Introduction

In recent decades, global climate change and human activities have accelerated the global hydrological cycle (Allen and Ingram 2002; Stocker and Raible 2005; Xu et al. 2013a, b, c; Millan 2014; Amin et al. 2017) and have changed the spatiotemporal pattern of rainfall and evaporation, which is called hydrological effect that has drawn wide attention (Beniston 2002; Coulibaly 2006; Chu et al. 2008; Barnett et al. 2008; Hsu and Li 2010; Piao et al. 2010; Liang et al. 2011; Liu et al. 2014; Romanowicz et al. 2016; Meresa et al. 2017). The runoffs of many rivers around the world exhibit a significant decreasing trend (Fu et al. 2004; Xu et al. 2010; Wang et al. 2012a; Li and Zhou 2016; Zhang et al. 2016) that greatly threatens global water security and brings great challenges to the conflict between water supply and demand during the city development. Compared with climate change, human activities exhibit a more significant contribution to the runoff change in China. It can be proved by the fact that in the past 100 years and in the past 50 years, there was no obvious change in annual precipitation, whereas runoffs exhibited a significant decreasing trend (Ding et al. 2006).

Qinzhou, being an important node of the “Belt and Road,” facing “China-ASEAN” and relying on Beibu Gulf Urban Agglomeration Plan (2017), will certainly usher in a leap-forward development. However, as a coastal city, the rivers in Qinzhou are mainly small and medium rivers with short source and rapid water flow. The rainfall in the flood season accounts for over 80% of the year. There are 393 different types of reservoirs in the city. However, the total capacity of all the reservoirs is only 4.85 × 108 m3 (Chen et al. 2016). The Survey and Evaluation of China’s Offshore Oceans in 2012 indicated that nearly 90% of the cities had different degrees of water shortage.

Under the background that the conflicts between the regional population growth, economic development and the shortage of water resources have become increasingly serious, analyzing the hydrological evolution under the dual influence of climate change and human activities and the contribution rate of rainfall, evaporation, temperature and other factors to runoff changes is the premise of analyzing the potential of water resources and carrying out correct evaluation and allocation of water resources.

Quantifying the impacts of climate change and human activities on runoff change has become a hot topic in climatic and hydrological researchers (Yuan et al. 2016; Zhang et al. 2016). Previous reports have indicated that human activities have altered the water cycle and have increased runoff loss. Researchers have utilized a variety of methods to evaluate the contribution rates of the main influencing factors to runoff changes and have revealed that the runoff changes involve the superposition of the effects of climate change and human activity with human activities being the dominant factor (Zhan et al. 2013; Gao et al. 2017). The effects of climate are mainly reflected in inter-annual variability, multi-timescale and future runoff sustainability (Wu et al. 2017). On the analysis of the contribution rate of each factor to runoff changes under the environmental change, the research methods that can quantitatively assess the hydrological influence of climate change and other factors are mainly hydrological model method (Wang et al. 2006; Xia et al. 2014; He et al. 2015) and quantitative assessment method in watershed scale. The quantitative assessment method includes mainly climate elasticity method (Fu et al. 2007; Hu et al. 2012), multivariate regressive method (Xu 2011), sensitivity analysis method (Jones et al. 2006), water balance method (Wang et al. 2009) and double-mass curve of runoff and precipitation (Guo et al. 2014; Zhao et al. 2014; Niu et al. 2016). Many researchers have conducted a wide range of studies on the Yangtze River Basin (Zhao et al. 2015), the Yellow River Basin (Wang et al. 2014; Kong et al. 2016), the Wei River Basin (Huang et al. 2016) the Haihe River Basin (Luo et al. 2016) and several typical watersheds in other countries (Buendia et al. 2016; Zare et al. 2016; Farenhorst et al. 2016; Griffioen 2017; Marcos et al. 2017). Although these studies have quantitatively assessed the comprehensive effects of climate change and human activities on runoff change, there are still some shortcomings and deficiencies in research methods and research areas. For example, the hydrological model method has a good physical foundation, but the sensitivity of the parameter has several uncertainties. If the simulation results are not verified, there will a great possibility that the contribution rate of climate change to runoff change is higher than its actual value (Legesse et al. 2003); although the quantitative evaluation method requires fewer data and needs a long series of data, the noise in the long series will cause interference to the evaluation results (Hu et al. 2012). As a result, the main driving factors in water resource change and the contribution of water resource researchers and managers to runoff change remain difficult (Wu et al. 2017).

In 2012, Wang et al. (2012b) formulated slope changing ratio of cumulative quantity (SCRCQ for short) which can effectively eliminate noise and separate the contribution rate of climate change and other factors in runoff change. With these advantages, it has preferable application in the middle and upper reaches of Heihe River Basin (He et al. 2012), the Mid-Route of South-to-North Water Transfer Project (Ling et al. 2014), Huangshui River Basin (Zhang et al. 2014), three outlets of the Dongting Lake (Li et al. 2016), three outlets along Jingjiang in the middle reach of Yangtze River (Shuai et al. 2016) and karst watersheds in Guizhou Province (Wu et al. 2017). However, the assumed reference periods of the studies based on SCRCQ method were the actual runoff series which did not fully consider the effects of air temperature on rainfall and evaporation and did not explore the complex internal relations of rainfall, evaporation and temperature on runoff change in depth. All these led to a smaller contribution rate of climatic factors than its actual value and led to a great result error. In view of the problems existing in the studies, it is necessary to make appropriate improvements to the SCRCQ method: Redefining the reference period, taking consideration of the influence of air temperature on rainfall and evaporation, exploring the internal relations between climatic factors and runoff change and getting the contribution rate of climatic factors indirectly make great contribution to more accurate data of study results with the application of SCRCQ method.

