significant regional precipitation variation as well as rapiddevelopment of local economy (e.g., Piao et al., 2010). A quantita-tive assessment revealed that local human activities since the1970s led to a decrease of the water diverted into the main streamof the Tarim River catchment, which has been aggravated in the2000s (Tao et al., 2011). In Haihe River catchment, an importanteconomic center of China, human activities were estimated to beresponsible for the decline in annual water discharge, which ac-counts for over 50% of runoff reduction, while the contribution ofclimate change is relatively small (Wang et al., 2012). In a recentstudy conducted by Zhang et al. (2011a), the trends of the annualstreamflow and precipitation and the relationship between themwere analyzed in nine large river basins of China during 1956–2005. The results indicated that annual runoff has the same chang-ing trend as precipitation in humid regions revealing a stationaryrainfall–runoff relationship is still held. However, in arid andsemi-arid regions of north China the decline in streamflow is fasterthan the decreases of precipitation since 1970s, indicating that therelationship between the annual precipitation and streamflow pre-sents a non-stationary state. This non-stationary relationship isstrongly influenced by human activities, especially by the increaseof irrigation water use.

Poyang Lake, the largest fresh water lake in China, is located inthe middle reaches of the Yangtze River with a catchment area of162,225 km2. The Lake and its surrounding catchments have suf-fered from frequent droughts and floods in recent decades, espe-cially in 1990s and 2000s (e.g., Wang et al., 2008; Min et al.,2011). These severe drought and flood events have raised concernsfor the lake ecology and local water resources management. Stud-ies on hydrological response suggested that the changes of annualstreamflow in the catchment were primarily caused by climateanomalies in the Yangtze River catchment, while human activitiessuch as land-use change and modifications to river systems includ-ing the Yangtze River also exerted some impacts (e.g., Min, 2002;Guo et al., 2008, 2011; Zhang et al., 2012). Several studies showed

that the variations of streamflow is much more strongly related toregional climate especially precipitation, but this is insufficient toexplain all the changes (e.g., Guo et al., 2007; Zhao et al., 2009;Ye et al., 2009). For example, the increased vulnerability of the laketo floods is further elevated by deforestation and change of land-scape in the basin. In addition, the construction of large-scalewater conservancy facilities (reservoir and irrigation system) inthe catchment is another important factor altering the annual hyd-rograph and increasing the water utilization (e.g., Zhang et al.,2011b; Liu et al., 2009b). Large amount of water demand severelydecreased the catchment discharges to Poyang Lake and elevatedthe drought in the lake area, especially in dry years.

Effects of climate change and human activities on runoff varia-tion are significantly sensitive, especially in arid and semi-arid re-gions, and these effects have resulted in severe environmentaldegradation and water crises. However, previous studies revealedthat the relative importance of the influence of climate variabil-ity/change and human activities varies from region to region. Toour knowledge, the relative contribution of climate change and hu-man activities to runoff change in the Poyang Lake catchment hasnot been well investigated. Further studies are needed in order toprovide a generalized and conclusive interpretation of the changesobserved. The answer to this is essential not only for an improvedunderstanding of the mechanism of hydrological response in thecatchment, but also for local water resources management as wellas floods and droughts protection and mitigation in the PoyangLake catchment and the lower reaches of the Yangtze River. Thepurposes of this study are: (1) to investigate the variability oflong-term historical records of climate and hydrological data inthe Poyang Lake catchment; and (2) to evaluate the relative im-pacts of climate change and human activities on catchment-scalestreamflow response under different spatial and temporal scales.

1.1. Overview of the Poyang Lake catchment

The Poyang Lake, connected to the Yangtze River, lies on thenorthern border of the Jiangxi Province, China. The lake receiveswater flows mainly from five rivers: Ganjiang, Fuhe, Xinjiang,Raohe and Xiushui, and discharges into the Yangtze River from anarrow outlet in the north (see Fig. 1). Among the five major rivers,the Ganjiang is the largest river in the region extending 750 kmand contributes almost 55% of the total discharge into the PoyangLake (Shankman et al., 2006). The topography of the Poyang Lakecatchment varies from highly mountainous regions (maximumelevation of about 2200 m above sea-level) to alluvial plains inthe lower reaches of the primary watercourses. Headwater of theserivers are located in boundaries of the east, south and west of theJiangxi Province that surrounded by high mountains. Stream gradi-ent decreases as these rivers flow onto the relatively flat regionsurrounding the Poyang Lake. The wide alluvial plains surroundingPoyang Lake and the broad alluvial valleys of the tributary streamsare important rice growing regions in Jiangxi Province as well as inChina; most notably the lower reaches of Ganjiang and Fuhe sub-catchments have large irrigation areas over 10,000 ha (see inFig. 1).

The Poyang Lake catchment belongs to a subtropical wet cli-mate zone with an annual mean precipitation of 1680 mm and an-nual mean temperature of 17.5 �C. Annual precipitation in thecatchment shows a wet and a dry season and a short transitionperiod in between (see Fig. 2). Water inputs from the five sub-catchments are particularly important during the wet season fromApril through June when heavy rainfall produces large surfaceflows from the sub-catchments to the lake (Shankman et al.,2006). Rainfall decreases sharply from July to September, whileevapotranspiration is still very strong in these months (Fig. 2).After September, the dry season sets in and lasts through

Fig. 1. Topography and river networks of the Poyang Lake catchment, with stream gauging stations (‘‘Hydrostation’’) and meteorological stations (‘‘Meterostation’’) aremarked.

Fig. 2. Mean monthly precipitation and evapotranspiration of the Poyang Lakecatchment for 1960–2007.

X. Ye et al. / Journal of Hydrology 494 (2013) 83–95 85

December, and surface flow of the catchment is very low duringthis period. In response to annual cycle of precipitation, about59.1% of the annual discharge arrives from March to June, but only13.7% arrives from October to next January. In normal years, thePoyang Lake can expand to a large water surface of 4000 km2 witha volume of 320 � 108 m3 in the wet season, but shrinks to littlemore than a river during the dry season (Xu et al., 2001).

As a typical agriculture catchment, about 13% of the land area isbeing irrigated. Although water resources in the Poyang Lakecatchment are pretty abundant, the difficulty of water utilizationis affected by the strong variability of annual and seasonal precip-itation. Furthermore, with the rapid economic development andpopulation explosion in the catchment, influence of human activi-ties on water resources has become increasingly important. Directeffects of human activities mainly include the soil conservationand water conservancy projects. The statistical data indicated thata total amount of 9530 reservoirs were built across the catchmentuntil 2007, and most of these projects were constructed before1980 (Min et al., 2011). Among which, 13 big reservoirs have a vol-

ume larger than 1.0 � 108 m3, including Zhelin Reservoir(50.17 � 108 m3) on the Xiushui River, Wan’an Reservoir(11.16 � 108 m3) on the Ganjiang River, and Hongmen Reservoir(5.24 � 108 m3) on the Fuhe River (see Fig. 1). Most water conser-vancy projects have special functions for flood protection, watersupply, agriculture irrigation, power generation and navigation,which dramatically increase water utilization and change the tem-poral and spatial distribution of the streamflow, but they do nottransfer water out of the Poyang Lake catchment. The recent waterresources bulletins from Water Conservancy Bureau of JiangxiProvince indicated that annual water consumption is about94.38 � 108–126.95 � 108 m3 during 1997–2007, of which 73%for agriculture irrigation and 21% for domestic and industrial utili-zation. During the past decades, land use/land cover of the catch-ment has changed dramatically, which may have affected thechange of catchment hydrology indirectly. There is an indicationthat the forest coverage of the catchment was reduced from over60% in 1950s to only 32.7% in 1970s. The Grain for Green (returningfarmland to forest) policy was launched by the government in ear-lier 1980s in order to restore the ecological environment, and as aresult, the forest coverage recovered to nearly 60% at the end of1990s (Liu et al., 2009b). Human activities have collectively chan-ged the natural condition of catchment hydrology, and introducedadditional challenges for local water resources management.

