Detecting the effects of climate changes and human activities on river regimes would help to identify the original driving forces of hydrological disturbances and highlight the mechanisms of such hydrological processes under changing conditions. Two non-parametric tests (Mann-Kendall and Pettitt) were applied to detect the change point of runoff (1964–2006) in the upper and middle reaches of the Heihe River basin. The change point was determined to be 1980, which divided the period into two parts (the baseline period and the change period). Double mass curve and hydrological sensitivity-based methods were used to separate the impacts of climate changes and human activities on runoff variation. The results demonstrated that human activities were the dominant force affecting runoff variation in the upper and middle reaches of the basin. At the sub-basin level, climate changes played a more significant role in the upstream region, while human activities dominated in the midstream region. Therefore, different countermeasures should be taken in the upstream and midstream regions to ensure sustainable water resource development in the Heihe River basin.

climate changes, Heihe River, human activities, runoff variation

Climate changes and human activities are the two major driving forces, which alter and complicate the water cycle (Ramanathan et al. 2001; Vogel 2011). Climate changes, induced by global warming, were predicted to enhance the variability of available water resources (Milly et al. 2005; Huntington 2006; Keller 2009). Generally, climate changes influence hydrological processes by altering the typical precipitation distribution and temperature fluctuation (Willett et al. 2007). These climate variations may cause adverse impacts on ecosystems, agriculture, water resources and other fields (Lin et al. 2007). Simultaneously, human activities play a key role in disturbing hydrological processes. Such activities as land use/cover change and hydraulic projects (e.g., dams and irrigation engineering) will directly change the underlying surface and reallocate streamflow within basins (Zhang & Schilling 2006; Chen et al. 2010). Anthropogenic activities mainly affect hydrological processes through exerting effects on water supply and demand. These alterations have significant ecological, environmental, social and even economic impacts (DeFries & Eshleman 2004). When considering the influences of climate changes and human activities simultaneously, hydrologic elements may significantly vary, especially runoff (Legesse et al. 2003; Wang et al. 2012; Shi et al. 2013).

Considering the coupled effects of climate changes and human activities on hydrological processes, it is important to detect streamflow variations and identify the mechanisms behind the two driving forces. Detection and identification at the watershed level will benefit water resources management. Some efforts have been made to reveal the impact of climate changes and human activities (e.g., Changnon & Demissie 1996; Li et al. 2007; Wang & Hejazi 2011; Tang et al. 2013; Peng et al. 2014). Previous studies applied a number of statistical analyses, such as the three-step method (Changnon & Demissie 1996) and linear regression method (Beguería et al. 2003; Jiang et al. 2011), and hydrological models, such as the PRMS model (Legesse et al. 2003), SWAT model (Li et al. 2009) and VIC-3L model (Jiang et al. 2011).

In this study, both statistic-based (the double mass curve method) and modelling (a simple water balance hydrological model) methods were used to evaluate the effects of climate changes and human activities on hydrological regimes. These two methods have been widely used to estimate the impacts of the different driving forces. For example, Li et al. (2007), Ma et al. (2008), Liu et al. (2009), Zhang et al. (2011) and Zhang et al. (2014) employed hydrological models to assess the effects of climate variability and human activities on streamflow in several basins. The double mass curve method has been applied in the Xijiang River basin (Zhang & Lu 2009), Zhengshui River basin (Du et al. 2011) and Yellow River basin (Gao et al. 2011).

The impacts of climate changes and human activities on streamflow are coupled. It is hard to separate the individual processes, because the response mechanisms between climate changes, human activities and hydrological processes are complex. Here, we assume that the impacts of climate changes and human activities are independent, to demonstrate their individual effects.