Study site and materials

Study site

The Qinjiang River is an outer river, and the Qinjiang River Basin (108.561°–109.512°E, 21.856°–22.587°E) is a small and medium watershed that is located in the southernmost part of Guangxi Zhuang Autonomous Region and near the Beibu Gulf (Fig. 1a). The Qinjiang River Basin is the largest basin in Qinzhou. The mainstream of the river basin is 191 km, the water catchment area is 2326 km2, and the slope of the mainstream is 0.36% (Fig. 1b). The altitude of the river is − 13 to 846 m. The terrain is high in the northeast and low in the southwest. The lower reaches of the basin are coastal alluvial plain. Because of these reasons, the lower reaches are the tidal reaches, and the tidal phenomenon is obvious (Fig. 1c).

Fig. 1
figure 1

Research regional geographic location (a), water system, station network and hydraulic engineering (b), topographic features (c), rainfall spatial distribution (d)

Full size image

The Qinjiang River Basin is located in the humid south subtropical monsoon climate zone, which has the features of oceanic monsoon climate, that transits from subtropics to tropics. The annual sunshine time of the Qinjiang River Basin is about 1800 h, the annual average temperature is 21.7–22 °C, and the annual evaporation is 929.6 mm (data from 1956 to 2016). Because the Qinjiang River Basin is near to the Beibu Gulf and it is often influenced by tropical cyclones in summer and autumn, the rainfall is plentiful with an annual average rainfall of 1743.7 mm. The spatial distribution of rainfall presents a trend of more in the south and less in the north (Fig. 1d). The annual rainfall in this basin focuses mainly on the flood season from April to September, and the rainfall in the flood season accounts for 81.3% of the annual rainfall.

With plentiful water resource, the Qinjiang River is one of the most important water source areas in Qinzhou. The unequal distribution of rainfall in time and space which focuses mainly on the flood season and lack of key water conservancy projects (one large reservoir with capacity of which 0.79 × 108 m3 and two medium-sized reservoirs with capacity of 0.204 × 108 m3) has led to difficult utilization of water resource and has led to serious conflicts between water supply and demand.

Materials

There are 12 rainfall stations, one hydrological station (Luwu hydrological station), two evaporation stations (Luwu and Huang Wutun evaporation station) and two temperature stations (Luwu and Qinzhou Meteorological station) in the study area. The hydrological data were mainly the monthly runoff data of the Qinjiang River Basin (from January 1956 to December 2016). The meteorological data included data on monthly rainfall, evaporation and air temperature. The monthly rainfall and evaporation data were interpolated through the combination of rainfall and evaporation data with data of the surrounding stations and application of the Kriging method of ArcGIS 10.

Data analysis method

Continuous wavelet analysis

Continuous wavelet analysis based on the Morlet function (Morlet et al. 1982; Christopher and Webster 1998) has been widely used to identify periodic oscillations of signals (Labat 2005; Werner 2008; Tongal et al. 2017). In recent years, continuous wavelet analysis has been applied to the analysis of multiple timescale features of hydrological and meteorological research. This method can clearly reveal a variety of changes in the time series and fully reflect the change trend of hydrological and meteorological data in different timescales (Wu et al. 2017). Therefore, based on Morlet wavelet analysis, this study intended to analyze the rainfall, runoff and air temperature sequences of the Qinjiang River since years and explore their evolution rules in specific timescale.

The basic concept of wavelet analysis is to approximate the desired function by a cluster of wavelet functions (Torrence et al. 1998). Therefore, the key to wavelet analysis is the correct choice of wavelet function. The selected wavelet function must have the characteristics of oscillation and rapid attenuation to zero. Besides, wavelet function \(\varPsi (t) \in L^{2} (R)\) should meet:

$$\int_{ - \infty }^{ + \infty } {\varPsi (t){\text{d}}t} = 0,$$
(1)
$$\varPsi_{n} (a,b) = \left| {a^{{ - \frac{1}{2}}} } \right|\int_{ - \infty }^{ + \infty } {x(t)\varPsi (\frac{t - b}{a}){\text{d}}t} ,$$
(2)

where \(\varPsi (t)\) is the wavelet basis function; \(\varPsi_{n} (a,b)\) is the daughter wavelet; \(a,b \in R\), \(a \ne 0\), a is scale factor that can reflect the length of the cycle; b is the shift factor that mainly reflects influence by time shifting.

The wavelet variance is to integrate the squared value of the selected wavelet coefficients on the b domain. The formula is as follows:

$$\text{var} (a) = \int_{ - \infty }^{ + \infty } {\left| {\psi_{n} (a,b)} \right|^{2} {\text{d}}b} .$$
(3)

Wavelet variogram can be used to identify the relative intensity of different scales of disturbance in a signal, the timescale and the dominant periods of the disturbance.

Hurst exponent

Rescaled range analysis is a time-series method which was formulated by British hydrologist Harold Edwin Hurst (1956). Mandelbrot and Wallis (1969) have improved this theory. The rescaled range analysis method has been widely applied in the fields of hydrology, climatology, economics, geology geochemistry, etc. (Wu et al. 2017; Wang et al. 2002; Tongal et al. 2013).