2. Data and method

2.1. Available data

Observed discharges (daily streamflow) from five gauging sta-tions on the lower reaches of the five rivers were obtained fromManagement Bureau of the Yangtze River (MBYR) catchment,accounting for 75.4% of the lake’s inflow from the catchment area.Among the five gauging stations, Waizhou, Lijiadu and Meigang are

Table 1List of hydrological stations and their features for 1960–2007.

Gaugingstation

Location Drainagearea (km2)

Mean annual discharge(�108 m3/year)

Specific discharge(mm/year)

Coefficient ofvariation (Cv)

Ratio of annual extremestreamflow (Qmax/Qmin)

Waizhou Gangjiang River (Lower)(28�370N, 115�490E)

80,948 685 846 0.28 4.85

Lijiadu Fuhe River (28�120N, 116�090E) 15,811 123 780 0.33 4.31Meigang Xinjiang River (28�250N,

116�480E)15,535 178 1144 0.36 6.18

Hushan Raohe River (Le’anjiang branch)(28�540N, 117�180E)

6474 74 1137 0.32 3.93

Wanjiabu Xiushui River (Liaohe branch)(28�510N, 115�380E)

3548 35 990 0.35 4.69

86 X. Ye et al. / Journal of Hydrology 494 (2013) 83–95

located at the lower reaches of Ganjiang, Fuhe and Xinjiang rivers,which contribute more than 90% of the total inflow of the five riv-ers. While Hushan and Wanjiabu are located at the branches ofRaohe and Xiushui rivers with relative small drainage areas, con-tributing less than 10% of the total inflow of the five rivers. The ba-sic features of these gauging stations are listed in Table 1.

The meteorological data from 19 weather stations inside thecatchment (see Fig. 1) were obtained from National Climate Centreof China Meteorological Administration (CMA). They provide dailyobservations of precipitation, temperature, relative humidity, sun-shine duration, actual vapour pressure, and wind speed, amongothers. The period of record of most weather stations used in thisstudy is 1960–2007, except for Nanxiong, Lichuan, Linchuan andShangrao stations where the available data period is 1980–2007.Climate variables of these four weather stations are further inter-polated for the period 1960–1979 using the nearest weatherstation. Based on the meteorological datasets, potential evapo-transpiration (PET) of the weather stations was estimated byapplying the Penman–Monteith equation (Allen et al., 1998). Thedaily records of streamflow and climate variables provided byMBYR and CMA had gone through a standard quality control pro-cess before delivery, and with no missing data on the variablesused in this study.

Before applying the data, all the hydro-meteorological variableswere aggregated from daily to monthly and to yearly. In consider-ation of the large degree of variation in topography and the unevendistribution of weather stations across the catchment, an area-based weighting method was used to calculate the average precip-itation, potential evapotranspiration for the whole catchment aswell as the individual sub-catchments. The weight coefficient, ex-pressed by the percentage of the area represented by each meteo-rological station, was calculated using the Thiessen Polygonmethod. Similarly, annual streamflow for the whole catchmentwas calculated from the five individual hydro-stations by usingthe same method.

2.2. Statistical analysis

In order to analyze the temporal variation of the streamflowdata, the Partial Mann–Kendall test (hereafter we call it MK test),which is known as Kendall’s statistic (Kahya and Partal, 2007),was applied in this study. The MK test is a rank-based non-para-metric method that has been widely applied for trend detectingin hydro-climatic time series due to its robustness against theinfluence of abnormal data and especially its reliability for biasedvariables (e.g., Burn and Hag Elnur, 2002; Chen et al., 2007; Zhangand Lu, 2009). Before the trend analysis was performed, autocorre-lation (serialcorrelation) was examined through the autocorrelationand partial autocorrelation function for all the hydrological data,and results revealed (not shown here) that no significant autocor-relation existed in the data.

In which, unused To detect the existence of any stepchange points in the hydro-climatic data Xt = (x1,x2,x3 . . .xn),sequential Mann–Kendall test was used, in which, the accumula-tive number ni of samples that xi > xj (1 6 j 6 i) should be first cal-culated. The normally distributed statistic dk can be calculated viathe following formula:

dk ¼Xk

i¼1

ni ð2 6 k 6 nÞ ð1Þ

Mean and variance of the normally distributed statistic dk aregiven by

EðdkÞ ¼ kðk� 1Þ=4 ð2Þ

varðdkÞ ¼ kðk� 1Þð2kþ 5Þ=72 ð3Þ

The normalized variable statistic UF(dk) is estimated as follows:

UFðdkÞ ¼½dk � EðdkÞ�ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

varðdkÞp ðk ¼ 1;2;3 . . . ;nÞ ð4Þ

where UF(dk) is the forward sequence, and the backward sequenceUB(dk) is calculated using the same equation but with a reversed ser-ies of data. The null hypothesis (no step change point) is rejected ifany of the points in the forward sequence (UF(dk)) are outside theconfidence interval. The sequential Mann–Kendall test was oftenused to determine the approximate time of occurrence of the changepoint by locating the intersection of the forward and backward curvesof the test statistic. An intersection point of UF(dk) and UB(dk) locatedwithin the confidence interval indicates the beginning of a stepchange point (Moraes et al., 1998; Zhang et al., 2011c).

2.3. Estimating the relative impact of climate change and humanactivities on streamflow

For a natural catchment, annual water balance can be quantifiedas:

DS=Dt ¼ P � ET � Q ð5Þ

where P is the precipitation, ET is the actual evapotranspiration, Q isthe streamflow, DS is the change in water storage and Dt is the timestep. Over a long period of time (i.e., 10 years or more), DS can bereasonably assumed as zero, and then ET can be estimated as thedifference between P and Q.

Among the water balance components, ET is mainly controlledby available water (P) and available energy (or evaporative de-mand – PET) of a catchment, and of course by the human activitiessuch as the landuse change and farmland irrigation. Based on the P,PET and ET values in a catchment, a coupled water and energy bud-gets analysis is available to be used to evaluate the efficiency ofwater and energy use by an ecosystem (Milne et al., 2002; Tomerand Schilliing, 2009). In which, unused water (P–ET) and unused

X. Ye et al. / Journal of Hydrology 494 (2013) 83–95 87

energy (PET–ET) available in the environment can be used to calcu-late the proportions of available water and energy, i.e. Pex and Eex,that are unused (i.e., in excess) as:

Pex ¼ ðP � ETÞ=P ð6Þ

Eex ¼ ðPET � ETÞ=PET ð7Þ

where Pex and Eex take values from 0 to 1. Based on the analysis ofPex and Eex, Tomer and Schilliing (2009) developed a conceptualmodel to simply distinguish land-use and climate change effectson watershed hydrology (see Fig. 2 in Tomer and Schilliing, 2009).In which, climate change influences P and PET, and causes the in-crease of excess water (Pex) and decrease of excess energy (Eex), orvice versa. However, changes in vegetation or its management willdirectly affect ET, but not P or PET, which result in a regime shift inexcess water and energy with either increase or decrease, depend-ing on the effect of the change on ET.

