Science of the Total Environment 693 (2019) 133556

Contents lists available at ScienceDirect

Science of the Total Environment

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

Hydrologic alteration and possible underlying causes in the Wuding River, China

Xiaojing Tian a,b, Guangju Zhao a,b,⁎, Xingmin Mu a,b, Pengfei Zhang a,b, Peng Tian a,c, Peng Gao a,b, Wenyi Sun a,b a State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, Northwest A&F University, Yangling, 712100, China b Institute of Soil and Water Conservation, Chinese Academy of Sciences & Ministry of Water Resources, Yangling, Shaanxi 712100, China c College of Natural Resources and Environment, Northwest A&F University, Yangling, Shaanxi 712100, China

HIGHLIGHTS GRAPHICAL ABSTRACT

• We assessed the hydrologic alteration and investigated possible underlying causes. • The hydrological regime altered highly since 1970s. • Index of Connectivity decreased gradu- ally, whereas index of check dams/res- ervoirs increasing. • Degree of hydrologic alteration was more sensitive to the land use changes. • Land use changes and construction of check dams/reservoirs greatly affected hydrological regime.

article info abstract

Article history: Understanding hydrological alteration of rivers and the potential driving factors are crucial for water resources Received 22 April 2019 management in the watershed. This study analyzed the daily runoff time series at six gauging stations during Received in revised form 6 July 2019 1960–2016 in Wuding River basin, northwestern China. The Mann–Kendall test and Lee-Heghinian method Accepted 22 July 2019 were employed to detect the temporal trends and abrupt changes in annual streamflow. The flow duration Available online 23 July 2019 curve (FDC) and the index of hydrologic alteration (IHA)/Range of Variability Approach (RVA) were applied to fl Editor: Ralf Ludwig assess the daily stream ow and degree of hydrologic alteration (DHA). In addition, we analyzed the changes of index of hydrological connectivity (IC) and reservoirs/dams (RI) in 1990, 1995, 2000 and 2015 in the basin. Keywords: The relationship between IC, RI and DHA were assessed to investigate the potential influences of land use changes Hydrologic alteration and constructions of reservoirs/dams on hydrological alteration. The results indicated that annual streamflow at IHA/RVA five stations showed significant downward trends (p b 0.01) from 1960 to 2016, and an abrupt changing point Temporal variation appeared in the beginning of 1970s in Wuding River basin. Exception is Qingyangcha station without significant Underlying causes changes, and Hanjiamao station with changing point in 1967. FDC analysis indicated that both high and low flow Wuding River basin indices reduced greatly. The integral DHA were higher than 70% at all the stations in the Wuding River basin, sug- gesting great variation in the magnitude, duration, frequency, timing and rate of change of daily streamflow. Both IC value and RI had close relationship with DHA, implying that DHA was highly affected by land use changes and dams/reservoirs constructions, and was more sensitive to the land use change (p b 0.01). This study provides good insight to understand the effects of soil and water conservation measures on hydrological regime. © 2019 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, Institute of Soil and Water Conservation, Northwest A&F University, Yangling, Shaanxi 712100, China. E-mail address: [email protected] (G. Zhao).

https://doi.org/10.1016/j.scitotenv.2019.07.362 0048-9697/© 2019 Elsevier B.V. All rights reserved. 2 X. Tian et al. / Science of the Total Environment 693 (2019) 133556