Hydrological and environmental systems respond more sensitively to climate changes and human activities in the arid and semi-arid regions (Ye et al. 2013; Xu et al. 2014). Thus, a typical (arid or semi-arid) basin, the Heihe River basin, was selected as a case study. As a grain production base in China, and also the socio-economic and political centre of Gansu Province, the middle reach of the basin suffers from intensive human activities. These activities have led to severe water resource crises, as well as environmental and ecological issues, especially in recent decades. A quantitative assessment of the effects of climate changes and human activities on runoff regimes would provide a theoretical basis for mitigation and adaptation of water resources systems in the Heihe River basin. Some related studies have been conducted in the basin. However, a theoretical and systematic estimation, based on statistical analysis and hydrological modelling, is still lacking. Such work is urgent for local water resources management. Therefore, the objectives of this study are to (1) detect the trends and change points of hydrological regimes and (2) identify the effects of the two driving forces on runoff regimes at the basin and sub-basin levels to explore the mechanisms affecting hydrologic regimes.

Study area

The Heihe River basin, the second largest inland river located in northwest China, is divided into three reaches, the upper, middle and lower reaches, by two hydrological stations (Yingluoxia and Zhengyixia). It originates at the northern foot of the Qilian Mountain, wanders through Qinghai, Gansu Province and the Inner Mongolia Autonomous Region, and finally flows into the east and west Juyan Lake. The basin encompasses mountain land, oasis and desert. The Heihe River basin, with a main stream length of approximately 821 km, has an area of nearly 143,000 km2.

In this study, the upper and middle reaches of the Heihe River basin (UMHB) were selected as the study area. Figure 1 shows the study area and all the hydro-meteorological stations. Analyses were also conducted at the sub-basin level in the upstream (UHB) and midstream (MHB) regions. In the UMHB, the mean annual precipitation, potential evapotranspiration and runoff are 226 mm, 924 mm and 88 mm, respectively. In addition, over 90% of the population, grain production and major industries are concentrated in the middle reaches. Approximately 84% of the total available water was consumed for irrigation, with demand consistently increasing (Ge et al. 2013). Land use/cover has also significantly changed in the basin. Urban land, cropland and woodland have increased, while water regions and grasslands have become increasingly degraded (Kun et al. 2007; Nian et al. 2014). Human activities, such as irrigation, dam construction and other types of water diversion engineering, are also intensive in this region. In this study, four hydrological stations (Yingluoxia, Binggou, Yuanyangchishuiku and Zhengyixia stations) were selected, among which, Yingluoxia and Binggou stations locate in the UHB and the others in the MHB. Eleven meteorological stations in or near the study area were selected (see Figure 1).
Figure 1
Figure 1. The upper and middle reaches of Heihe River basin.
View largeDownload slide

The upper and middle reaches of Heihe River basin.

Figure 1
Figure 1. The upper and middle reaches of Heihe River basin.
View largeDownload slide

The upper and middle reaches of Heihe River basin.

Close modal

Data description

Hydro-meteorological data in this study were provided by the Cold and Arid Regions Science Data Centre at Lanzhou (http://westdc.westgis.ac.cn/), and the Hydrology and Water Resources Survey Bureau of Gansu Province. Daily climate data (e.g., precipitation, temperature, humidity, radiation, wind speed and day length) were collected from 11 meteorological stations from 1964–2006 (Figure 1). Actual evapotranspiration data were unavailable in the study area. Instead, potential evapotranspiration, which was estimated using the Penman-Monteith method (Allen et al. 2006) by these climate data, was used. Annual precipitation and potential evapotranspiration of UMHB, UHB and MHB were used by interpolating the data of relevant meteorological stations with the Inverse Distance Weighted (IDW) method. Annual runoff values from 1964 to 2006 for UMHB, UHB and MHB were aggregated from the runoff measurements of the four hydrological stations (Yingluoxia, Binggou, Yuanyangchishuiku and Zhengyixia stations) using a weighted average method. Here, runoff depth was used and its unit is mm, as well as units of precipitation and potential evapotranspiration, and the time interval is one year.

In order to estimate the impacts of climate change and human activities on runoff variation, a change point is needed to divide the time series into a baseline period and a change period. In this study, change points of runoff in four sub-basins (i.e., Yingluoxia, Binggou, Zhengyixia and Yuanyangchishuiku sub-basins, which were called according to the outlets of the sub-basins) were detected. Runoff at the outlet represents that of the relevant sub-basin, for example, runoff at Yingluoxia hydrological station was used as runoff of Yingluoxia sub-basin. Then annual precipitation and potential evapotranspiration of the four sub-basins were also calculated by interpolating the data of relevant meteorological stations by the IDW method.