The plots of excess water (Pex) versus excess energy (Eex) pro-vide an effective way to empirically distinguish the relative effectsof climate change and human activities on watershed hydrologyfor any two different periods. It should be noted that the applica-bility of the conceptual model is based on the assumptions that hu-man activities are independent of climate change, and the landusechange affects only ET. However, the effects of human activitiesand climate are commonly interrelated with each other indirectlyat broad scales. For example, in large catchments, climate changeeffects on hydrology have coincided with land cover changes(Zheng et al., 2009; Tomer and Schilliing, 2009), and constructionof river regulation system or big reservoirs will affect regional pre-cipitation and temperature and consequently results in changes inthe hydrological regime (Wu et al., 2006; Li et al., 2002). Despitethat human activities and climate system may interact with eachother, it is commonly not considered in most quantitative assess-ment studies (e.g. Li et al., 2007; Fan et al., 2010; Jiang et al.,2011; Wang et al., 2012).

In the Poyang Lake catchment, human activities are complexwhich include a wide range of landuse change, river regulationand rapid development of local society and economy. Due to thefact that specific human activities (e.g. land use change, farmlandirrigation and river regulation) may exist simultaneously in acatchment and their effects may accumulate or counteract eachother, we extended this assumption that all anthropogenic stressesin the Poyang Lake catchment will result in a change of ET. Then,changes associated with local climate and human impacts arelikely to result in a shift in excess water and energy, and the rela-

Fig. 3. Conceptual model of hydrological shifts associated with climate change andhuman activities (modified from Tomer and Schilliing, 2009).

tionship can be illustrated in Fig. 3. The directions of these hydro-logical shifts may indicate the relative impacts of climate changeand human activities on catchment hydrology. Here we appliedthis empirical method aimed to analyze the hydrological shifts inthe Lake catchment associated with climate change and humanactivities as an auxiliary analysis for demonstration of the relativeimportance of climate change and human activities, and the resultis afterwards inspected from a quantitative assessment.

Under the assumption that water balance is controlled bywater availability and atmospheric demand, Zhang et al. (2001)noted that long-term mean annual evapotranspiration has thefollowing relationship with local precipitation and potentialevapotranspiration:

ETP¼ 1þwðPET=PÞ

1þwðPET=PÞ þ ðPET=PÞ�1 ð8Þ

where w is a model parameter of available water coefficient relatedto vegetation type (Zhang et al., 2001), and can be calibrated usingannual hydro-climatic data.

Hydrologic sensitivity can be described as the percentagechange in mean annual runoff in response to changes in mean an-nual precipitation and potential evapotranspiration (Jones et al.,2006; Li et al., 2007). Variation in both precipitation and potentialevapotranspiration can lead to changes in water balance. It can beassumed that a change in mean annual runoff due to climatechange can be approximated as follows (Koster and Suarez,1999; Milly and Dunne, 2002):

DQc lim ¼ a � DP þ b � DPET ð9Þ

where DQclim, DP, DPET are the changes in streamflow, precipitationand potential evapotranspiration respectively; a and b are the sen-sitivity parameters and can be further expressed as (Li et al., 2007):

a ¼ 1þ 2xþ 3wx

ð1þ xþwx2Þ2ð10Þ

b ¼ 1þ 2wx

ð1þ xþwx2Þ2ð11Þ

where x is the dryness index (equal to PET/P), w is same as in Eq. (8).A change in mean annual streamflow can be calculated as

follows:

DQobs ¼ Q�

obs2 � Q�

obs1 ð12Þ

where DQobs indicates the observed change in mean annual stream-flow between two different periods, Q

�obs1 is the average annual

streamflow during the reference period, and Q�

obs2 is the average an-nual streamflow during the other period.

The change of catchment hydrology is mainly affected by thetwo driving factors of climate change and human activities. As afirst-order approximation, the change in mean annual runoff canbe estimated as follows:

DQobs ¼ DQ c lim þ DQhum ð13Þ

where DQclim and DQhum are the changes in the mean annualstreamflow due to climate change and human activities, respec-tively. The relative contributions of climate change and humanactivities on streamflow can be further expressed as:

gc lim ¼DQc lim

jDQ obsj� 100% ð14Þ

ghum ¼DQhum

jDQobsj� 100% ð15Þ

where =clim and =hum are the percentages of the impact of climatechange and human activities on streamflow, respectively.

Table 2Results of MK test for seven variables on seasonal and annual basis 1960–2007.

Parameter Annual mean Cv (%) Trends

Spring (MAM) Summer (JJA) Autumn (SON) Winter (DJF) Annual

T 17.7 3 2.30* 0.84 1.47 2.78* 3.40**

P 1656.6 15 �0.44 1.31 �1.02 1.68 1.11RH 78.5 2 �2.21* 0.12 �1.31 0.42 �0.56VP 9.9 0 �2.21* �0.12 �1.15 0.40 �1.59W 2.1 13 �7.54** �4.92** �6.66** �6.82** �7.83**

SD 4.8 10 �0.63 �4.24** �1.48 �2.85** �4.47**

PET 1050.6 5 0.67 �2.82** �2.04* �1.52 �2.59**

Note: � Delineates negative trends based on the MK test. Cv: coefficient of variation (%); P: precipitation (mm); T: temperature (�C); VP: vapour pressure (kPa); RH: relativehumidity (%); SD: sunshine duration (h); W: wind speed (m/s); PET: potential evapotranspiration (mm/y).* Delineate significance at 0.05 significance level.** Delineate significance at 0.01 significance level.

88 X. Ye et al. / Journal of Hydrology 494 (2013) 83–95

3. Results

3.1. Changes of catchment climate

In order to investigate the variation of regional climate in thePoyang Lake catchment, basin-scale averaged time-series 1960–2007 of six variables, i.e., precipitation, temperature, relativehumidity, sunshine duration, vapour pressure and wind speedwere examined by the MK test. The potential evapotranspiration(PET) estimated by Penmam–Monteith method was also analyzed.Results of the trend test for the seven variables are displayed in Ta-ble 2. It is seen that on annual basis, temperature and precipitationshow positive trends; relative humidity, vapour pressure, windspeed, sunshine duration and PET show negative trends. Amongwhich, four variables, i.e., temperature, wind speed, sunshine dura-tion and PET have undergone significant trends at 0.01 significancelevel. Precipitation and wind speed show the highest coefficient ofvariation with 15% and 13%, respectively, showing the highestvariability.

On seasonal basis, statistically significant (0.01) positive trendsof temperature were detected for spring and winter; however, thepositive trends in summer and autumn are not significant.Although precipitation shows an upward trend for summer andwinter, and downward trend for spring and autumn, no significant

Fig. 4. Time series of catchment annual precipitation (a) and PET (c) and corresponding Mthis period, and the horizontal dashed lines in the right figures represent the critical va

trend was detected through four seasons. Both relative humidityand vapour pressure decreased significantly in spring, but trendsin other seasons are not significant. Among the seven variables,wind speed is the only one which decreased significantly in all sea-sons. In summer and winter, sunshine duration decreased signifi-cantly during 1960–2007, but the negative trends in spring andautumn are not significant. PET decreased through all seasons ex-cept spring, and the negative trends in summer and autumn arestatistically significant at 0.01 significance level, respectively.