The basic principle of the R/S analysis is as follows: Define a time series \(\left\{ {\xi (t)} \right\},t = 1,2, \ldots .\), and for any positive integer \(\tau \ge 1\), the mean series can be defined as

$$\bar{\xi }_{\tau } = \frac{1}{\tau }\sum\limits_{t = 1}^{\tau } {\xi (t),\tau = 1,2, \ldots .}$$
(4)

\(X(t)\) represents accumulated deviation

$$X(t,\tau ) = \sum\limits_{u = 1}^{\tau } {(\xi (u) - \bar{\xi }_{\tau } ),\;1 \le t \le } \tau$$
(5)

Range \(R(\tau )\) can be defined as

$$R(\tau ) = \mathop {\hbox{max} }\limits_{1 \le t \le \tau } X(t,\tau ) - \mathop {\hbox{min} }\limits_{1 \le t \le \tau } X(t,\tau ),\;\tau = 1,2, \ldots .$$
(6)

Standard deviation \(S(\tau )\) can be defined as

$$S(\tau ) = \left( {\frac{1}{\tau }\sum\limits_{t = 1}^{\tau } {(\xi (t) - \bar{\xi }_{\tau } )^{2} } } \right)^{{\frac{1}{2}}} ,\;\tau = 1,2, \ldots .$$
(7)

Considering ratio \({{R(\tau )} \mathord{\left/ {\vphantom {{R(\tau )} {S(\tau ) \triangleq {R \mathord{\left/ {\vphantom {R S}} \right. \kern-0pt} S}}}} \right. \kern-0pt} {S(\tau ) \triangleq {R \mathord{\left/ {\vphantom {R S}} \right. \kern-0pt} S}}}\), Hurst found the following empirical correlation

$$\frac{R}{S} \propto \left( {\frac{\tau }{2}} \right)^{H} .$$
(8)

In the formula, H (0 < H < 1) is Hurst index.

If \(\left\{ {\xi (t)} \right\}\) is a random sequence which is independent of each other and of which the variance is limited, the following result was proved by Hurst (1951) and Feller (1951):

$$\frac{R}{S} = \left( {\frac{\pi \tau }{2}} \right)^{{{1 \mathord{\left/ {\vphantom {1 2}} \right. \kern-0pt} 2}}} ,$$
(9)

that is, H = 1/2.

Hurst phenomenon exists in a change in rainfall, runoff, evaporation and temperature series. When 0.5 < H < 1, the series present persistence that means that time series are not independent of each other but have a certain memory or present positive correlation; when 0 < H < 0.5, the series present anti-persistence and the negative correlation, which indicates that the change in the future time series presents an opposite trend to the past. The grading intensity table (Table 1) of the Hurst exponent describes quantitatively the intensity of future trend components. The grading intensity of the Hurst exponent is divided into five grades from weak to very strong (Fan et al. 2008).

Table 1 Classification of Hurst exponent
Full size table

Mann–Kendall

Trend analysis is generally divided into parameters and nonparametric tests. Parameter test is more powerful than nonparametric test but requires independent and normally distributed data. On the other hand, the nonparametric test only needs independent data and can tolerate abnormal data (Hamed et al. 1998). With the combination of least square method and Mann–Kendall, this study quantitatively analyzed the trend of time series (Pingale et al. 2013).

If \(x_{i}\) is a certain climate variable with a sample size of n and \(t_{i}\) is the time that is corresponding to \(x_{i}\), the following linear regression equation can be formulated.

$$x_{i} = a + bt_{i} ,$$
(10)
$$b = {{\left[ {\sum\limits_{{i = 1}}^{n} {x_{i} t_{i} - \frac{1}{n}\left( {\sum\limits_{{i = 1}}^{n} {x_{i} } } \right)\left( {\sum\limits_{{i = 1}}^{n} {t_{i} } } \right)} } \right]} \mathord{\left/ {\vphantom {{\left[ {\sum\limits_{{i = 1}}^{n} {x_{i} t_{i} - \frac{1}{n}\left( {\sum\limits_{{i = 1}}^{n} {x_{i} } } \right)\left( {\sum\limits_{{i = 1}}^{n} {t_{i} } } \right)} } \right]} {\left[ {\sum\limits_{{i = 1}}^{n} {t_{i}^{2} - \left( {\sum\limits_{{i = 1}}^{n} {t_{i} } } \right)^{2} } } \right]}}} \right. \kern-\nulldelimiterspace} {\left[ {\sum\limits_{{i = 1}}^{n} {t_{i}^{2} - \left( {\sum\limits_{{i = 1}}^{n} {t_{i} } } \right)^{2} } } \right]}}.$$
(11)

Mann–Kendall trend test (Mann 1945; Kendall 1975) is a method to estimate the correlation between series based on the order of time series (Hamed 2008). Mann–Kendall is the nonparametric test. Because it is not influenced by sample value and distribution type, it has been widely used to identify the trend of time series of precipitation, runoff, air temperature and water quality trends (Wu et al. 2017; Yu and Chen 2011). The formula of the statistical amount (S) is as follows:

$$S = \sum\limits_{i = 2}^{n} {\sum\limits_{j = 1}^{i - 1} {{\text{sign}}(x_{i} - x_{j} )} } ,$$
(12)

where \({\text{sign}}\) is symbol function and when \(x_{i} - x_{j}\) is less than, equal to or greater than 0, \({\text{sign}}(x_{i} - x_{j} )\) is − 1, 0 or 1, respectively.