Change point analysis

Two nonparametric test methods, the Mann-Kendall (MK) test and the Pettitt test, were separately employed to detect the change points. Then, the cumulative departure curve was plotted to verify the results.

Mann-Kendall test

The MK method (Mann 1945; Kendall 1975) has been widely used to detect the changing trends and change points of a time series (Huo et al. 2008; Peng & Xu 2010; Peng et al. 2014). For an independent time series (x1, x2, …, xn), Sk is established by:
formula
1
where k = 2, …, n, and, in the equation, of two different observations (i, j and j= 1, 2, …, i), when xi > xj, then ri = 1; when xi ≤ xj, then ri = 0. The statistical variable ⁠, with a standard normal distribution, is defined as:
formula
2
where ⁠, and and are the variance and mean of Sk. When the time series are in the same continuous distribution, then:
formula
3
formula
4

If with the given significance level ⁠, it indicates that the statistical series exhibits a clear trend. Then, the process was repeated using the inverse time series (xn, …, x2, x1) with (k = n, n − 1, …, 1) and ⁠. Finally, the interaction point of the two graphs (UFk, UBk) is considered to be the change point if it is within the critical limits (using a 95% confidence level, UBk = 1.96 and UFk = −1.96, which were selected as the critical limits in the study) (Mann 1945; Kendall 1975).

Pettitt test

The Pettitt test (Pettitt 1979) detects a change in a sample by arbitrarily splitting it into two samples. The Pettitt statistic, Ut,T, is defined as:
formula
5
when xi > xj, then sgn (xi – xj) = 1; when xi = xj, then sgn (xi – xj) = 0; when xi < xj, then sgn (xi – xj) = −1. It sets the null hypothesis, H0, so that the two samples, x1, …, xt and xt+1, …, xT, exhibit no change. The following statistics apply:
formula
6
formula
7
The final Pettitt statistic, M, describes the change point with the date, T. The significance probability of rejecting H0 was:
formula
8

If D < 0.05, then the change point is significant.

The cumulative departure curve is a variable curve of cumulative departures. Departures were measured as follows:
formula
9
where x is the time series. The extreme values of the cumulative departure curves are typically the change points.

Double mass curve method

The double mass curve (DMC) method (Searcy & Hardison 1960) was used to verify the consistency of the hydrological data. In this study, the three variables (runoff, precipitation and potential evapotranspiration) were used in the double mass curve method. The linear relationship between cumulative runoff, cumulative precipitation and cumulative potential evapotranspiration in the baseline period was established as:
formula
10
where Q refers to runoff, P refers to precipitation and E0 refers to potential evapotranspiration. k1, k2, b are the parameters. The cumulative runoff in the change period was calculated using the linear relationship defined in the baseline period. Two differences were calculated as:
formula
11
formula
12
where Q2m and Q1m are the measured mean annual runoff in the change period and baseline period, respectively, and Q2s is the simulated mean annual runoff in the change period. indicates the runoff variation induced by climate changes and human activities, which can be described as follows:
formula
13
and describe the variations primarily induced by human activities and climate changes, respectively. Finally, the ratios of chuman and cclim are calculated as:
formula
14
formula
15

The value of chuman characterises the contribution from human activities, and the value of cclim represents the proportion of the contribution from climate changes.

Hydrological sensitivity-based method

The hydrological sensitivity-based (HSB) method (Li et al. 2007) was used to estimate the impact of climate changes (e.g., by precipitation P and potential evapotranspiration E0) on the runoff using the water balance (Dooge et al. 1999). The water balance in a basin is described as:
formula
16
where E is evapotranspiration, and ΔS is the variation of water storage in the basin, which is considered to be zero over a long period of time, such as 5–10 years. The long-term annual evapotranspiration could be calculated as (Zhang et al. 2001):
formula
17
where w is a parameter relating to vegetation types. The relationship between runoff, precipitation and potential evapotranspiration changes is defined as (Milly & Dunne 2002):
formula
18
where and are the variations of precipitation and potential evapotranspiration, respectively. β is the sensitivity coefficient of runoff to precipitation, and γ is the sensitivity coefficient to potential evapotranspiration. The two parameters (β and γ) can be calculated by:
formula
19
formula
20
where z is a dryness index (E0/P).