It can be summarized that most climate variables have experi-enced significant trends during the past decades. For illustrativepurposes, the variation of annual precipitation and calculated po-tential evapotranspiration (PET) series and their correspondingMK sequential test in 1960–2007 are shown in Fig. 4a, c and b, d,respectively. As depicted in Fig. 4a and c, catchment annual precip-itation and PET show a long-term increase and decrease lineartrend, respectively. Fig. 4b and d shows that an abrupt change ofthe two variables occurred in 1969 at 0.05 significance level asthe intersection point of the two curves located within the confi-dence interval. Catchment annual precipitation showed a decreas-ing trend from 1960 to 1969, while opposite trend was found forthe other periods. There is an obvious increasing tendency since1990, which is significant during the period 1999–2002 as the val-ues of UF are above the critical limit. On the contrary, the values of

K trend test (b and d). The long dashed line in the left figures means linear trend forlue of 0.05 significance level.

X. Ye et al. / Journal of Hydrology 494 (2013) 83–95 89

UF for catchment annual PET are below zero for almost the wholeperiod especially after 1969, and the long term decrease trend be-comes significant after 1982.

3.2. Variation of annual and monthly streamflows

Fig. 5 shows the MK test of the annual and monthly streamflowsof each station in the five sub-catchments. As shown in Fig. 5a, theUF curve of Waizhou station indicates a decreasing trend ofstreamflow from 1960 to 1972, while an increasing trend is found

Fig. 5. (a) Trends variation and change point test for annual streamflow; the horizontamonthly streamflow, the horizontal solid lines and dashed lines represent the critical va

for the other periods. The variation in annual streamflow is smallbefore 1990, and after that streamflow increased obviously. ForMeigang station, the UF curve exceeds zero for almost the wholeperiod and it intersects with the critical value lines for a short per-iod, which confirmed an increasing trend especially in the end of1990s. No obvious trend was detected for the streamflow at Lijiadustation with UF fluctuates between the two critical value lines.However, streamflow trend for Lijiadu station shows a slight de-crease during 1987–1996 with UF < 0, which is obviously differentfrom that of the other stations. Annual streamflow of Wanjiabu and

l dashed lines represent the critical value of 0.05 significance level; (b) trends oflues of 0.01 and 0.05 significance level, respectively.

90 X. Ye et al. / Journal of Hydrology 494 (2013) 83–95

Hushan stations has similar changing trends which increased foralmost the whole period. Furthermore, streamflow of these twostations increased significantly in 1970s and 1990s with UF curvesintersect with the critical value lines during these two periods.Change point analysis for annual streamflow indicates that exceptfor the Wanjiabu station where the intersection of UF and UBcurves occurred in 1969 (at 0.05 significance level), no clear stepchange can be identified for the other stations.

Estimated results for the long-term trends of annual streamflowindicate that most gauging stations show increasing trends exceptfor Lijiadu station, but monthly variations are obviously different.As shown in Fig. 5b, most stations present long-term decreasebut not significant trends during the wet season from April to June.However, increasing trends are detected for the other months. Ex-cept for Hushan station, streamflow of most stations increasesobviously at 0.05 or 0.01 significance level for August and Septem-ber. Also, streamflow of Wanjiabu and Hushan stations increasessignificantly in January at 0.01 significance level. The monthly var-iability of streamflow may indicate the change of seasonal climateregime and catchment management practices.

3.3. Hydrological response associated with changes in climate andhuman activities

The spatial and temporal variation of streamflow in Fig. 5 re-flects the combined effects of climate change and human activities,which provides valuable information in water resources planningand management in the catchment. Within the study period, wecannot distinguish a natural period during which the effect of hu-man activities on streamflow was less recognized like in most pre-vious studies (e.g. Li et al., 2007; Wang et al., 2012). In this study,attempts, therefore, are made to investigate the influence ofanthropogenic activities and climate change on the streamflow ofthe sub-catchments as well as the whole catchment under differ-ent decades. The changes of precipitation, PET, ET and streamflowrelative to baseline period 1960s for the whole Poyang Lake catch-ment and the three largest sub-catchments are analysed in detailsand the results are plotted in Fig. 6. Compared to 1960s, precipita-tion increased during all the other decades, especially in 1990s anincrease of 168.7–316.1 mm was detected for the three sub-catch-ments. The range of decrease in PET is 63.9–86.9 mm for Ganjiang,42.3–130.8 mm for Xinjiang and 48.0–83.4 mm for Fuhe. Change of

Fig. 6. Change of precipitation, PET, ET and stream

streamflow is consistent with precipitation except for a decrease of67.1 mm in Fuhe sub-catchment during 2000s. Calculated ET in-creases obviously in Fuhe especially in 1990s and 2000s with69.1 mm and 83.2 mm, respectively, but decreases in Xinjiang inrecent decades.

Based on the water-energy budget analysis, excess water (Pex)and excess energy (Eex) of the three sub-catchments as well asthe whole catchment were calculated under different decades.Using the background of regional climate condition and anthropo-genic stresses of 1960s as the reference baseline period (bench-mark), the shifts of Pex and Eex relative to 1960s are illustrated inFig. 7. In the figure, an increasing trend for Pex and a decreasingtrend for Eex are detected in the other decades for Ganjiang andXinjiang sub-catchments, as well as the whole catchment. Theseresults indicate that compared to the baseline period 1960s,changes of streamflow in 1970s, 1980s, 1990s and 2000s were pri-marily affected by climate change, while human activities played acomplementary role. However, there is some difference in the FuheRiver sub-catchment with decreasing Pex and Eex in 2000s, implyingthat relative changes of streamflow in 2000s were primarily af-fected by human activities rather than climate change.

3.4. Quantitative assessment of relative impacts of climate change andhuman activities on streamflow trends

The effect of climate change on streamflow can be estimatedusing the hydrologic sensitivity analysis method. In this method,w is the main model parameter that needs to be calibrated. Foreach river catchment, the calibration was conducted by comparingthe observed with simulated annual streamflow by using Eqs. (5)and (8) for the baseline period 1960–1969. Generally, optimizedw values are 0.40, 0.20, 1.20 and 0.45 for Ganjiang, Xinjiang, Fuheand the whole Poyang Lake catchment, respectively. As w repre-sents available water stress coefficient that related to vegetationtype (Zhang et al., 2001), optimized w values for each sub-catch-ment also indicate the inhomogeneous distribution of vegetationin the Poyang Lake catchment. Fig. 8 shows the correlation be-tween observed and simulated annual streamflow, in which, scat-ter points are concentrated around the 1:1 line. All the R2 valuesare greater than 0.82 with small mean absolute error (MAE),indicating that the simulated results are acceptable. With the opti-mized w values of Ganjiang, Xinjiang, Fuhe and the whole Poyang

flow compared to the baseline period 1960s.

Fig. 7. Variation of excess water (Pex) and excess evaporative demand (Eex) relative to 1960s.

Fig. 8. Correlation between observed and simulated annual streamflow for the baseline period 1960s.