1. Introduction which the ecological restoration measures were responsible for the sig- nificant reduction in annual streamflow. There are several studies focus- Water resources always play a critical role in the process of social de- ing on runoff and sediment load variation and their potential driving velopment of human being and ecological civilization construction factors (Xu, 2009; Zhang et al., 2010). However, very few studies paid (Zhao et al., 2015; Forootan et al., 2019). The spatiotemporal distribu- attention to hydrological regimes alteration. tion of water resources in many basins around the word is influenced Therefore, the objectives of this study were to analyze temporal var- by uneven precipitation and complex topography, and has experienced iation of the magnitude, duration, frequency, timing and rate of change significant changes in the past few decades (Gao and Huang, 2001; Piao of discharge by using a long-term series with IHA/RVA approach. The et al., 2010). These changes are related to many aspects—including potential effects of human activities on hydrological alteration were in- water supply and allocation, agricultural development, flood disaster, vestigated by assessing the relationship between hydrological indices droughts, ecological restoration, and they directly influence regional and hillslope connectivity and reservoir index. The finding of this sustainable development (Li et al., 2009; Li et al., 2017). It is worth not- study can provide a good reference for water resources allocation and ing that the reduction of surface/underground water will aggravate the future soil and water conservation. degree of drought and decrease agricultural production, as well as dete- rioration of ecological environment in arid areas (Van Loon and Laaha, 2. Study area and dataset 2015; Wanders and Wada, 2015). Therefore, assessing hydrological al- teration at regional and local scales is valuable for water redistribution 2.1. Study area and allocation. The most effective approach for assessing hydrologic alteration is The Wuding River, with a total length of 491.1 km and a drainage IHA/RVA, which includes 33 indices in five groups and analyzed hy- area of 30,261 km2,isafirst order tributary of the in drological alteration comprehensively and deeply (Richter et al., China (Fig. 1). The river originates from the north of Shaanxi Prov- 1996). In recent years, IHA/RVA has drawn more attention from ince, flows through the Autonomous Region, and many researchers in ecological hydrology and was widely applied discharged into the Yellow River in Shaanxi Province. It has several to many basins in the world (Gao et al., 2009; Lin et al., 2016). branches of Hailiutu, Luhe and Dali Rivers (Fig. 1). The watershed Pfeiffer and Ionita (2017) found that some hydrological indices has typical temperate continental arid and semi-arid climate pattern were highly altered and hydrological regime changed significantly with mean annual temperature and precipitation of 7.9–11.2 °C and since the middle of the 20th century, a probable consequence of cli- 491.1 mm, respectively. Most of precipitation concentrated between mate change, in the Elbe River and Rhine River. Yang et al. (2008) June and September with rainstorms. Approximately 45.65% of the evaluated the degree of hydrologic alteration in the middle and basin is covered by fine loess in the south and eastern part, and lower Yellow River and results showed that hydrologic regime al- 54.35% is relative coarse sandy area (Musu desert). The average an- tered highly at Huayuankou gauging station causing by Xiaolangdi nual soil erosion rate is higher than 8000 t/km2/a, and the annual reservoir. The alteration of the hydrologic regime across the Pearl runoff depth is 36.54 mm/a. In the basin, arable land, grassland and River Delta is moderate and high in monthly mean maximum and desert area dominated the land use and cover, and forest land minimum water level (Zhang et al., 2009). A study in the accounted for b10% of the whole basin. River showed that the hydrological regime exhibited high degree of alteration in the middle and lower reaches due to the construction 2.2. Data collection of Three Gorges Dam (Guo et al., 2019). Since the 1950s, large amount of soil and water conservation mea- The daily runoff at six hydrological stations (Fig. 1): Baijiachuan sures, i.e. check dams and terraces, and converting slope arable land to (BJC), Suide (SD), Lijiahe (LJH), Qingyangcha (QYC), Hanjiamao forest and grass, have been carried out to improve vegetation cover (HJM) and Dingjiagou (DJG) from 1960 to 2016 were collected (Xu, 2003; Huang and Zhang, 2004; Mu et al., 2007; Zhang et al., from the Hydrological Year book of the Yellow River basin, which is 2008; Gao et al., 2013). Consequently, these measures greatly affected published by the Yellow River Conservancy Committee (Table 1). land use and vegetation cover, thereby altered hydrological regime in The data quality of the daily discharge has been checked by the offi- the Yellow River. Numerous studies have investigated the variation of cial authorities. There were no gaps among these data at all the hy- streamflow and sediment discharge in response to climate change and drological stations. human activities to understand the changing characteristics and poten- The Digital Elevation Model (DEM) with a resolution of 90 × 90 m tial driving factors (Liu and Zheng, 2004; Wang et al., 2013; Zhao et al., was downloaded from USGS (http://www.usgs.gov). The land use 2014). Zhao et al. (2015) found that annual streamflow along the main data in four different periods (1990, 2000, 2005 and 2015) were ob- Yellow River gauging stations presented significant decreasing trends (p tained from the Data-Sharing Network of China Earth System Sci- b 0.01), and an abrupt-change point was examined in 1986 due to large ence (http://www.geodata.cn). The Yellow River Conservancy reservoir operation. Gao et al. (2013) showed that from 1932 to 2008 in Committee provided spatial distribution of check dams and reser- the the contribution rate of human activity to streamflow and voirs, including the locations, storage capacity, construction dates, sediment discharge was found to be 82.8% and 95.6%, respectively. Li height of dams, status of operating and etc. These data were ob- et al. (2009) addressed that climate variability influenced the surface tained from field survey by local government in 2011. All the data hydrology more significantly than land use changes in the Heihe catch- used in the study have been checked to guarantee their consistency ment during 1981–2000. These findings can provide deep insights on and quality. the effects of soil and water conservation measures on river runoff. The Wuding River is one of the large tributaries in the middle 3. Method reaches of the Yellow River. In the past few decades, annual runoff showed significant decreasing trend as a consequence of climate change 3.1. Non-parametric Mann-Kendall test and Lee-Heghinian and human activities. The study from Zhou et al. (2012) suggested that the annual streamflow was continuously decreasing, and abrupt- In this study, the non-parametric Mann-Kendall test (MK test) was changing points occurred in 1971 and 1997. They also found that employed to examine the changing trends and significant level for an- human activities were the main influencing factors for streamflow re- nual runoff in the Wuding River basin during 1960–2016, which was duction during the period of 1972–2009. Liang et al. (2015) suggested proposed by Mann (1945) and Kendall (1955). The MK test has been that streamflow variability is mainly sensitive to human activities, of commonly applied for trend detection due to its robustness for non- X. Tian et al. / Science of the Total Environment 693 (2019) 133556 3

Fig. 1. Study area and location of hydrological stations in the Wuding River basin. normally distributed time series. The test statistics S can be expressed Lee-Heghinian test (Lee and Heghinian, 1977) is based on the theory as: of Bayes and was applied to determine potential abrupt-change year in the hydrological time series during 1960–2016 in the Wuding River nX−1 Xn basin. It is assumed that population distribution is Gaussian distribution ¼ − ; b b ; − S sgn x j xk k j n where sgn x j xk and the period distribution of abrupt changes (τ) is uniform distribu- − − þ 8k 1 j k 1 tion. After calculating posterior distribution f(τ) of all possible points < 1; x Nx j k in a given period, abrupt change can be identified by the peaks of ¼ 0; x j ¼ xk ð1Þ : curve of f(τ). Its advantage is that outliers presented in time series will −1; x jbxk be removed. The posterior distribution f(τ) is expressed as: where sgn is the sign function, n is the length of the time series, and x , x j k 1½ðÞτ −ðÞn−2 2 ðÞ≤τ≤ − are observed value. S is a normal distribution and its mean value is 0, R 1 n 1 f ðÞ¼τjx1; x2; ⋯; xn kn½=τðÞn−τ 2 ð4Þ then hiXn τ 2 n 2 2 nnðÞ−1 ðÞ2n þ 5 R ¼ ∑ ¼ ðÞxt −xτ þ ∑ ¼τþ ðÞxt−xn−τ = ðÞxt −xn ð5Þ varðÞ¼S ð2Þ t 1 t 1 18 t¼1 where k is a proportional constant, n is sample size. If τ satisfies A Z statistic obtained from MK test can expressed as: max1≤τ≤n−1{f(τ|x1,x2,⋯,xn)}, it will possibly be an abrupt-change point.