The sequence of Z (trend changes) is, respectively, as follows:

$$Z = \left\{ \begin{aligned} {{(S - 1)} \mathord{\left/ {\vphantom {{(S - 1)} {\sqrt {\frac{n(n - 1)(2n - 1)}{18}} \begin{array}{*{20}c} , & {S > 0} \\ \end{array} }}} \right. \kern-0pt} {\sqrt {\frac{n(n - 1)(2n - 1)}{18}} \begin{array}{*{20}c} , & {S > 0} \\ \end{array} }} \hfill \\ 0\begin{array}{*{20}c} {} & {} & {} & {\begin{array}{*{20}c} {} & {} & {\begin{array}{*{20}c} {} & {} & , &\quad {S = 0} \\ \end{array} } \\ \end{array} } \\ \end{array} \hfill \\ {{(S + 1)} \mathord{\left/ {\vphantom {{(S + 1)} {\sqrt {\frac{n(n - 1)(2n - 1)}{18}\begin{array}{*{20}c} , &\quad {S < 0} \\ \end{array} } }}} \right. \kern-0pt} {\sqrt {\frac{n(n - 1)(2n - 1)}{18}\begin{array}{*{20}c} , &\quad {S < 0} \\ \end{array} } }} \hfill \\ \end{aligned} \right..$$
(13)

When Z is a positive value, it presents an increasing trend, and when Z is a negative value, it presents a decreasing trend. When the absolute value of Z is 1.28, 1.64 or 2.32, it indicates that the trend change passes the significance test with the confidence of 90%, 95% or 99%, respectively, and can be marked with *, ** or ***, respectively.

In addition to trend analysis, Mann–Kendall can also be used to test time-series mutations. Mutation test statistic calculation: According to \(x_{1} ,x_{2} , \ldots ,x_{n}\), construct a rank order \(m_{i}\), \(m_{1} ,m_{2} , \ldots ,m_{n}\) and \(m_{i}\) is the cumulative number of sample \(x_{i} > x_{j} (1 \le j \le i)\).

Define variables \(d_{k}\):

$$d_{k} = \sum\limits_{i}^{k} {m_{i} } (2 \le k \le N).$$
(14)

Define average value and variance of \(d_{k}\) as follows:

$$E(d_{k} ) = {{k(k - 1)} \mathord{\left/ {\vphantom {{k(k - 1)} 4}} \right. \kern-0pt} 4},$$
(15)
$${\text{Var}}(d_{k} ) = {{(k(k - 1)(2k + 5))} \mathord{\left/ {\vphantom {{(k(k - 1)(2k + 5))} {72}}} \right. \kern-0pt} {72}}\;(2 \le k \le N).$$
(16)

Under the stochastic independent of time series, \({\text{UF}}_{k}\) can be defined as follows:

$${\text{UF}}_{k} = {{(d_{k} - E(d_{k} ))} \mathord{\left/ {\vphantom {{(d_{k} - E(d_{k} ))} {\sqrt {{\text{Var}}(d_{k} )} }}} \right. \kern-0pt} {\sqrt {{\text{Var}}(d_{k} )} }}\;(k = 1,2, \ldots ,n),$$
(17)

where \({\text{UF}}_{k}\) is standard normal distribution. Given a certain significant level, a critical value can be found in normal distribution table. When \({\text{UF}}_{k} > U_{\alpha }\), it indicates that the series presents a significant increasing or decreasing trend and all \({\text{UF}}_{k}\) will compose a curve UF. By applying this method to antitone sequence \((x_{n} ,x_{n - 1} , \ldots ,x_{1} )\), \({\text{UB}}_{k}\) will be obtained and all values of \({\text{UB}}_{k}\) compose the curve \(U_{B}\).

In mutation test analysis chart, when the curve \(U_{F}\) exceeds the confidence test critical line of confidence test, it indicates that the trend of change is significant, and the range beyond the critical line is the time zone where the mutation occurs. If \(U_{F}\) and \(U_{B}\) have an intersection and the intersection point lies between the critical lines of the confidence test, the time corresponding to the intersection point is the time when the mutation starts.

Cumulative anomaly

Cumulative anomalies are commonly used to determine the change in hydrological and meteorological factors such as precipitation, evaporation, runoff and air temperature series. The core of this method is to judge the discrete amplitude of the discrete data to its mean value, and if the cumulative distance value increases, it indicates that the discrete data are larger than its average value. Instead, the discrete data are smaller than the average value (Ran et al. 2010). For discrete sequence \(x_{i}\), cumulative anomalies \(S_{t}\) of hydrological and meteorological data points \(x_{i}\) can be represented as

$$S_{t} = \sum\limits_{i = 1}^{t} {(x_{i} - \bar{x})} \;t = 1,2, \ldots ,n,$$
(18)
$$\bar{x} = \frac{1}{n}\sum\limits_{i = 1}^{n} {x_{i} } ,$$
(19)

where \(\bar{x}\) is the average of the discrete point series \(x_{i}\) and n is the number of discrete points (Wang et al. 2012a).

Slide t test technique

Because of inconsistent results of abrupt change points between Mann–Kendall and cumulative anomaly methods, it is necessary to use slide t test as an assistant method to test the significance of the time-series mutation. Slide t test is used to test the significant difference between the mean values of the two random samples (Afifi et al. 1972; Fu et al. 1999). Therefore, we divide the continuous random variable \(x\) into two subsample sets \(x_{1}\) and \(x_{2}\) and define the mean, variance and sample length of \(x_{1}\) as \(\mu_{i}\), \(S_{i}^{2}\) and \(n_{i}\), respectively, where \(n_{i}\) requires a man-defined length.