X. Ye et al. / Journal of Hydrology 494 (2013) 83–95 91

Lake catchment, the calculated values for the sensitivity coefficienta were 0.86, 0.88, 0.96, 0.90 and b were�0.63, �0.75,�0.47,�0.64,respectively. In addition, the absolute values of sensitivity coeffi-cient also reveal that the change in streamflow was more sensitiveto precipitation (P) than to potential evapotranspiration (PET) inthis region.

Based on the method of hydrological sensitivity analysis, all thecalculated parameters were then used to estimate the impacts ofclimate change and human activities on variation of streamflowreference to 1960s. As presented in Table 3, impacts of climatechange across the catchment were generally positive for thestreamflow compared to the reference period of 1960s, i.e. climate

change resulted in an increased streamflow. Take the whole catch-ment as an example, changes in regional climate (precipitation andpotential evapotranspiration) in 1970–2007 were the main factorsthat increased runoff with a contribution of 150.7% relative to1960s, while the reduction percentages due to human activitieswere only 50.7%. Specific impacts of climate change on the increaseof annual streamflow were estimated to be 105.0 mm, 118.6 mm,261.7 mm and 75.3 mm for 1970s, 1980s, 1990s and 2000s, respec-tively (Table 3). On the contrary, the impacts of human activitieswere always negative on the water volume, leading to a decreaseof annual streamflow of 5.4 mm, 56.3 mm, 47.6 mm and 39.8 mmduring these periods. The proportional change in annual

Table 3Contribution of climate change and human activities on the change of annual streamflow compared to the baseline period 1960–1969.

Catchments Periods Q (mm) DQ (mm) Contribution of climate change Contribution of human activities

DQclim (mm) gclim (%) DQhum (mm) ghum (%)

Ganjiang 1960–1969 760.7 – – – – –1970–1979 876.3 115.7 119.3 103.2 �3.6 �3.21980–1989 813.0 52.3 108.2 206.6 �55.8 �106.61990–1999 954.8 180.3 231.4 128.4 �51.2 �28.42000–2007 822.9 62.2 116.1 186.5 �53.9 �86.5

Xinjiang 1960–1969 1017.3 – – – –1970–1979 1158.8 141.4 154.1 109.0 �12.7 �9.01980–1989 1100.4 83.1 120.7 145.3 �37.6 �45.31990–1999 1372.2 354.8 378.4 106.6 �23.6 �6.62000–2007 1046.9 29.5 64.8 219.6 �35.3 �119.6

Fuhe 1960–1969 749.4 – – – – –1970–1979 803.4 54.0 69.8 129.2 �15.8 �29.21980–1989 796.3 46.9 102.6 218.8 �55.7 �118.81990–1999 849.1 99.7 201.8 202.5 �102.2 �102.52000–2007 682.3 �67.1 38.2 �56.9 �105.3 156.9

The whole Poyang Lake Catchment 1960–1969 801.9 – – – –1970–1979 911.4 109.5 115.0 105.0 �5.4 �5.01980–1989 864.2 62.3 118.6 190.3 �56.3 �90.31990–1999 1029.6 214.1 261.7 122.3 �47.6 �22.32000–2007 837.4 35.5 75.3 212.1 �39.8 �112.11970–2007 914.5 112.6 169.7 150.7 �57.1 �50.7

Fig. 9. (a) The sensitivity of the potential evapotranspiration to the four major meteorological variables; (b) trends analysis by MK test. The horizontal dashed lines in theright figure represent the critical value of 0.01 significance level.

92 X. Ye et al. / Journal of Hydrology 494 (2013) 83–95

streamflow due to climate change (gclim) ranges from 105.0–212.1%, while the proportional change due to human activities(ghum) ranges from �5.0% to �112.1%. Relative effect of climatechange is greater than that of human activities, which is consistentwith the results from the coupled water-energy budgets analysis.

Due to the discrepancy in regional climate and the intensive hu-man activities in the catchment, the contribution of individual im-pacts varied under different spatial and temporal scales. Accordingto Table 3, increased streamflows in Ganjiang, Xinjiang and Fuhesub-catchments due to climate change (DQclim) were estimated to108.2–231.4 mm, 64.8–378.4 mm and 38.2–201.8 mm, while thespecific reduction impacts of human activities (DQhum) were 3.6–55.8 mm, 12.7–37.6 mm and 15.8–105.3 mm, respectively. Rela-tive to 1960s, climate change has the largest impact on the increasein streamflow in 1990s, while the impact is the least in 2000s,which is consistent with the increased frequency and severity offloods and droughts during these time periods (Min et al., 2011;Guo et al., 2007). However, percentage of the impact is bigger in1980s in Ganjiang River sub-catchment and in 2000s in XinjiangRiver sub-catchment. The Fuhe River sub-catchment is the onlycatchment where the reduction in streamflow during 2000s shouldbe mainly attributed to the intensive human activities. As shown inTable 3, the contribution of climate change accounted for 56.9% ofthe changes in streamflow, while human activities were responsi-ble for �156.9% of the change.

4. Discussions

The empirical estimation and hydrological sensitivity analysisperformed in this study are mainly for studying the effect of meanannual climate change on the variations in annual streamflow.Although the long-term trends of annual streamflow of the fivegauging stations are not significant, monthly variations were obvi-ous (as shown in Fig. 5b). Previous studies indicate that the increaseof precipitation, especially the increased frequency of summerstorm is the main reason for the increase of streamflow and for thecause of floods in 1990s (e.g., Guo et al., 2007). Additionally, in bothmethods, effect of climate change on streamflow is mainly throughthe changes in precipitation and potential evapotranspiration. Be-cause of the calculated potential evapotranspiration represents theintegrated effect of climate variables, further analysis indicated thatin the Poyang Lake catchment, the most important predictor for thedecreasing trend in potential evapotranspiration is the net radiationand wind speed, since they are not only sensitive variables in deter-mining PET, but also decrease significantly in the catchment (seeFig. 9). Comparable result was given in Xu et al. (2006), who con-cluded that the decreasing trend of reference evapotranspirationin the Yangtze River catchment is most attributed to the changesin net radiation and wind speed.

Like many regions in China, Poyang Lake catchment hasundergone intensive human activities since 1950s, including

Fig. 11. Changes of total reservoir amount and storage in the Poyang Lakecatchment.

X. Ye et al. / Journal of Hydrology 494 (2013) 83–95 93

afforestation and deforestation, land reclamation, river regulation,agriculture intensification and extensive infrastructure construc-tion, which may have exerted considerable impacts on catchmenthydrology. A modeling study in Xinjiang River sub-catchment indi-cated that when agriculture land changed into forest, which ac-counts for up to 23.3% of the catchment area, it would result in adecrease of annual discharge by up to 3.2% (Guo et al., 2008). Thedecreases of streamflow caused by increased forest cover are par-ticularly strong in the wet season through the increase of evapo-transpiration, but streamflow increase in the dry season has beenprimarily resulted from the increased groundwater contribution(Guo et al., 2007; Ye et al., 2011). Fig. 10 shows the changes of for-est coverage and the corresponding area of soil erosion in the Poy-ang Lake catchment. The increase of forest coverage since 1980smay play an important role in reducing streamflow during floodseason. In addition, the variation of annual streamflow has beenfurther elevated by the extensive water utilization according tothe local social and economic development. Due to the lack ofinformation of long-term water consumption, we here presentthe changes of reservoir construction and the storage volume inthe catchment, which may well indicate the variation of water uti-lization. As shown in Fig. 11, although there is litter change in thetotal amount of reservoirs in the catchment since 1980s, the vol-ume of water storage increased steadily for the same period, espe-cially after 1990s. The reasons for this are the rapid changes insociety and social life, population, economic forces and technolog-ical development, which increased the demand and the ability ofwater supply for industrial and domestic consumptions. The differ-ent changing patterns in annual streamflow of Lijiadu station inFuhe River sub-catchment since 1990s (Fig. 5a and Table 3) ascompared with other sub-catchments are the results of highwater-use efficiency. The biggest irrigation farmland as well asthe irrigation systems of Jiangxi Province are located in the middleand lower reach of Fuhe, which will notably increase the water uti-lization and directly decrease streamflow, especially in those dryyears. The results of this study indicated that the impact of humanactivities on the change of catchment streamflow was negativecompared to the baseline period of 1960s. However, the impactof human activities on water resources at local scale is complexand different human activities may accumulate or counteract eachother; attempts have not been made in this study to distinguishthem from each other.