¼ S ð Þ Z : 3 3.2. Range of variability approach and degree of hydrologic alteration ½varðÞS 0 5 The index of hydrologic alteration (IHA) was used to evaluate the A negative value of Z indicates a downward trend, and vice versa. degree of hydrologic alteration in terms of flow magnitude, timing,

Table 1 Information about hydrological stations in this study.

Station Longitude (E) Latitude (N) Drainage area (km2) Annual streamflow (m3/s) Time span

Average Std Cv

Baijiachuan (BJC) 110.42 37.23 29,662 12,322.24 3580.71 0.29 1960–2016 Suide (SD) 110.23 37.50 3893 1572.02 558.04 0.35 1960–2016 Qingyangcha (QYC) 109.22 37.37 662 288.77 114.24 0.40 1960–2016 Lijiahe (LJH) 109.83 37.62 807 309.14 109.70 0.35 1960–2016 Hanjiamao (HJM) 109.15 38.07 2452 957.39 228.47 0.24 1960–2016 Dingjiagou (DJG) 110.25 37.55 23,422 9584.4 2604.02 0.27 1960–2016 4 X. Tian et al. / Science of the Total Environment 693 (2019) 133556

Fig. 2. Changing trends in the annual runoff during 1960–2016 at six gauging stations in the Wuding River basin. BJC (a), SD (b), QYC (c), LJH (d), HJM (e), DJG (f). frequency, duration and rate of change with 33 hydrological variables 3.3. Index of Connectivity classified into five groups. Range of variability approach (RVA) was used to assess the changes of streamflow in different periods, and thus Hydrological processes can be influenced by land use/cover changes was applied to analyze the DHA before and after the changing points on the hillslope and hydraulic structures in the stream. Landscape het- (Richter et al., 1996). The estimation was consisted of four steps: (1) a erogeneities have direct effects on hydrological connectivity, and may total of 33 IHA indicators were calculated by using the daily time series alter runoff volume and flow movement path. Dams and reservoirs in the pre-abrupt time period, (2) the RVA target range of each IHA in- along the river can trap large amount of water in rainy season and re- dicator can be defined as 75% and 25% of their probability, (3) and then lease water for irrigation. Thus, we selected both hillslope hydrological the IHA indices were estimated with daily streamflow in the post- connectivity index and reservoir index to identify their potential effects abrupt change period, (4) the degree of hydrologic alteration were fi- on hydrological alternation. nally evaluated with RVA target range, and can be expressed as: The Index of Connectivity (IC) denotes the probability of a unit amount of material at a catchment scale from source through slopes − Y0i Y f to channels/sinks (Borselli et al., 2008). It is consisted of upslope and Di ¼ 100% ð6Þ Y f downslope components in the landscape, and can be expressed as:

where Di is the degree of hydrologic alteration of ith IHA, Y0i and Yf are 0 the number of observed and expected years of ith IHA locating in RVA pffiffiffi D BWS A target range after change. If D between 0 and 33% denotes low-degree ¼ up ¼ B Þ ð Þ i IC log10 log10@P 7 change in streamflow time series, and between 67% and 100% means Ddn di high-degree change, else moderate-degree change. wisi

Fig. 3. Abrupt-change years for annual runoff at six gauging stations during 1960–2016 in the Wuding River basin. BJC (a), SD (b), QYC (c), LJH (d), HJM (e), DJG (f). X. Tian et al. / Science of the Total Environment 693 (2019) 133556 5

Table 2 4. Results Changes for the average annual runoff in different periods at six stations in the Wuding River basin. 4.1. Changes in annual streamflow Station Abrupt-change year Mean annual runoff (m3/s) Changes (%) fl Baseline period Changing period 4.1.1. Temporal variation in annual stream ow As shown in Fig. 2,significant downward trends (p b 0.01) were ex- BJC 1971 17,665.39 11,044.53 −37.48 SD 1971 2165.22 1430.17 −33.95 amined in the annual runoff from 1960 to 2016 at six gauging stations in QYC 1970 391.76 291.56 −25.58 the Wuding River basin. Relative high reduction rates were found at BJC − − LJH 1970 424.02 259.99 −38.68 (0.475 mm·a 1) and DJG (0.411 mm·a 1)stations(Fig. 2a and f), re- HJM 1967 1368.90 899.78 −34.27 spectively. However, the decreasing trend was not significant (Z value DJG 1970 13,891.90 8669.35 −37.59 of-0.38 from MK test) in annual runoff at QYC station with the reducing rate of (0.0345 mm·a−1)(Fig. 2c). It can be clearly seen that the annual runoff had a sharp reduction in the end of 1960s, and remained relative where Dup is the upslope component, denoting the possibility of water stable during 1970–2010 (Fig. 2c). from the upstream slope moving downstream; Ddn is the downslope component, denoting the probability that water reaches the nearest set- 4.1.2. Abrupt changes in annual streamflow tlement point downstream through the flow path; W is the average Fig. 3 exhibited the Lee-Heghinian test for annual runoff. The poste- weighing factor of the upslope contributing area (dimensionless); S is rior distribution curve of f(τ) within the period of 1960–2016 suggested the average slope gradient of the upslope contributing area (m/m); A significant changes occurred in 1971 at BJC and SD station (Fig. 3aand 2 is the upslope contributing area (m ); di is the length of the ith cell b), in 1970 at QYC, LJH, and DJG stations (Fig. 3c, d and f), and around along the downslope path (m); wi is the weight of the ith cell (dimen- 1967 at HJM station (Fig. 3e). Another abrupt change can be seen in τ sionless); and si is the slope gradient of the ith cell (m/m). To avoid ex- 2010 at QYC station, while its f( ) value is less than that in 1970. Accord- fl treme and error value, si = 0.05 (the upper slope is si = 1) is used as ing to the abrupt changing points, we divided the stream ow time se- critical value to replace when slope is b0.05. In addition, IC is defined ries into two periods. The former one can be regarded as baseline in the range of [−∞,+∞] and connectivity increases when IC grows to- period with relatively limited human activities influences on hydrolog- wards +∞. ical regime, and the latter is changing period with strong human activ- The reservoir index (RI) is a dimensionless indicator proposed by ities impacts. Lopez and Frances (2013), which represent as: Table 2 showed the average annual runoff within two periods. We can clearly see that the average annual runoff in the changing period re- duced by 37.48%, 33.95%, 25.58%, 36.68%, 34.27% and 37.59%, respec- XN A C tively, at BJC, SD, QYC, LJH, HJM and DJG stations compared to those in RI ¼ i i ð8Þ A C baseline period (Table 2). This indicated that the annual runoff has i¼1 T T been greatly altered since 1970s in the Wuding River basin.