Determination of change point and hydrological regimes

As listed in Table 1, the change points of runoff and precipitation at the four sub-basins were tested using a cumulative departure curve, MK test and Pettitt test. The three test methods demonstrated a consistent runoff change point of 1980 and precipitation change point of 1978 in Yingluoxia sub-basin. As demonstrated in Table 1, the runoff change points in Yingluoxia, Zhengyixia, Binggou and Yuanyangchishuiku sub-basins likely occurred in 1980, 1989, 1983 and 1985, respectively, while the precipitation change points in the four sub-basins were 1978, 1976, 1980 and 1983, respectively. Following the development of the local economy, the intensity of human activities increased, and hydrological processes were increasingly impacted. Thus, the earliest runoff change point of the four sub-basins, in 1980, was selected as the dividing point of the baseline period and the change period to estimate the impacts of driving forces to runoff variation in MHB, UHB and UMHB.

Table 1

Change points of the four sub-basins via the cumulative departure curve, MK test and Pettitt test

Sub-basinRunoff
Precipitation
Cumulative departure curveMK testPettitt testCumulative departure curveMK testPettitt test
Yingluoxia 1980 1980 1980 1978 1978 1978 
Zhengyixia 1989 – 1989 1976 – 1976 
Binggou 1983 – 1983 1980 – 1980 
Yuanyangchishuiku – – 1985 – – 1983 
Sub-basinRunoff
Precipitation
Cumulative departure curveMK testPettitt testCumulative departure curveMK testPettitt test
Yingluoxia 1980 1980 1980 1978 1978 1978 
Zhengyixia 1989 – 1989 1976 – 1976 
Binggou 1983 – 1983 1980 – 1980 
Yuanyangchishuiku – – 1985 – – 1983 

Note: ‘–‘means change point not statistically significant at a 95% confidence level.

View Large

Vegetative cover and land use have changed significantly since the 1980s (Ma & Frank 2006; Qi & Luo 2006; Wang et al. 2007), and a more complicated landscape has evolved (Lu et al. 2003). In addition, many projects in the basin were constructed after 1980, e.g., the Caotanzhuang and Dadunmen projects in 1988 and 1990, and the water quantity allocation scheme of the Heihe River basin, which was issued by the Central Government in 1997. Thus, the selection of 1980 as the division of the baseline period and change period is justified.

Figure 2 presents the change curves of precipitation, potential evapotranspiration and runoff in the different study regions. The dotted lines in Figure 2 are moving average plots in the baseline and change period, respectively. Precipitation increased in all basins, but the difference between the two periods in the MHB was fairly small (see Figure 2(a)). Potential evapotranspiration in the MHB and UMHB was much larger than in the UHB. In addition, the moving-average of potential evapotranspiration increased in the MHB and UMHB, but declined in the UHB during the two periods (see Figure 2(b)). Runoff in the UHB significantly increased in the change period, when compared to the baseline period. Runoff decreased in the MHB, but the runoff difference in the UMHB was very small (see Figure 2(c)). Figure 3 shows the comparison of hydrological elements between the baseline period and the change period in the UMHB. As seen in Figure 3, runoff and potential evapotranspiration slightly declined in the change period (−2.3% and −1.2%, respectively) when compared to the baseline period, whereas precipitation and actual evapotranspiration exhibited a slight increase (2.9% and 6.2%, respectively). The dryness ratio (E0/P) and the evapotranspiration ratio (ET/P) in these two periods changed inversely, with variation ratios of −3.8% and 3.3%, respectively.
Figure 2
Figure 2. The change curves of (a) precipitation, (b) potential evapotranspiration and (c) runoff in the UMHB, UHB and MHB.
View largeDownload slide

The change curves of (a) precipitation, (b) potential evapotranspiration and (c) runoff in the UMHB, UHB and MHB.