It is worth of noting that the concept of the model by Tomer andSchilliing (2009) is based on long-term observations of landusechange in the catchment. Although, the application of empiricalmethod in the Poyang Lake catchment gives a comparative resultin distinguishing the effects of human activities and climatechange on catchment hydrology, the applicability and implicationsneed more documentation, as suggested by Tomer and Schilliing(2009), and this study contributes to this call.

Fig. 10. Changes of forest coverage and soil erosion in the Poyang Lake catchment.

In this study, the variability of climate indices is consistent withthat of the data time series. The 1960s were used as the baselineperiod to analyze the relative impacts of climate change and hu-man activities in other decades, which is because the intensity ofhuman activities in 1960s is relatively small. In addition, the cou-pled water-energy budget analysis is an empirical method whichheavily depends on the quality of P, PET and ET measurements orestimates, especially the latter two being the most difficult compo-nents to estimate. The possible influence of data error on the re-sults is yet to be investigated in the future study. Meteorologicaldata from 19 weather stations in the study area might not be suf-ficient coverage for such a large-scale catchment. Influence of res-ervoir storage on the estimation of mean annual ET is notconsidered. Additionally, some uncertainties also exist in thehydrologic sensitivity analysis method which separates the effectsof climate change and human activities on streamflow. The perfor-mance of the hydrologic sensitivity analysis depends on thestreamflow data of the long-term baseline period, with no effectof human activities, for model calibration. In reality, there was alack of detailed long-term observation data in the Poyang Lakecatchment to distinguish a natural period, and even during thebaseline period of 1960s, there were some human disturbancessuch as farmland irrigation, river regulation and land coverchanges. Although calibrated available water coefficient (w) in thisstudy well reflects the average vegetation condition of the catch-ment during 1960s, this could still affect the estimation resultsto some extent.

5. Conclusions

Under the background of global warming, increased regionalclimate change and human activities were identified as the mainfactors causing the floods and droughts in the Poyang Lake catch-ment, especially since 1990s. In this study, we performed anassessment of the relative effects of climate change and humanactivities on the changes of streamflow in the catchment duringthe past five decades. MK test indicated a long term increase trendof precipitation and a decrease trend of potential evapotranspira-tion. An increasing trend of annual streamflow of the catchmentexcept for Fuhe sub-catchment was detected. The findings suggestthat relative effects of climate change and human activities on thechange of streamflow varied among sub-catchments as well as thewhole catchment under different decades, due to the discrepanciesin regional climate and anthropogenic stresses across the PoyangLake catchment. The increase of annual streamflow (52.35–180.28 mm) was primarily affected by climate change for thewhole catchment compared to the reference period of 1960s, whilehuman activities played a complementary role. However, due tothe intensified water utilization for industrial and agricultural

94 X. Ye et al. / Journal of Hydrology 494 (2013) 83–95

development, especially in those dry years, the decrease of stream-flow (�67.13 mm) in the Fuhe sub-catchment in 2000s was pri-marily affected by human activities.

Quantitative assessment revealed that climate change resultedin an increase in runoff of 75.3–261.7 mm in 1970s–2000s forthe whole catchment, accounting for 105.0–212.1% of runoffchanges relative to 1960s. However, human activities may beresponsible for the decrease in runoff of 5.4–56.3 mm in 1970s–2000s, which accounts for �5.0% to �112.1% of runoff changes.Moreover, human activities such as farmland irrigation, river regu-lation and deforestation were the main anthropogenic stresses inaltering hydrological processes before 1990s. Afterwards, affores-tation and rapid local socio-economic development becameincreasingly apparent. However, the effects of specific humanactivities may accumulate or counteract each other, and attemptswere not made in this study to distinguish them from each other.

As a sub-tropical humid catchment with strong seasonal varia-tions of river discharge, current and future water resource manage-ment and planning in the Poyang Lake catchment may take a widerange of afforestation, soil conservation, and construction of waterconservancy engineering system into account. Implementation of apossible integrated water regulation system incorporating the Yan-gtze River is needed in order to reduce flood and drought disastersin the lake region and to maintain a healthy ecosystem of the lake.

Acknowledgements

This work is supported by the National Basic Research Programof China (2012CB417003 and 2012CB956103-5), National NaturalScience Foundation of China (41201026), State Key Laboratory ofLake Science and Environment (2010SKL014), and Science Founda-tion of Nanjing Institute of Geography and Limnology (NI-GLAS2012135001 and NIGLAS2010XK02). The authors extendtheir thanks to two anonymous reviewers for their constructivecomments, which greatly improved the quality of this paper.

References

Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop EvapotranspirationGuidelines for Computing Crop Water Requirements. FAO Irrigation andDrainage Paper No. 56. FAO, Rome.

Bates, B.C., Kundzewicz, Z.W., Wu, S., Palutikof, J.P., 2008. Climate Change andWater. Technical Paper of the Intergovernmental Panel on Climate Change. IPCCSecretariat, Geneva. p. 210.

Brown, A., Zhang, L., McMahon, T., Western, A., Vertessy, R., 2005. A review of pairedcatchment studies with reference to the seasonal flows and climate variability.J. Hydrol. 310, 28–61.

Burn, D.H., Hag Elnur, M.A., 2002. Detection of hydrologic trends and variability. J.Hydrol. 255, 107–122.

Carless, D., Whitehead, P. G., in press. The potential impacts of climate change onhydropower generation in Mid Wales. Hydrol. Res. http://dx.doi.org/10.2166/nh.2012.012.

Chen, H., Guo, S.L., Xu, C.Y., Singh, V.P., 2007. Historical temporal trends of hydro-climatic variables and runoff response to climate variability and their relevancein water resource management in the Hanjiang basin. J. Hydrol. 344, 171–184.

Déry, S.J., Hernández-Henríquez, M.A., Burford, J.E., Wood, E.F., 2009. Observationalevidence of an intensifying hydrological cycle in northern Canada. Geophys.Res. Lett. 36, L13402. http://dx.doi.org/10.1029/2009GL038852.

Du, J.K., Li, Q., Rui, H.Y., Zuo, T.H., Zheng, D.P., Xu, Y.P., Xu, C.-Y., 2012. Assessing theeffects of urbanization on annual runoff and flood events using an integratedhydrological modelling system for Qinhuai River basin, China. J. Hydrol. 464–465, 127–139.