where N is the number of reservoirs upstream of the gauge station, Ai is 4.2. Daily runoff changes within different periods the catchment area of each reservoir, AT is the catchment area of the gauge station, Ci is the total capacity of each reservoir, and CT is the The flow duration curve (FDC) method is a simple, but comprehen- mean annual streamflow at the gauge station. sive graphical view of the overall variability of the cumulative

Fig. 4. FDCs for daily runoff in different periods in the Wuding River. BJC (a), SD (b), QYC (c), LJH (d), HJM (e), DJG (f). 6 X. Tian et al. / Science of the Total Environment 693 (2019) 133556

Fig. 5. Changing rates (a) and DHA (b) of the IHA 33 indicators at six gauging stations in the Wuding River basin. distribution of the time series. The FDC approach has been widely ap- period reduced compared with those in the baseline period. The reduc- plied in the hydrological field due to its simple calculation procedure. tion rates of low-flow (Q90) were higher than high-flow (Q10)atsix To quantify temporal changes of the daily runoff in the magnitude and gauging stations. For instance, the indices of Q10, Q50 and Q90 reduced frequency, we selected three indices (i.e. exceeds 50% of the time dramatically by 36.9%, 31.7% and 51.2% at DJG station, respectively

(Q50), the indexes of RQ10:Q50 and RQ90:Q50) for analysis in different (Fig. 4e). In addition, the changes in the low-flow index (RQ90:Q50) periods. were greater than the high-flow index (RQ10:Q50). Among all the sta- Fig. 4 showed the FDCs of daily discharge in two periods at six sta- tions, we found the lowest reduction in both high-flow and low-flow in- tions. The results indicated that the daily runoff indices in the changing dices (−1.28% and −3.70%) at LJH station. Extreme high reduction was

Fig. 6. The changes of IC value in different periods in the Wuding River basin. X. Tian et al. / Science of the Total Environment 693 (2019) 133556 7

detected in RQ90:Q50 (approximately −30%) at BJC, HJM and DJG Table 3 stations. The changes of IC value in the Wuding River basin. Year BJC SD QYC LJH HJM DJG 4.3. Hydrologic alteration evaluation with IHA/RVA 1995 −0.83 0.83 0.78 0.73 −1.89 −1.29 2000 −1.28 0.38 0.35 0.42 −2.43 −1.74 We assessed the alteration of runoff regime using the IHA/RVA ap- 2005 −1.48 0.2 0.15 0.26 −2.63 −1.94 proach, which has been widely applied to assess the hydrological alter- 2015 −2.46 −0.92 −0.91 −0.87 −3.34 −2.87 − − − − − − ation. As can been seen in Fig. 5a, the monthly average runoff decreased Slope 0.0798 0.0859 0.0831 0.0797 0.0695 0.0773 at six stations, particularly during April–August with average changing rates lower than −40%. However, magnitude dramatically rose in Janu- ary (64.6%) and December (14.7%) at SD station, and increasing rates reservoirs in the whole basin. The varied RI denotes spatial distribution were 12.1% and 16.9% in January at QYC and LJH station, respectively. of dams/reservoir in different regions. For instance, the lowest RI is lo- Similar to the monthly average runoff, the annual extreme high and cated in the sub-basin of HJM, indicating relative sparse dams distrib- low magnitude (1-, 3-, 7-, 30-and 90-maximum and minimum) de- uted in this region. In this sub-basin, the land surface is dominated by clined, but there were exceptional rising in 1- and 3-minimum magni- sandy soil with flat topography, which is not suitable for check dam con- tude (137.5% and 50.0%, respectively) at SD station. The date of struction. The sub-basins of QYC and DJG belong to hilly-gully loess re- minimum increased by 186% at QYC station, while no changes were de- gion, and more check dams were built to control soil erosion. This may tected in date of minimum and maximum at other stations. The fre- be the main reason for high RI in these regions. quencies of high and low pulse declined at most stations, and the changing rates in duration of low pulse were higher than 50% at six sta- 4.4.3. Relationship between degree of hydrologic alteration and IC and RI tions, even reached up to 125% at QYC station. The simple linear regression method was used herein to explore the According to Fig. 5b, it is obvious to see that the DHAs of monthly av- correlation between DHA and IC. As shown in Table 5, the IC have ex- erage runoff in flood reason from June to September were lower than tremely strong negative correlation with DHA (p b 0.01) in the sub- −67% at five gauging stations (BJC, SD, QYC, LJH and DJG), which are basins of BJC, SD, LJH, and DJG, which implies a decreasing IC with in- moderate-degree alteration or low-degree alteration (|DHA|b33%). creasing DHA. As for the sub-basin of QYC and HJM, positive relationship High alteration can be examined in March–May and October– can be detected between DHA and IC, but not significant. This indicates November at BJC and DJG stations and in January at LJH station with that the changes in IC might not be the dominant reason leading to high DHA of 86.2%. The HJM station showed high alternation as with extreme DHA. low DHA of −100% as presented by extreme low/high magnitude indi- Table 6 shows the relationship between DHA and RI. A relative good ces, i.e. 1-, 3-, 7-, 30- and 90-minimum. correlation can be detected between DHA and RI in the sub-basins of To sum up, a majority of indices showed high hydrologic alteration. BJC, DJG, LJH and HJM. The correlation coefficientsarehigherthan The integral DHAs at all stations was examined and N69%. Among them, 0.58, with the highest values of 0.78 at LJH. Furthermore, we found HJM station showed the highest value of 81.71%, followed by DJG station that DHA were negatively correlated with RI due to their deceasing with 77.68%, and the lowest DHA was found at LJH station (69.79%). trends in most sub-basins (Table 4), but except the DJG.