Figure 2
Figure 2. The change curves of (a) precipitation, (b) potential evapotranspiration and (c) runoff in the UMHB, UHB and MHB.
View largeDownload slide

The change curves of (a) precipitation, (b) potential evapotranspiration and (c) runoff in the UMHB, UHB and MHB.

Close modal
Figure 3
Figure 3. Comparison of hydrological elements between the baseline period and the change period in the UMHB.
View largeDownload slide

Comparison of hydrological elements between the baseline period and the change period in the UMHB.

Figure 3
Figure 3. Comparison of hydrological elements between the baseline period and the change period in the UMHB.
View largeDownload slide

Comparison of hydrological elements between the baseline period and the change period in the UMHB.

Close modal

Adaptability of the two methods and attribution in UMHB

The linear relationship between the cumulative runoff, cumulative precipitation and cumulative potential evapotranspiration was established in the baseline period as ⁠, with a correlation coefficient of 0.9998. Runoff was calculated in the change period based on the regression equation. Figure 4 shows the accumulative curve of runoff during the baseline period and change period. During the change period, the simulated cumulative runoff fluctuated and had quite a difference with the measured runoff (Figure 4), which indicated the effects of human interference. Mean annual runoff in the change period decreased by 2.0 mm, of which 1.1 mm was induced by human activities and 0.9 mm by climate changes. Therefore, according to Equations (14) and (15), the contributions of human activities and climate changes to runoff variation were estimated and the results listed in Table 2 were 53% and 47%, respectively, using the DMC method.
Table 2

Quantitative assessment of impacts of climate changes and human activities on runoff using DMC method

 Qm/mmQs/mm△Qtot/mm△Qhuman/mm△Qclim/mmchuman/%cclim/%
The baseline period 89.6 89.9 −2.0 −1.1 −0.9 53 47 
The change period 87.6 89.3           
 Qm/mmQs/mm△Qtot/mm△Qhuman/mm△Qclim/mmchuman/%cclim/%
The baseline period 89.6 89.9 −2.0 −1.1 −0.9 53 47 
The change period 87.6 89.3           
View Large
Figure 4
Figure 4. Accumulative curve of runoff during the baseline period and change period.
View largeDownload slide

Accumulative curve of runoff during the baseline period and change period.

Figure 4
Figure 4. Accumulative curve of runoff during the baseline period and change period.
View largeDownload slide

Accumulative curve of runoff during the baseline period and change period.

Close modal
For the HSB method, w was determined to be 1.2 (Zhang et al. 2001), and β and γ were 0.11 and −0.02. The results suggest that the runoff variations induced by climate changes and human activities were 0.9 mm and 1.1 mm, and Figure 5 shows the quantitative assessment of impacts of the two driving forces which correlate with contributions of 45% and 55%, respectively.
Figure 5
Figure 5. Quantitative assessment of impacts of climate changes and human activities on runoff using HSB method.
View largeDownload slide

Quantitative assessment of impacts of climate changes and human activities on runoff using HSB method.

Figure 5
Figure 5. Quantitative assessment of impacts of climate changes and human activities on runoff using HSB method.
View largeDownload slide

Quantitative assessment of impacts of climate changes and human activities on runoff using HSB method.

Close modal

The two methods assess the impacts of climate changes and human activities with different theories. The DMC method mainly considers the impacts of human activities, while the HSB method considers climate change impacts. Therefore, if the assessments of the two methods are similar, then the reliability of the results is higher. The difference between the two methods was 2%, which indicates that the assessments are consistent and verify each other. Setting the results of the two methods as a range, the contribution of the impacts of climate changes and human activities on runoff variation in the UMHB were 45–47% and 53–55%, respectively. These results illustrate that the impact of human activities was only slightly dominant in the runoff regimes of the UMHB. Climate changes also account for a large proportion of runoff variation.