The null hypothesis \(H_{0}\): \(\mu_{1} - \mu_{2} = 0\). The statistic is defined as follows:

$$t_{0} = \frac{{\bar{x}_{1} - \bar{x}_{2} }}{{S_{P} \left( {\frac{1}{{n_{1} }} + \frac{1}{{n_{2} }}} \right)^{{{1 \mathord{\left/ {\vphantom {1 2}} \right. \kern-0pt} 2}}} }},$$
(20)

where \(S_{{_{P} }}^{2}\) is joint sample variance.

$$S_{{_{P} }}^{2} = \frac{{(n_{1} - 1)S_{{_{1} }}^{2} + (n_{2} - 1)S_{2}^{2} }}{{n_{1} + n_{2} - 2}}.$$
(21)

\(S_{{_{P} }}^{2}\) is an unbiased estimation of \(\sigma^{2}\)\((E[S_{{_{P} }}^{2} ] = \sigma^{2} )\) and \(t_{0}\) obeys normal distribution \(t_{0} \sim t(n_{1} + n_{2} - 2)\). Given the reliability \(\alpha\), the critical value \(t_{\alpha }\) can be obtained. After figuring out \(t_{0}\), compare \(t_{0}\) and \(t_{\alpha }\) under \(H_{0}\). When \(\left| {t_{0} } \right| \ge t_{\alpha }\), the null hypothesis \(H_{0}\) should be rejected. This indicates that \(t_{0}\) and \(t_{\alpha }\) have significant difference. When \(\left| {t_{0} } \right| < t_{\alpha }\), the null hypothesis \(H_{0}\) should be accepted. Because of artificiality in the selection of \(n_{i}\), there will be a certain degree of confusion. In a real application, \(n_{i}\) should be determined by real need and should be constantly changed in order to enhance the reliability of inspection results (Fu et al. 1992).

Evaporation interpolation method

Evaporation difference method obeys the principle of water balance (Li et al. 2004). The changes in surface water resources \(R\) are tightly associated with rainfall \(P\), the basin evaporation \(E_{\text{before}}\) before the changes in the underlying surface of the basin and soil water storage capacity \(W\). \(W\) can be expressed as a function of temperature \(T\) and rainfall \(P\) as shown in formula (22).

$$R = P - E_{\text{before}} - W = f(P,E_{\text{before}} ,T),$$
(22)

where \(E_{\text{before}}\) can be expressed as:

$$E_{\text{before}} { = }{{\left[ {(A - A_{1} )E_{\text{land}} + \alpha A_{1} E_{\text{water}} } \right]} \mathord{\left/ {\vphantom {{\left[ {(A - A_{1} )E_{\text{land}} + \alpha A_{1} E_{\text{water}} } \right]} A}} \right. \kern-0pt} A},$$
(23)

where \(A\) is the total area of the tested basin (km2); \(A_{1}\) is the surface area of reservoirs, lakes and ponds in the basin area before human activities (km2); \(\alpha\) is the converted coefficient of evaporation of water surface; \(E_{\text{water}}\) is the evaporation of evaporating dish (mm); \(E_{\text{land}}\) is the land evaporation.

The study area is located in the southern coastal area. Therefore, the land evaporation should refer to the Kaijiang River evaporation formula (Jin et al. 1981):

$$E_{\text{land}} = {{(\alpha T + \beta )\theta } \mathord{\left/ {\vphantom {{(\alpha T + \beta )\theta } L}} \right. \kern-0pt} L},$$
(24)
$$\theta = \theta_{0} \left[ {0.65 - 0.35(n - s)} \right],$$
(25)

where \(\theta\) is total solar radiation (cal/cm2); \(T\) is daily mean air temperature (°C); \(L\) is latent heat of vaporization; \(\alpha\) and \(\beta\) is the coefficient and the constant, respectively; \(\theta_{0}\) is total solar radiation when the sky is cloudless and blue (cal/cm2); \(n\) is monthly mean cloud cover; \(s\) is monthly mean amount of sunshine.

SCQCR

Proposed by Wang et al. (2012b), the SCRCQ method was revised from the slope changing ratio of accumulative quantity (SCRAQ) to account for the contributions of influence factors to runoff change in different periods. The influence factors of runoff change in river basin usually cover climatic factors and human activities (Wang et al. 2012a).

Based on the SCRCQ method, the base period is redefined as the reduced runoff series before the first abrupt change point of the measured runoff series. The study period is measured runoff series after the mutation point. The slope of the linear equation of the cumulative rainfall and year during the base period and the study period is \(K_{\text{Pa}}\)(mm/a) and \(K_{\text{Pb}}\)(mm/a), respectively, and the slope varying ratio is \(S_{P}\). The slope of the linear equation of the cumulative evaporation and year linear equation during the base period and the study period is \(K_{Ea}\)(mm/a) and \(K_{Eb}\)(mm/a), respectively, and the slope varying ratio is \(S_{E}\). The slope of the linear equation of the cumulative runoff and year during the base period and the study period is \(K_{\text{Ra}}\)(108 m3/a) and \(K_{\text{Rb}}\)(108 m3/a), respectively, and the slope varying ratio is \(S_{R}\); the slope of the linear equation of the cumulative air temperature and year during the base period and the study period is \(K_{\text{Ta}}\)(°C/a) and \(K_{\text{Tb}}\) (°C/a), respectively, and the slope varying ratio is \(S_{T}\).

$$S_{P} = (K_{\text{Pb}} - K_{\text{Pa}} )/K_{\text{Pa}} ,$$
(26)
$$S_{E} = (K_{\text{Eb}} - K_{\text{Ea}} )/K_{\text{Ea}} ,$$
(27)
$$S_{R} = (K_{\text{Rb}} - K_{\text{Ra}} )/K_{\text{Ra}} ,$$
(28)
$$S_{T} = (K_{\text{Tb}} - K_{\text{Ta}} )/K_{\text{Ta}} .$$
(29)