4.4. Relationship between DHA and IC/RI 5. Discussion

4.4.1. Changes of IC value 5.1. Changes in streamflow regimes Fig. 6 showed the spatial distribution of hydrologic connectivity for the whole Wuding River basin in different year (1995, 2000, 2005 and Hydrological regime had been altered in the past decades on the 2015). The IC values were smaller (negative) in upstream regions cov- Loess Plateau, which has also been addressed in previous studies ered by coarse sandy, and higher (positive) in midstream and down- (Chen et al., 2010; Guo et al., 2019; Wang et al., 2013; Yang et al., stream covered with fine loess. This implied that the possibility of 2012). This study applied Man-Kendall test and Lee-Heghinian method runoff transport was higher in loess region in downstream. The IC to detect the changing trend and abrupt changes for annual runoff in the values were higher in the valley than in the hillslopes. The IC value Wuding River basin during 1960–2016. The findings are similar to Li has a relative broader range from −10.1 to 6.2 in 1995 than those in et al. (2007) and Zhou et al. (2012), indicating a significant downward 2000, 2005 and 2015. The basin-averaged IC also showed a decreasing trend in annual streamflow. The abrupt changing points occurred in trend from 1995 (−0.82) to 2015 (−2.44) due to significant changes 1971/1970 at most stations, except HJM station with the changing in land use. point in 1967. A series of soil and water conservation measures had It has been mentioned that land use changed dramatically due to the been implemented in the Loess Plateau since 1950s. Hydrological pro- implementation of soil and water conservation measures and cesses would be altered with increasing amount of the measures until converting farmland to forests for grass. However, these changes varied the beginning of 1970s (Mou et al., 2017; Yang et al., 2017). Therefore, in different sub-basins. Table 3 showed changes of IC in different sub- the changing points in 1970s were consistent with progress of soil and basins. Among them, we found that the average IC values were positive water conservation measures in the Wuding River basin. in the sub-basins controlled by SD, QYC and LJH stations, suggesting The FDC curves exhibited general view of daily time series (Vogel higher hydrological connectivity in these areas. Whereas the IC values and Fennessey, 1994; Brown et al., 2005). This study analyzed the were negative in the sub-basin of BJC, HJM and DJG. IC showed decreas- changes of the streamflow regime in two different periods. By contrast, ing trends from 1995 to 2015 in the whole basin. This suggested that the hydrological connectivity decreased within the study period due to landscape changes. Table 4 RI changes in the Wuding River basin (10−3).

4.4.2. Changes in index of check-dam/reservoir Year BJC SD QYC LJH HJM DJG Table 4 showed the temporal changes of RI in different periods. The 1995 0.17 0.41 4.71 1.92 0.11 3.82 result indicated that an upward trend could be examined in the sub- 2000 0.18 0.21 1.52 1.98 0.10 4.00 basins of BJC and DJG station, which are located in the downstream of 2005 0.19 0.22 1.50 1.88 0.09 4.25 2015 1.24 0.24 1.51 1.77 0.10 29.6 the basin. The upward RI denotes an increasing construction of dams/ 8 X. Tian et al. / Science of the Total Environment 693 (2019) 133556

Table 5 Correlation between degree of hydrologic alteration and Index of Connectivity.

Independent Dependent Sub-basin Correlation model R2 ⁎⁎ BJC y = −0.0139x + 0.7546 0.9366 ⁎⁎ SD y = −0.0044x + 0.6968 0.9566 QYC y = 0.0006x + 0.727 0.0205 Degree of hydrologic alteration Index of Connectivity ⁎⁎ LJH y = −0.0098x + 0.6967 0.9544 HJM y = 0.0007x + 0.822 0.0581 ⁎⁎ DJG y = −0.0147x + 0.7314 0.9385

⁎⁎ Correlation is significant at the 0.01 level.

the indices of median streamflow (Q50), high flow (RQ10:Q50) and low 123 million t sediment per year on average during 2007–2013 in the re- flow (RQ90:Q50) showed a reduction in all percentile streamflow, and gion upstream of Tongguan gauging station in the Loess Plateau. slopes in changing period were slightly declined than those in baseline Among these measures check dam/reservoir was considered to period. These changes are consistent with the results from Gao et al. be the dominant factor influencing the hydrological regime