Furthermore, these contributions were analysed for different change points. Figure 6 presents the contribution of human activities with different change points. Using different change points, from 1980 to 1995, to divide the baseline period and the change period would result in significant variation of the estimation of the impacts of climate changes and human activities on runoff variation (see Figure 6). The parameter w remained constant for various change points, when using the HSB method. The performance coefficients, R2, of the DMC method were all larger than 0.9995 for various change points. No significant trends were found for the different change points using either of the two methods. The contribution difference of human activities before 1991 was small, and sometimes zero, as calculated by the two methods. The difference was larger after 1991, which may be attributed to the land use/cover change. These changes would have caused w to vary in the HSB method (Ma & Frank 2006; Wang et al. 2007).
Figure 6
Figure 6. Variation of chuman with different change points.
View largeDownload slide

Variation of chuman with different change points.

Figure 6
Figure 6. Variation of chuman with different change points.
View largeDownload slide

Variation of chuman with different change points.

Close modal
Figures 7 and 8 show the variation of the impacts of climate changes and the two sensitivity coefficients of runoff to precipitation and potential evapotranspiration (i.e., β and γ, respectively) following changing w and E0/P. In the HSB method, w was sensitive, and a small change led to a significant change in cclim, especially when w was smaller than 1.0, as seen in Figure 7(a). In the basin, cclim typically decreased as w increased. As w was a parameter related to vegetation, variations in w indicates changes linked to vegetative cover (Zhang et al. 2001). Thus, the increase in w indicated an increase in the vegetative cover. It can be concluded that the impact of climate changes on runoff regimes would decrease as the amount of vegetative cover in the basin increased. In addition, as the dryness index, E0/P, increased, cclim declined (see Figure 7(b)). This indicates that when the basin was drier, climate changes had less of an effect on runoff variation in the basin. The sensitivity coefficient of runoff to precipitation β decreased as w or E0/P increased while the opposite trend was illustrated for the sensitivity coefficient of runoff to potential evapotranspiration γ, as seen in Figure 8(a) and 8(b). Therefore, Figure 8 shows that, with more vegetative cover or a drier climate, runoff variation would be less sensitive to changes in precipitation and evapotranspiration.
Figure 7
Figure 7. Variation of cclim following changes to (a) w and (b) E0/P.
View largeDownload slide

Variation of cclim following changes to (a) w and (b) E0/P.

Figure 7
Figure 7. Variation of cclim following changes to (a) w and (b) E0/P.
View largeDownload slide

Variation of cclim following changes to (a) w and (b) E0/P.

Close modal
Figure 8
Figure 8. Variation of the sensitivity coefficients of runoff to precipitation β and potential evapotranspiraiton γ following changes to (a) w and (b) E0/P.
View largeDownload slide

Variation of the sensitivity coefficients of runoff to precipitation β and potential evapotranspiraiton γ following changes to (a) w and (b) E0/P.

Figure 8
Figure 8. Variation of the sensitivity coefficients of runoff to precipitation β and potential evapotranspiraiton γ following changes to (a) w and (b) E0/P.
View largeDownload slide

Variation of the sensitivity coefficients of runoff to precipitation β and potential evapotranspiraiton γ following changes to (a) w and (b) E0/P.

Close modal

Impacts of the two driving forces on runoff regimes in the UHB and MHB

The DMC method was applied to assess the impacts of climate change and human activities in the upstream and midstream regions of the basin, respectively. The relationship equation during the baseline period was ∑Q = 0.58*∑P − 0.05*∑E0 + 12.71 in the UHB and ∑Q = 0.13*∑P + 0.02*∑E0 + 26.61 in the MHB, respectively. The R2 performance indices were 0.9997 and 0.9991 in the two sub-basins. Figure 9 shows impact estimations using the DMC method in the UHB and MHB. In the UHB, climate changes and human activities contributed to 71% and 29% of the runoff variation, respectively (see Figure 9(a)). In addition 17% and 83% of the runoff variation in the MHB can be attributed to climate change and human activities, respectively (see Figure 9(b)). It can be concluded that climate change was the dominant driving force affecting hydrological variation in the UHB, while human activities dominated in the MHB. He et al. (2012) reported that climate change contributed to 60% of the upstream runoff variation, while human activities contributed to 75% of the midstream variation, based on a simple statistical method. Based on our evaluations using the two methods in this study, and verification using a previous study, we believe that our estimations are accurate and reasonable.
Figure 9
Figure 9. Impacts of climate changes and human activities on runoff variation using the DMC method in the (a) UHB and (b) MHB. Note: ‘ + ’ and ‘-’means that runoff increased or decreased, respectively.
View largeDownload slide

Impacts of climate changes and human activities on runoff variation using the DMC method in the (a) UHB and (b) MHB. Note: ‘ + ’ and ‘-’means that runoff increased or decreased, respectively.