Fan, J., Tian, F., Yang, Y.H., Han, S.M., Qiu, G.Y., 2010. Quantifying the magnitude ofthe impact of climate change and human activity on runoff decline in MianRiver Basin, China. Water Sci. Technol. 62, 783–791.

Guo, H., Hu, Q., Jiang, T., 2008. Annual and seasonal stream flow responses toclimate and land-cover changes in the Poyang Lake basin, China. J. Hydrol. 33,172–186.

Guo, H., Hu, Q., Zhang, Q., 2011. Changes in hydrological interactions of the YangtzeRiver and the Poyang Lake in China during 1957–2008. Acta Geograph. Sin. 66(5), 609–618 (in Chinese).

Guo, H., Su, B.D., Wang, Y.J., Jiang, T., 2007. Runoff coefficients change and theanalysis of the relationship between climate factors and runoff coefficients inPoyang Lake Basin (China):1955–2002. J. Lake Sci. 19 (2), 163–169 (in Chinese).

Huntington, T.G., 2006. Evidence for intensification of the global water cycle:review and synthesis. J. Hydrol. 319, 83–95.

IPCC, 2007. Climate change 2007: the physical science basis. In: Solomon, S., Qin, D.,Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.),Contribution of Working Group I to the Fourth Assessment Report of theIntergovernmental Panel on Climate Change. Cambridge University Press,Cambridge.

Jiang, S.H., Ren, L.L., Yong, B., Singh, V.P., Yang, X.L., Yuan, F., 2011. Quantifying theeffects of climate variability and human activities on runoff from the Laohahebasin in northern China using three different methods. Hydrol. Pro. 25, 2492–2505.

Jiang, S.H., Ren, L.L., Yong, B., Fu, C.B., Yang, X.L., 2012. Analyzing the effects ofclimate variability and human activities on runoff from the Laohahe basin innorthern China. Hydrol. Res. 43 (1–2), 3–13.

Jin, L., Whitehead, P.G., Futter, M.N., Lu, Z.L., 2012. Modelling the impacts of climatechange on flow and nitrate in the River Thames: assessing potential adaptationstrategies. Hydrol. Res. 43 (6), 902–916.

Jones, R., Chiew, F., Boughton, W., Zhang, L., 2006. Estimating the sensitivity of meanannual runoff to climate change using selected hydrological model. Adv. WaterRes. 29, 1419–1429.

Jung, G., Wagner, S., Kunstmann, H., 2012. Joint climate-hydrology modeling: animpact study for the data-sparse environment of the Volta Basin in West Africa.Hydrol. Res. 43 (3), 231–248.

Kahya, E., Partal, T., 2007. Is seasonal precipitation decreasing or increasing inTurkey? Online J. Earth Sci. 1 (1), 43–46.

Kizza, M., Guerrero, J.-L., Rodhe, A., Xu, C.-Y., Ntale, H.K., in press. Modellingcatchment inflows into Lake Victoria: regionalisation of the parameters of aconceptual water balance model. Hydrol. Res. http://dx.doi.org/10.2166/nh.2012.152.

Koster, R.D., Suarez, M.J., 1999. A simple framework for examining the interannualvariability of land surface moisture fluxes. J. Clim. 12, 1911–1917.

Lahmer, W., Pfiitzner, B., Becker, A., 2001. Assessment of land use and climatechange impacts on the mesoscale. Phys. Chem. Earth (B) 26, 565–575.

Lakshmi, V., Hong, S.B., Small, E.E., Chen, F., 2012. The influence of the land surfaceon hydrometeorology and ecology: new advances from modeling and satelliteremote sensing. Hydrol. Res. 42 (2–3), 95–112.

Legesse, D., Vallet-Coulomb, C., Gasse, F., 2003. Hydrological response of acatchment to climate and land use changes in Tropical Africa: case studySouth Central Ethiopia. J. Hydrol. 275, 67–85.

Li, D.F., Ning, D.T., Liu, C.M., Yang, Z.F., 2002. Impacts of west route project of south-to-north water transfer on local climate of exported region: a preliminaryanalysis. J. Nat. Resour. 17 (1), 16–21 (in Chinese).

Li, L., Ngongondo, C.S., Xu, C.-Y., Gong, L., in press. Comparison of the global TRMMand WFD precipitation datasets in driving a large-scale hydrological model inSouthern Africa. Hydrol. Res. http://dx.doi.org/10.2166/nh.2012.175.

Li, L.J., Zhang, L., Wang, H., Wang, J., Yang, J.W., Jiang, D.J., Li, J.Y., Qin, D.Y., 2007.Assessing the impact of climate variability and human activities on streamflowfrom the Wuding River basin in China. Hydrol. Process. 21, 3485–3491.

Liu, Q., Yang, Z., Cui, B., Sun, T., 2009a. Temporal trends of hydro-climatic variablesand runoff response to climatic variability and vegetation changes in the YiluoRiver basin, China. Hydrol. Process. 23, 3030–3039.

Liu, J., Zhang, Q., Xu, C.Y., Zhang, Z.X., 2009b. Research on runoff variationcharacteristics of Poyang Lake watershed in the past 50 years. Trop. Geogr. 29(3), 213–224 (in Chinese).

Ma, Z., Kang, S., Zhang, L., Tong, L., Su, X., 2008. Analysis of impacts of climatevariability and human activity on streamflow for a river basin in arid region ofnorthwest China. J. Hydrol. 352, 239–249.

Milliman, J.D., Farnsworth, K.L., Jones, P.D., Xu, K.H., Smith, L.C., 2008. Climatic andanthropogenic factors affecting river discharge to the global ocean, 1951–2000.Global Planet. Change 62, 187–194.

Milly, P.C.D., Dunne, K.A., 2002. Macroscale water fluxes 2. Water and energy supplycontrol of their inter-annual variability. Water Resour. Res. 38 (10), 1206.http://dx.doi.org/10.1029/2001WR000760.

Milne, B.T., Gupta, V.K., Restrepo, C., 2002. A scale invariant coupling of plants,water, energy, and terrain. Ecoscience 9 (2), 191–199.

Min, Q., 2002. Analysis on the flood characters in 1990s, Poyang Lake. J. Lake Sci. 14(4), 323–330 (in Chinese).

Min, Q., Shi, J.G., Min, D., 2011. Characteristics of sediment into and out of PoyanghuLake from 1956 to 2005. J. China Hydrol. 31 (1), 54–58 (in Chinese).

Moraes, J.M., Pellegrino, H.Q., Ballester, M.V., Martinelli, L.A., Victoria, R., Krusche,A.V., 1998. Trends in hydrological parameters of a southern Brazilian watershedand its relation to human induced changes. Water Resour. Manage. 12, 295–311.

Piao, S., Ciais, P., Huang, Y., Shen, Z., Peng, S., Li, J., Zhou, L., Liu, H., Ma, Y.,Ding, Y., Friedlingstein, P., Liu, C., Tan, K., Yu, Y., Zhang, T., Fang, J., 2010. Theimpacts of climate change on water resources and agriculture in China. Nature467, 43–51.

Raymond, P.A., Oh, N.H., Turner, R.E., Broussard, W., 2008. Anthropogenicallyenhanced fluxes of water and carbon from the Mississippi River. Nature 451(24), 449–452.