(2015), which found that high flow (Q10) and median flow (Q50) de- (Bonacci et al., 2010; Hu et al., 2008). In this study, the liner regres- creased significantly during 1953–2010 in the Yanhe River basin, sion analysis indicated that the IC and RI had good relationship whereas the low flow index (Q90) increased. Soil conservation measures with DHA. This suggested that changes in land use/cover and dams/ played a dominant role in runoff indices reduction, and this has also reservoirs constructions would influenced hydrologic regime. This been confirmed by Li et al. (2007) and Zhou et al. (2012). is consistent with Zhao et al. (2012) and Liang et al. (2015),which The DHAs analysis suggested that the average monthly runoff varied found that the DHA increased gradually with the land use/cover greatly in March–May in six sub-basins (|DHA|N67%), particularly the changes, however, the conservation measures in the river networks annual extreme low magnitude (1-, 3-, 7-, 30- and 90-minimum) at (dams/reservoirs) were mainly responsible for decreasing runoff. HJM station with DHA of −100%. The integral DHA were higher than Similar results were also found by Yang et al. (2008).Amajorityof 70% at all the stations in the Wuding River watershed, suggesting strong researches attempted to investigate the influences of land use/ variation in the magnitude, duration, frequency, timing and rate of cover changes on hydrology regime (Guo et al., 2008; Xie and Cui, change of daily runoff. The variation of hydrologic indices was largely 2011), whereas more studies should be undertaken to address the caused by the uneven distribution of precipitation and frequent rain- influence of terraces, check-dams and others factors. storms. This is quite different from previous work like Mou et al. (2017), who found that runoff in January and February changed greatly in Wuding River basin during 1960–2007. 6. Conclusion

5.2. Underlying causes on hydrologic changes In this study, we applied the IHA/RVA approach to access the hydro- logic alteration of the daily streamflow within period from 1960 to 2016 In general, the changes in hydrological regime in the Wuding River in the Wuding River basin. The effects of land use changes and construc- basin can be attributed to the climate change and human activities (Li tion of dams/reservoirs were also investigated. The main conclusions et al., 2007; Zhou et al., 2012). The Loess Plateau has experienced a rel- can be drawn as following: atively warm and dry period according to the historical observations During 1960–2016, significant downward trend was examined in since 1950s (Zhao et al., 2013). This may lead to more water to be evap- annual streamflow, and abrupt changes occurred in 1971(1970) in the orated. He et al. (2017) addressed that the rainfall amount, intensity and Wuding River basin at most stations. The QYC station had no significant storm frequency did not change greatly over the past decades, suggest- changes, and an abrupt change was detected in 1967 at HJM station. FDC ing limited contribution to changes in hydrologic alternation from pre- analysis indicated that both high and low flow indices (RQ5:Q50 and RQ90: cipitation variation. Since 1950s, a series of soil and water conservation Q50) reduced significantly. The integral DHAs were higher than 70% at all measures have been implemented in the Wuding River Basin. These the stations in the Wuding River basin, suggesting that great changes measures include terraces, check-dams/reservoirs, fish-scale pits, affor- occurred in the magnitude, duration, frequency, timing and rate of estation and grass plantation. Furthermore, the Chinese government change of daily runoff. launched a large project “Grain for Green” in 1999, which made evident DHA had good correlation with both IC and RI, implying that hydro- changes in land use/cover from 2000 to 2015 (Fu et al., 2017). The arable logical regime was highly affected by land use changes and dams/reser- land reduced from 8736 km2 in 1995 to 5972 km2 in 2015, and the area voirs construction in the study area. The IC has extremely strong of grass land increased from 14,183km2 in 1995 to 19,996km2 in 2015. A negative correlation with DHA (p b 0.01) in the sub-basins of BJC, SD, recent survey suggested that the number of check-dam increased to LJH, and DJG, and positive correlation in the sub-basins of QYC and 11,602, including 1155 key dams, 3747 medium-sized dams and 6700 HJM, but not significant. The correlation coefficients between DHA and small-scale dams in the Wuding River basin (Han et al., 2018)and RI are higher than 0.58 in the sub-basins of BJC, DJG, LJH and HJM, reporting that check dams have trapped a total of 41.3 million t sedi- with the highest values of 0.78 at LJH. By contrast, the hydrological alter- ment during 2011–2017 in the Wuding River basin. A study from Gao nation was more sensitive to land use changes rather than dams et al. (2014) found that check dams could trapped approximately construction.

Table 6 Correlation between degree of hydrologic alteration and Reservoirs/Dams index.

Independent Dependent Sub-basin Correlation model R2

BJC y = 15.337x + 0.7688 0.6669 SD y = −11.88x + 0.6995 0.1133 QYC y = −0.8047x + 0.729 0.1957 Degree of hydrologic alteration Reservoirs/Dams index LJH y = −69.538x + 0.8268 0.7827 HJM y = −148.04x + 0.8352 0.5859 DJG y = 0.6351x + 0.7537 0.6495 X. Tian et al. / Science of the Total Environment 693 (2019) 133556 9