According to formula (1–4), the contribution rate of rainfall, evaporation and air temperature to runoff change can be obtained as follows, respectively: \(C_{P}\), \(C_{E}\) and \(C_{T}\):

$$C_{p} = 100 \times S_{P} /S_{R} ,$$
(30)
$$C_{E} = 100 \times S_{E} /S_{R} ,$$
(31)
$$C_{T} = 100 \times S_{T} /S_{R} .$$
(32)

According to \(C_{P}\), \(C_{E}\) and \(C_{T}\), we can indirectly obtain the contribution rate of human activities to runoff. According to the principle of water balance, as formula (22), T is independent variable and R is the dependent variable; T can directly cause a change in R and the contribution rate is \(C_{T}\); besides, change in T can cause a change in P and E and affect R indirectly, and the contribution rate is also \(C_{T}\). However, the \(C_{T}\) here is calculated repeatedly. Therefore, \(C_{O}\) is not a constant value and should change in the range of lower limit that is under the influence of T and an upper limit that is not under the influence of T. \(C_{O}\) is shown as formula (33).

$$C_{O} \in [100 - C_{p} - C_{E} - C_{T} ,100 - C_{p} - C_{E} ].$$
(33)

Results

Analysis of the evolution of hydrological sequence

Inter-annual change characteristics

Figure 2 presents the inter-annual change trend of runoff depth (Luwu station), rainfall, evaporation and air temperature in the Qinjiang River Basin. The black line, red line, green line and orange line represent the inter-annual change, 3-year running mean value, the average for many years and time-series inter-annual trend line, respectively. The pink rectangle area and the green rectangle represent the above mean period (AMP) and below mean period (BMP), respectively.

Fig. 2
figure 2

Inter-annual change characteristics of hydrological and meteorological time series (a) runoff, (b) rainfall, (c) evaporation, (d) temperature

Full size image

According to 3-year running mean, it can be concluded that the rainfall, runoff depth and evaporation in the Qinjiang River Basin presented obvious inter-annual change. Around 1986, the rainfall, runoff depth and evaporation in the Qinjiang River Basin changed from to AMP to BMP while the air temperature changed from BMP to AMP. From 1956 to 2016, the rainfall amount in the Qinjiang River Basin overall showed an increasing trend of 0.808 mm/a (Fig. 2a). The result of M–K (Table 2) showed that its increasing trend was not significant and the significance level (Z = 0.467, \(\left| Z \right| \le 1.28\)) was less than 90%. The result of Hurst indicated that the sequence presented strong persistence (H = 0.655). The runoff depth showed a significant (Z = − 2.172**) decreasing trend of − 3.827 mm/a (Fig. 2b). However, its decreasing trend will present a weak persistence in the future. The evaporation series showed a decreasing trend of − 1.404 mm/a (Fig. 2c), and this decreasing trend exhibited significance (Z = − 2.259**) and strong persistence (H = 0.824). Despite the significant decrease in evaporation, the air temperature of the Qinjiang River basin was increasing (Fig. 2d). In the period from 1956 to 2016, the air temperature of the Qinjiang River Basin increased by 1.513 °C, and this significant increasing trend (Z = 5–5.364***) will present strong persistence in the future (H = 0.777).

Table 2 Sequence change trend and continuity
Full size table

In general, rising temperatures cause an increasing in evaporation (Roderick et al. 2002; Ohmura and Wild 2002). However, observations data indicate that the evaporation in the evaporating dish shows a significant downward trend (Peterson et al. 1995). The phenomenon is called the evaporation paradox (Roderick et al. 2002). At present, it is believed that the change in evaporation is mainly related to temperature, solar radiation, cloud amount, relative humidity, wind speed, etc. (Qi et al. 2015). Qi et al. (2015) found that although the temperature in southern China increased significantly, the evaporation of evaporating dish showed a significant downward trend. In this study, the evaporation of the basin is based on the temperature, further considering solar radiation, the average cloud amount and the amount of sunshine shown in Eq. (23), and the research results are similar to the studies above.

To sum up, in the future, the rainfall and air temperature series in the Qinjiang River Basin will present a weak increasing trend and a significant increasing trend, respectively, whereas the runoff and evaporation series will present a significant decreasing trend in the future.

Abrupt change characteristics

By using Mann–Kendall method (Fig. 3a–d) and cumulative anomaly method (Fig. 3e–h), the abrupt change points of rainfall, runoff depth, evaporation and air temperature series from 1956 to 2015 in the Qinjiang River Basin were detected.

Fig. 3
figure 3

Detection of abrupt change points of rainfall, runoff depth, evaporation and air temperature series of the Qinjiang River Basin based on the Mann–Kendall method (a–d) and cumulative anomaly method (e–h)

Full size image

As shown in Fig. 3a, the curve Uf and Ub intersected at the point of the year 1959, 1962, 1964, 1989, 2003, 2012 and 2014. As shown in Fig. 3e, the cumulative anomaly curves reached the extreme points in 1964, 1976, 1980, 1987, 1992, 2003 and 2011. Abrupt changes were likely to occur among the above time points. However, as the results of M–K and the cumulative anomaly method were not consistent, sliding t test was needed as an assistant method to make a further significance level test. The results of sliding t test showed that there was no significant abrupt change in rainfall series from 1956 to 2016 (all the possible abrupt change points were with a significance level of no more than 85% and with a value of 1.458), and there were only transitions at the above points.

Through the further test of the results of the Mann–Kendall method (Fig. 3b–d) and cumulative anomaly method by sliding t test (Fig. 3f–g), it could be obtained that significant abrupt changes in the runoff series of the Qinjiang River Basin occurred in 1989 (t = 2.725, with significance level over 99%) and 2003 (t = 2.391, 98%). Significant abrupt changes in evaporation occurred in 1974 (t = 3.697, 99%), 1986 (t = 2.241, 98%) and 2011 (t = 3.783, 99.9%). Significant abrupt changes in air temperature series only occurred in 1996 (t = − 7.551, 99.9%).