Figure 9
Figure 9. Impacts of climate changes and human activities on runoff variation using the DMC method in the (a) UHB and (b) MHB. Note: ‘ + ’ and ‘-’means that runoff increased or decreased, respectively.
View largeDownload slide

Impacts of climate changes and human activities on runoff variation using the DMC method in the (a) UHB and (b) MHB. Note: ‘ + ’ and ‘-’means that runoff increased or decreased, respectively.

Close modal

Analysis of the performances

In the UMHB, runoff variation was mainly induced by human activities. However, the influences of climate change reached over 70% in the UHB, and human activities over 83% in the MHB. The difference between the basin and its sub-basins may be ascribed to scaling factors. Still, the study was consistent with the fact that climate change has played a principal role in the alteration of hydrological processes in the mountainous region of the UHB, and human activities acted vitally in the MHB (Wang et al. 2007; Nian et al. 2014). In addition, uncertainties in the methods contributed to performance variations. Furthermore, the trends of precipitation and runoff in the UHB were opposite to those in the MHB. Then the variations of hydrological elements in the upstream and midstream regions offset. Some anthropogenic effects increased streamflow, such as afforestation and crop replacement for saving water, while others decreased streamflow, such as irrigation and damming. However, the performances of each method quantitatively documented the general facts associated with the study area. Within areas of rapid socio-economic and agricultural development, the impacts of human activities and demand for water resources would intensify. Meanwhile, climate change has increased the probability of extreme climate events, especially droughts in northwestern China. In recent decades, runoff in the midstream region exhibited a decreasing trend. Countermeasures should be taken to improve the situation and promote sustainable water resources development.

Identifying the impacts of climate changes and human activities on runoff would help to demonstrate the key driving forces affecting hydrological process and reveal the essential mechanisms of such hydrological disturbances. This study separated the effects of climate changes and human activities on the runoff regimes in the upper and middle reaches of the Heihe River basin.

MK and Pettitt tests determined the runoff change point to be 1980, based on a time series from 1964–2006. The DMC and HSB methods were applied to quantitatively assess the impacts of climate changes and human activities on runoff variation in the upstream and midstream reaches of the basin, as well as their sub-basins, to analyse the major processes affecting the basin. The application of the two methods in the study exhibited good adaptability.

Human activities accounted for 53–55% and climate changes for 45–47% of the runoff variation in the upper and middle reaches of the Heihe River basin. Climate changes contributed to 71% of the runoff variation in the upstream region, while human activities contributed to 83% of the variation in the midstream region, respectively. The results indicate that human activities play a key role in runoff variation over the entire study area. Climate changes are the major driving forces in the upstream region, while human activities have more significant impacts in the midstream region.

This study is financially supported by the State Key Program of National Natural Science of China (91125015). We also acknowledge the support of the Cold and Arid Regions Science Data Centre at Lanzhou and the Hydrology and Water Resources Survey Bureau of Gansu Province, China.