Acknowledgement sediment concentrated region in the Loess Plateau, China. Water Resour. Manag. 29 (11), 4045–4060. Lin, K., Lin, Y., Liu, P., He, Y., Tu, X., 2016. Considering the order and symmetry to improve The work is financially supported by the National Key ScientificRe- the traditional RVA for evaluation of hydrologic alteration of river systems. Water search Project (2016YFC0402401), the National Natural Science Foun- Resour. Manag. 30 (14), 5501–5516. Liu, C.M., Zheng, H.X., 2004. Changes in components of the hydrological cycle in the Yel- dation of China (Grant Nos.: 41671279, 51509206), the West Light low River basin during the second half of the 20th century. Hydrol. Process. 18 (12), foundation of the Chinese Academy of Sciences (XAB2017A03) and In- 2337–2345. ternational Scientific and Technological Cooperation and Exchange Pro- Lopez, J., Frances, F., 2013. Non-stationary flood frequency analysis in continental Spanish jects in Shaanxi Province (2017KW-042). Special thanks are also given rivers, using climate and reservoir indices as external covariates. Hydrol. Earth Syst. Sci. 17 (8), 3189–3203. to the Yellow River Conservation Committee for providing the hydro- Mann, H.B., 1945. Non-parametric test against trend. Econometrica 13 (3), 245–259. logical data. The author would also thank for the very valuable com- Mou, X., Liu, Q., Ping, F., Liang, J., 2017. Analysis of changes in annual streamflow and its fl ments from reviewers which greatly improved the quality of the paper. in uence factors in Wuding River basin in the last 50 years (1960-2007). J. Beijing Normal Univ. (Nat. Sci.) 53 (4), 465–471. Mu, X.M., Zhang, L., McVicar, T.R., Chille, B., Gau, P., 2007. Analysis of the impact of conser- References vation measures on stream flow regime in catchments of the Loess Plateau, China. Hydrol. Process. 21 (16), 2124–2134. Bonacci, O., Tadic, Z., Trninic, D., 2010. Effects of dams and reservoirs on the hydrological Pfeiffer, M., Ionita, M., 2017. Assessment of hydrologic alterations in Elbe and Rhine Riv- characteristics of the lower Drava River. River Res. Appl. 7 (4), 349–357. ers, Germany. Water 9 (9), 1–18. Borselli, L., Cassi, P., Torri, D., 2008. Prolegomena to sediment and flow connectivity in the Piao, S., Ciais, P., Huang, Y., Shen, Z., Peng, S., Li, J., Zhou, L., Liu, H., Ma, Y., Ding, Y., landscape: a GIS and field numerical assessment. Catena 75 (3), 268–277. Friedlingstein, P., Liu, C., Tan, K., Yu, Y., Zhang, T., Fang, J., 2010. The impacts of climate Brown, A., Zhang, L., McMahon, T., Weatern, A., Vertessy, R., 2005. A critical review of change on water resources and agriculture in China. Nature 467 (7311), 43–51. paired catchment studies with reference to the seasonal flows and climate variability. Richter, B.D., Baumgartner, J.V., Powell, J., Braun, D.P., 1996. A method for assessing hydro- J. Hydrol. 310, 28–61. logic alteration within ecosystems. Conserv. Biol. 10, 1163–1174. Chen, Y.Q.D., Yang, T., Xu, C.Y., Zhang, Q., Chen, X., Hao, Z.C., 2010. Hydrologic alteration Van Loon, A.F., Laaha, G., 2015. Hydrological drought severity explained by climate and along the Middle and Upper East River (Dongjiang) basin, South China: a visually en- catchment characteristics. J. Hydrol. 526, 3–14. hanced mining on the results of RVA method. Stoch. Env. Res. Risk A. 24 (1), 9–18. Vogel, R., Fennessey, N., 1994. Flow-duration curves. I: new interpretation and confidence Forootan, E., Khaki, M., Schumacher, M., Wulfmeyer, V., Mehrnegar, N., Van Dijk, A.I.J.M, intervals. J. Water Res. 120, 485–504. 2019. Understanding the global hydrological droughts of 2003–2016 and their rela- Wanders, N., Wada, Y., 2015. Human and climate impacts on the 21st century hydrolog- tionships with teleconnections. Sci. Total Environ. 650, 2587–2604. ical drought. J. Hydrol. 526, 208–220. Fu, B.J., Wang, S., Liu, Y., Liu, J.B., Liang, W., Miao, C.Y., 2017. Hydrogeomorphic ecosystem Wang, Y., Ding, Y.J., Ye, B.S., Liu, F.J., Wang, J., Wang, J., 2013. Contributions of climate and responses to natural and anthropogenic changes in the Loess Plateau of China. Annu. human activities to changes in runoff of the Yellow and Yangtze rivers from 1950 to Rev. Earth Planet. Sci. 45 (1), 223–243. 2008. Sci. China Earth Sci. 56 (8), 1398–1412. Gao, G., Huang, C.Y., 2001. Climate change and its impact on water resources in North Xie, X.H., Cui, Y.L., 2011. Development and test of SWAT for modeling hydrological pro- China. Adv. Atmos. Sci. 18 (5), 718–732. cesses in irrigation districts with paddy rice. J. Hydrol. 396 (1–2), 61–71. Gao, Y., Vogel, R.M., Kroll, C.N., Poff, N.L., Olden, J.D., 2009. Development of representative Xu, J.X., 2003. Sediment flux to the sea as influenced by changing human activities and indicators of hydrologic alteration. J. Hydrol. 374 (1–2), 136–147. precipitation: example of the Yellow River, China. Environ. Manag. 