Multiple timescale characteristics

The multiple timescale analysis of rainfall, runoff depth, evaporation and temperature series of the Qinjiang River Basin based on Morlet wavelet method is shown in Fig. 4. The real part of wavelet coefficients in different timescales can reflect the changing characteristics of the system: The positive phase (> 0) of the real part of wavelet coefficients corresponds to years of abundant and the negative phase (< 0) of the real part of wavelet coefficients corresponds to years of rare. The bigger the absolute value of the wavelet coefficients is, the more significant the change in timescale is. The wavelet variance can reflect the distribution of wave energy of the rainfall series with a change in timescale.

Fig. 4
figure 4

Real-part contours and variogram of wavelet transform coefficient: (a1–d1) is the real-part contours of wavelet transform coefficient of rainfall, runoff depth, evaporation and air temperature, respectively, and (a2–d2) is the variogram of rainfall, runoff depth, evaporation, and temperature, respectively

Full size image

The rainfall series of the Qinjiang River Basin presented five obvious timescale changes, respectively, 3–8a, 9–17a, 18–29a, 30–42a and 43–64a (Fig. 4a1) and the period center was 6a, 13a, 23a, 34a and 53a, respectively (Fig. 4a2). The 53a timescale periodic change was extremely stable in the whole time domain with overall influence and was the first peak of the wavelet variance and the first main period of the control of rainfall series. The timescale of 34a presented an obvious conversion of abundant and rare year before 1990, whereas the center of this period went down to the period center of 23a that led to increase in the period center of 23a from 22a to 25a and led to the reappearance of timescale 9–17a which disappeared around 1965 and a more obvious periodic oscillation.

The timescale characteristic of runoff series was similar to that of rainfall. However, the difference between them was obvious. The rainfall series had four obvious timescale changes, respectively, 3–8a, 9–17a, 18–31a and 32–64a (Fig. 4b1) and the period center was 4a, 13a, 22a and 51a, respectively (Fig. 4b2). Because in runoff series, the timescale of 32–64a occupied the oscillation range of the timescale 30–42a, there were only four types of characteristic timescales. Besides the differences, the conversion of the abundant and rare year of runoff series was in good synchronization with rainfall.

The similarity between evaporation and air temperature as well as rainfall and runoff series was that their first main periods were located at 50–55a and occupied the entire time domain. Besides, they had a complete conversion of abundant and rare year and the conversion centers were almost synchronous. The difference between them was that there were only three timescale changes in the evaporation series, respectively, 7–12a, 25–35a and 40–64a (Fig. 4c1) and the period center was 8a, 30a and 55a, respectively. The 30a timescale was the second main period and presented no disappearance or oscillation weakening. The two timescale changes in the temperature series were 23–47a and 48–64a and the period centers were 43a and 55a.

Figure 4 shows that rainfall, runoff, evaporation and temperature controlled the whole period change through 53a and 23a, 51a and 22a, 55a and 30a, and 55a, respectively, namely the first and second main periods of rainfall, runoff and evaporation and the first main period of temperature.

According to the real-part changing curve of the wavelet transformation, the first main periods of rainfall (Fig. 5a) and runoff (Fig. 5b) series experienced one complete conversion from years of rare (1980–1995) and years of abundant (1996–2013) and the second main periods experienced three conversions of years of abundant (1967–1973, 1981–1987, 1996–2003) and years of rare (1974–1980, 1988–1995, 2004–2012). According to the changing rule of the first and second main periods, it can be speculated that 2014–2031 will be years of rare and 2032–2049 will be years of abundant in the timescale of 53a, whereas 2013–2019 and 2027–2033 will be years of abundant and 2020–2026 and 2034–2040 will be years of rare in the timescale of 23a. Therefore, rainfall and runoff evolution trends are predicted as follows: There will be the conversion of the abundant and rare year in 2018–2021, 2022–2030 will be years of rare and 2031–2039 will be years of abundant. Evaporation evolution trend is predicted as follows: 2019–2021 will be years of rare, there will be fluctuations of abundant and rare year and 2031 will be the year of abundant. The air temperature will experience its years of lower in 2018–2031 and its years of higher in 2031–2049. However, it will still present a significant increasing trend.

Fig. 5
figure 5

Real-part changing curve of the wavelet transformation of different main periods, a rainfall, b runoff, c evaporation, d temperature

Full size image

Attribution analysis of runoff variation

Through evaporation difference method (22), it can be obtained that the average annual surface water resource amount in the Qinjiang River Basin was 22.733 × 108 m3. Therefore, the curves of difference between natural and measured runoff and changes in the mean value of each age in the basin were obtained as shown in Fig. 6. According to Fig. 6, it can be obtained that the curves of difference presented a significant increasing trend (Z = 5.397**) with 5.096 mm/a and the mean of presented an increasing trend which indicated that human activities exerted more and more influence on runoff reduction. Therefore, it is necessary to select the typical watershed of Qinzhou to further quantitatively analyze the contribution rate of rainfall, evaporation, air temperature and human activities to runoff reduction.

Fig. 6
figure 6

Trend analysis of different curves between natural and measured runoff

Full size image

The improved SCRCQ method was used to quantitatively analyze the contribution rate of rainfall, evaporation and air temperature to runoff reduction. Meanwhile, the contribution rate of human activities to runoff reduction was indirectly analyzed.