Ren, L.L., Wang, M.R., Li, C.H., Zhang, W., 2002. Impacts of human activity on riverrunoff in the northern area of China. J. Hydrol. 261, 204–217.

Schilling, K.E., 2004. Relation of baseflow to row crop intensity in Iowa. Agric.Ecosyst. Environ. 105, 433–438.

Shankman, D., Keim, B.D., Song, J., 2006. Flood frequency in China’s Poyang Lakeregion: trends and teleconnections. Int. J. Climatol. 26, 1255–1266.

X. Ye et al. / Journal of Hydrology 494 (2013) 83–95 95

Siriwardena, L., Finlayson, B.L., McMahon, T.A., 2006. The impact of land use changeon catchment hydrology in large catchment: the Comet River, centralQueensland, Australia. J. Hydrol. 326, 199–214.

St. Jacques, J.M., Sauchyn, D.J., Zhao, Y., 2010. Northern Rocky Mountain streamflowrecords: global warming trends, human impacts or natural variability?Geophys. Res. Lett. 37, L06407. http://dx.doi.org/10.1029/2009GL042045.

Tao, H., Gemmer, M., Bai, Y.G., Su, B.D., Mao, W.Y., 2011. Trends of streamflow in theTarim River Basin during the past 50 years: human impact or climate change? J.Hydrol. http://dx.doi.org/10.1016/j.jhydrol.2011.01.016.

Thompson, J.R., 2012. Modelling the impacts of climate change on uplandcatchments in southwest Scotland using MIKE SHE and the UKCP09probabilistic projections. Hydrol. Res. 43 (4), 507–530.

Tomer, M.D., Meek, D.W., Kramer, L.A., 2005. Agricultural practices influence flowregimes of headwater streams in western Iowa. J. Environ. Qual. 34 (5), 1547–1558.

Tomer, M.D., Schilliing, K.E., 2009. A simple approach to distinguish land-use andclimate-change effects on watershed hydrology. J. Hydrol. 376, 24–33.

Wang, W.J., Shao, Q.X., Yang, T., Peng, S.Z., Xing, W.Q., Sun, F.C., Luo, Y.F., 2012.Quantitative assessment of the impact of climate variability and humanactivities on runoff changes: a case study in four catchments of the HaiheRiver basin, China. Hydrol. Process. http://dx.doi.org/10.1002/hyp.9299.

Wang, F., Wu, D.Y., Li, R.F., 2008. Analysis on flood disaster characteristic in LakePoyang region. J. Lake Sci. 20 (4), 500–506 (in Chinese).

Wang, J., Meng, J.J., 2008. Characteristics and tendencies of annual runoff variationsin the Heihe River basin during the past 60 years. Sci. Geogr. Sin. 28 (1), 83–88(in Chinese).

Wu, L.G., Zhang, Q., Jiang, Z.H., 2006. Three Gorges Dam affects regionalprecipitation. Geophys. Res. Lett. http://dx.doi.org/10.1029/2006GL026780.

Xiong, L.H., Yu, K.X., Zhang, H.G., Zhang, L.P., in press. Annual runoff change in theheadstream of Yangtze River and its relation to precipitation and airtemperature. Hydrol. Res. http://dx.doi.org/10.2166/nh.2012.120.

Xu, C.Y., Gong, L.B., Jiang, T., Chen, D.L., Singh, V.P., 2006. Analysis of spatialdistribution and temporal trend of reference evapotranspiration in Changjiang(Yangtze River) catchment. J. Hydrol. 327, 81–93.

Xu, D.L., Xiong, M., Zhang, J., 2001. Analysis of hydrological characteristic of PoyangLake. Yangtze River 32 (2), 21–23 (in Chinese).

Yang, C.G., Yu, Z.B., Hao, Z.C., Zhang, J.Y., Zhu, J.T., 2012a. Impact of climate changeon flood and drought events in Huaihe River Basin, China. Hydrol. Res. 43 (1–2),14–22.

Yang, D., Li, C., Hu, H., Lei, Z., Yang, S., Kusuda, T., Koike, T., Musiake, K., 2004.Analysis of water resources variability in the Yellow River of China during the

last half century using historical data. Water Resour. Res. 40, W06502. http://dx.doi.org/10.1029/2003WR002763.

Yang, X.L., Ren, L.L., Singh, V.P., Liu, X.F., Yuan, F., Jiang, S.H., Yong, B., 2012b. Impactsof land use and land cover changes on evapotranspiration and runoff atShalamulun River watershed, China. Hydrol. Res. 43 (1–2), 23–37.

Ye, X.C., Zhang, Q., Liu, J., Li, L.J., Guo, H., 2009. Impacts of climate change and humanactivities on runoff of Poyang Lake catchment. J. Glaciol. Geocryol. 31 (5), 835–842 (in Chinese).

Ye, X.C., Liu, J., Zhang, Q., 2011. A modeling study of hydrological response to landuse changes based on hypothetical scenarios for the Poyang Lake catchment.ISWREP 2011. http://dx.doi.org/10.1109/ISWREP.2011.5893066.

Zhan, C.S., Niu, C.W., Song, X.M., Xu, C.-Y., in press. The impacts of climate variabilityand human activities on streamflow in Bai River basin, northern China. Hydrol.Res. http://dx.doi.org/10.2166/nh.2012.146.

Zhang, L., Dawes, W.R., Walker, G.R., 2001. The response of mean annualevapotranspiration to vegetation changes at catchment scale. Water Resour.Res. 37, 701–708.

Zhang, Q., Li, L., Wang, Y.-G., Werner, A.D., Xin, P., Jiang, T., Barry, D.A., 2012. Has theThree-Gorges Dam made the Poyang Lake wetlands wetter and drier? Geophys.Res. Lett. 39, L20402. http://dx.doi.org/10.1029/2012GL053431.

Zhang, Q., Sun, P., Jiang, T., 2011b. Changing properties, causes and impacts ofextreme streamflow in Lake Poyang basin, China. J. Lake Sci. 23 (3), 445–453 (inChinese).

Zhang, S.R., Lu, X.X., 2009. Hydrological responses to precipitation variation anddiverse human activities in a mountainous tributary of the lower Xijiang.Catena 77, 130–142.

Zhang, Y., Guan, D., Jin, C., Wang, A., Wu, J., Yuan, F., 2011c. Analysis of impacts ofclimate variability and human activity on streamflow for a river basin innortheast China. J. Hydrol. http://dx.doi.org/10.1016/j.jhydrol.2011.09.023.

Zhang, Z.X., Chen, X., Xu, C.-Y., Yuan, L.F., Yong, B., Yan, S.F., 2011a. Evaluating thenon-stationary relationship between precipitation and streamflow in NineMajor Basins of China during the past 50 years. J. Hydrol. 409, 81–93.

Zhao, G.J., Hörmann, G., Fohrer, N., Zhang, Z.X., Zhai, J.Q., 2009. Streamflow trendsand climate variability impacts in Poyang Lake basin, China. Water Resour.Manage. http://dx.doi.org/10.1007/s11269-009-9465-7.

Zheng, H., Zhang, L., Zhu, R., Liu, C., Sato, Y., Fukushima, Y., 2009. Responses ofstreamflow to climate and land surface change in the headwaters of the YellowRiver Basin. Water Resour. Res. 45, W00A19. http://dx.doi.org/10.1029/2007WR006665.