31 (3), 328–341. Gao, P., Geissen, V., Ritsema, C.J., Mu, X.M., Wang, F., 2013. Impact of climate change and Xu, J.X., 2009. Influence of soil and water conservation on sediment budget of channel anthropogenic activities on stream flow and sediment discharge in the Wei River system: a case study of the Wudinghe River basin. Sci. Soil Water Conserv. 7 (4), basin, China. Hydrol. Earth Syst. Sci. 17 (3), 961–972. 7–13. Gao, Y.F., Guo, Y.T., Liu, X.Y., He, X.Z., 2014. Amount of the intercepted sediment by the Yang, T., Zhang, Q., Chen, Y.D., Tao, X., Xu, C.Y., Chen, X., 2008. A spatial assessment of hy- warping dams in the regions above Tongguan in the Yellow River Basin. Yellow drologic alteration caused by dam construction in the middle and lower Yellow River, River 36 (7), 97–99. China. Hydrol. Process. 22 (18), 3829–3843. Gao, P., Jiang, G., Wei, Y., Mu, X., Wang, F., Zhao, G., Sun, W., 2015. Streamflow regimes of Yang, Z.F., Yan, Y., Liu, Q., 2012. Assessment of the flow regime alteration in the lower Yel- the Yanhe River under climate and land use change, Loess Plateau, China. Hydrol. Pro- low River, China. Eco. Inform. 10, 56–64. cess. 29 (10), 2402–2413. Yang, Y.Y., Li, Z.B., Ren, Z.P., Gao, H.D., 2017. The influence of human activities on the Guo, H., Hu, Q., Jiang, T., 2008. Annual and seasonal streamflow responses to climate and changes of water and sediment in different landforms. J. Sediment. Res. 42 (05), land-cover changes in the Poyang Lake basin, China. J. Hydrol. 355 (1–4), 106–122. 50–56. Guo, W.X., Li, Y., Zhuo, Z.Y., Zhang, L., 2019. Impact Assessment of Three Gorges Reservoir Zhang, X.P., Zhang, L., Zhao, J., Rustomji, P., Hairsine, P., 2008. Responses of streamflow to on hydrological regimes in the middle and lower reaches of Yangtze River. Water changes in climate and land use/cover in the Loess Plateau, China. Water Resour. Res. Power 45 (05), 22–27. 44 (7), 1–12. Han, X.N., Xie, S.Y., Gao, Y.F., 2018. Study on sediment retaining effect of check dams in Zhang, Q., Xu, C.Y., Chen, Y.D., Yang, T., 2009. Spatial assessment of hydrologic alteration Wuding River basin in recent years. Yellow River 40 (11), 5–8(37). across the Delta, China, and possible underlying causes. Hydrol. Process. He, Y., Tian, P., Mu, X.M., Gao, P., Zhao, G.J., Wang, F., Li, P.F., 2017. Changes in daily and 23 (11), 1565–1574. monthly rainfall in the middle Yellow River, China. Theor. Appl. Climatol. 129 Zhang, X., Cai, H.J., Yin, X.N., 2010. Variation trends analysis of runoff and sediment time (1–2), 139–148. series based on r/s method in Wuding River basin. Trans. Chin. Soc. Agric. Eng. 26 Hu, W.W., Wang, G.X., Deng, W., Li, S.N., 2008. The influence of dams on ecohydrological (14), 212–217. conditions in the Huaihe River basin, China. Ecol. Eng. 33 (3–4), 233–241. Zhao, Q.H., Liu, S.L., Deng, L., Dong, S.K., Wang, C., Yang, Z.F., Yang, J.J., 2012. Landscape Huang, M., Zhang, L., 2004. Hydrological responses to conservation practices in a catch- change and hydrologic alteration associated with dam construction. Int. J. Appl. ment of the Loess Plateau, China. Hydrol. Process. 18 (10), 1885–1898. Earth Obs. Geoinf. 16, 17–26. Kendall, M.G., 1955. Rank correlation methods. Br. J. Psychol. 25 (1), 86–91. Zhao, G.J., Mu, X.M., Wen, Z.M., Wang, F., Gao, P., 2013. Soil erosion, conservation, and eco- Lee, A.F.S., Heghinian, S.M., 1977. A shift of the mean level in a sequence of independent environment changes in the Loess Plateau of China. Land Degrad. Dev. 24 (5), normal random variables: a Bayesian approach. Technometrics 19 (4), 36–41. 499–510. Li, L., Zhang, L., Wang, H., Wang, J., Yang, J., Jiang, D., Li, J., Qin, D., 2007. Assessing the im- Zhao, G.J., Tian, P., Mu, X.M., Jiao, J.Y., Wang, F., Gao, P., 2014. Quantifying the impact of pact of climate variability and human activities on streamflow from the Wuding River climate variability and human activities on streamflow in the middle reaches of the basin in China. Hydrol. Process. 21 (25), 3485–3491. Yellow River basin, China. J. Hydrol. 519, 387–398. Li, Z., Liu, W.Z., Zhang, X.-C., Zheng, F.L., 2009. Impacts of land use change and climate var- Zhao, G.J., Li, E.H., Mu, X.M., Wen, Z.M., Rayburg, S., Tian, P., 2015. Changing trends and re- iability on hydrology in an agricultural catchment on the Loess Plateau of China. gime shift of streamflow in the Yellow River basin. Stoch. Env. Res. Risk A. 29 (5), J. Hydrol. 377 (1–2), 35–42. 1331–1343. Li, P.F., Mu, X.M., Holden, J., Wu, Y.P., Irvine, B., Wang, F., Gao, P., Zhao, G.J., Sun, W.Y., 2017. Zhou, Y.Y., Shi, C.X., Du, J., Fan, X.L., 2012. Characteristics and causes of changes in annual Comparison of soil erosion models used to study the Chinese Loess Plateau. Earth Sci. runoff of the Wuding River in 1956–2009. Environ. Earth Sci. 69 (1), 225–234. Rev. 170, 17–30. Liang, W., Bai, D., Jin, Z., You, Y., Li, J., Yang, Y., 2015. A study on the streamflow change and its relationship with climate change and ecological restoration measures in a