Allen
R. G.
Pereira
L. S.
Raes
D.
Smith
M.
2006
FAO Irrigation and Drainage Paper No. 56
.
Beguería
S.
López-Moreno
J. I.
Lorente
A.
Seeger
M.
García-Ruiz
J. M.
2003
Assessing the effects of climate oscillations and land-use changes on streamflow in the Central Spanish Pyrenees
.
Ambio
32
,
283
–
286
.
Google Scholar
Crossref
Search ADS
PubMed
 
DeFries
R.
Eshleman
N. K.
2004
Land-use change and hydrologic processes: a major focus for the future
.
Hydrol. Process
.
18
,
2183
–
2186
.
Google Scholar
Crossref
Search ADS
 
Du
J.
He
F.
Zhang
Z.
Shi
P. J.
2011
Precipitation change and human impacts on hydrologic variables in Zhengshui River Basin, China
.
Stoch. Environ. Res. Risk Assess
.
25
,
1013
–
1025
.
Google Scholar
Crossref
Search ADS
 
He
X. Q.
Zhang
B.
Sun
L. W.
Jin
S. L.
Zhao
Y. F.
An
M. L.
2012
Contribution rates of climate change and human activity on the runoff in upper and middle reaches of Heihe River basin
.
Chin. J. Ecol.
31
,
2884
–
2890
(
in Chinese
).
Google Scholar
 
Keller
C. F.
2009
Global warming: a review of this mostly settled issue
.
Stoch. Environ. Res. Risk Assess
.
23
,
643
–
676
.
Google Scholar
Crossref
Search ADS
 
Kendall
M. G.
1975
Rank Correlation Methods
. 4th edn,
Griffin
,
London
.
Google Scholar
 
Kun
F.
Chen
X. P.
Liu
Q. G.
Li
C. H.
2007
Land use and land cover changes based on remote sensing and GIS in Heihe River basin, China. IGARSS: 2007 IEEE International Geoscience and Remote Sensing Symposium
. In:
Sensing and Understanding our Planet. pp.
1790
–
1793
. doi:10.1109/IGARSS.2007.4423168.
Google Scholar
 
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 streamflow from the Wuding River basin in China
.
Hydrol. Process
.
21
,
3485
–
3491
.
Google Scholar
Crossref
Search ADS
 
Lin
E.
Xu
Y.
Wu
S.
Ju
H.
Ma
S.
2007
China's National Assessment Report on Climate Change (II): climate change impacts and adaptation
.
Adv. Clim. Change Res.
3
(
Suppl.
),
6
–
11
.
Google Scholar
 
Mann
H. B.
1945
Nonparametric tests against trend
.
Econometrica
13
,
245
–
259
.
Google Scholar
Crossref
Search ADS
 
Milly
P.
Dunne
K. A.
2002
Macroscale water fluxes - 2. Water and energy supply control of their interannual variability
.
Water Resour. Res
.
38
,
241
–
249
. doi:10.1029/2001WR000760.
Google Scholar
Crossref
Search ADS
 
Peng
D. Z.
Qiu
L. H.
Lin
H. J.
2014
Identification for the impact of climate change and human activities on streamflow in Xitiaoxi river basin, China
. In:
6th International Conference on Flood Management. ICFM6
.
Brazilian Water Resources Association
,
Sao Paulo
,
Brazil
.
Google Scholar
 
Ramanathan
V.
Crutzen
P. J.
Kiehl
J. T.
Rosenfeld
D.
2001
Atmosphere-aerosols, climate, and the hydrological cycle
.
Science
294
,
2119
–
2124
.
Google Scholar
Crossref
Search ADS
PubMed
 
Searcy
J. K.
Hardison
C. H.
1960
Double mass curves
.
Water-Supply Paper 1541-B
.
US Geological Survey
,
Reston, VA
.
Google Scholar
 
Vogel
R. M.
2011
Hydromorphology
.
J. Water Resour. Plann. Manage
.
137
,
147
–
149
.
Google Scholar
Crossref
Search ADS
 
Willett
K. M.
Gillett
N. P.
Jones
P. D.
Thorne
P. W.
2007
Attribution of observed surface humidity changes to human influence
.
Nature
449
,
710-U6
.
Google Scholar
Crossref
Search ADS
 
Zhang
Y. F.
Guan
D. X.
Jin
C. J.
Wang
A. Z.
Wu
J. B.
Yuan
F. H.
2011
Analysis of impacts of climate variability and human activity on streamflow for a river basin in northeast China
.
J. Hydrol
.
410
,
239
–
247
.
Google Scholar
Crossref
Search ADS
 
© IWA Publishing 2016