The abrupt change years of each factor of upper catchment of the Lu Wu Station in the Qinjiang River Basin are shown in Fig. 3. The abrupt change years of runoff series were 1986 and 2003. The rainfall series presented no significant abrupt change points (for easy calculation, assuming that the abrupt change points were corresponding to that of runoff series). The abrupt change years of evaporation series were 1974, 1986 and 2011. The abrupt change year of air temperature series was 1996. If we only consider the abrupt change points of runoff series, the period from 1956 to 2013 can be divided into three periods: period AR from 1956 to 1986, period BR from 1987 to 2003 and period CR from 2004 to 2013. Period AR is base period and period BR and period CR are studied periods. Based on that, the results present bigger deviations. Therefore, according to abrupt change points, the time division of evaporation and air temperature series should be taken into consideration: Period AR of evaporation series was only from 1956 to 1973 and the evaporation series was short, which will increase contingency of the results. Therefore, this abrupt change point is eliminated. The period BR of the air temperature series was only 1997–2003, as shown in Fig. 7.

Fig. 7
figure 7

Curves of annual cumulative runoff (a), rainfall (b), evaporation (c) and temperature (d)

Full size image

Analysis of the contribution rate of climatic factors to runoff change of the upper catchment of the Lu Wu Station in the Qinjiang River Basin are shown in Table 3. Compared with period AR, the slope of linear equation of the cumulative runoff depth of period BR reduced by 2216 mm/a with the reduction rate of 23.75%; the slope of linear equation of the cumulative rainfall of period BR increased by 49.30 mm/a with the increase rate of 2.93%; the slope of the linear equation of the cumulative evaporation of period BR reduced by 69.37 mm/a with the reduction rate 7%; the slope of the linear equation of cumulative air temperature increased by 0.97 °C/a with the increase rate of 4.41%. According to the equation of water balance (22), it can be concluded that the change in natural runoff is a function of rainfall, evaporation and air temperature (all climatic factors), whereas the measured runoff is influenced by human activities. It can be obtained from the relationship between cumulative runoff and year in period AR (Fig. 7) when human activities exerted a weak influence on runoff that the sum of the slope change rate of the above climatic factors was equal to that of runoff. Actually, the changes in runoff depth were influenced by both climatic and other factors.

Table 3 Analysis of the contribution rate of climatic factors to runoff change in the upper catchment of the Lu Wu Station in the Qinjiang River Basin
Full size table

According to the results of the formula (26–33), it can be seen that the contribution rate of the factors of the upper catchment of the Lu Wu Station in the Qinjiang River Basin to the runoff reduction is shown in Table 4. According to the results of formula (26–33), compared with period AR, the contribution rate of rainfall, evaporation and air temperature to the runoff reduction in period BR was 12.33%, 29.48% and 18.58%, respectively. Formula (33) shows that the contribution rate of human activities is not a constant value, whereas it should change in the range of lower limit (39.61%) that is under the influence of temperature and upper limit (58.19%) that is not under the influence of Temperature. According to the original method, based on measured runoff series and taking the influence of air temperature into consideration in the base period, the contribution rate of other factors to runoff reduction was 30.16% and the contribution rate of climatic factors was obviously larger. Similarly, compared with period AR, the contribution rate of rainfall reduction, evaporation reduction and air temperature increase to runoff reduction in period CR was 12.80%, 17.69% and 11.15%, respectively. The contribution rate of human activities to the runoff reduction was within the range of 58.35–69.50%.

Table 4 Contribution rate of the factors of the upper catchment of the Lu Wu Station in the Qinjiang River Basin to the runoff reduction
Full size table

Compared with the period from 1986 to 2003, the contribution rate of runoff, evaporation and air temperature to the runoff reduction in the upper catchment of Lu Wu Station in the Qinjiang River Basin decreased in the period from 2004 to 2016, whereas the contribution rate of human activities to the runoff reduction increased that accelerated the runoff reduction.

Although the upper catchment of the Lu Wu Station accounts for a larger proportion of the Qinjiang River Basin, it is mainly located in middle and upper reaches of the watershed upstream, whereas the human activities exert a greater influence in the lower reaches. Table 5 shows the contribution rate of each factor in the Qinjiang River Basin to the runoff reduction.

Table 5 Contribution rate of various factors in the Qinjiang River Basin to runoff reduction
Full size table

Table 5 shows that the contribution rate of human activities in the Qinjiang River Basin to the runoff reduction became larger. In period BR, the contribution rate increased from 39.61–58.19 to 43.78–61.17% and in period CR, the contribution rate increased from 58.35–69.50 to 61.17–72.66%.

To sum up, the runoff reduction in the Qinjiang River Basin is mainly influenced by human activities, followed by evaporation. Rainfall and air temperature are not the main factors to the runoff reduction. The contribution rate of the human activities in the Qinjiang River Basin is larger than that in the upper catchment of the Lu Wu Station. Besides, the human activities dominate the runoff reduction. All these indicate that human life and production activities have caused a great change in the underlying surface and have seriously affected the balance of the water circulation system.

Discussions

Impacts of climatic on runoff change

The double-mass curve of the measured runoff and rainfall, evaporation and temperature in Qinjiang (Fig. 8) shows that there is a strong correlation between runoff, rainfall and evaporation. Figure 2a, b shows that the rainfall has almost synchronized inter-annual abundance changes with the runoff. In addition, Qinjiang located in the south of China contains with many small- and medium-sized rivers flow into the sea. According to the principle of water balance (22), rainfall is the main source of runoff, evaporation is one of the main output sources, and the rainfall has similar multi-timescale characteristics with runoff, which is shown in Fig. 4. It can be seen that the rainfall and evaporation are one of the main factors to control the runoff depth in Qinjiang River Basin.

Fig. 8
figure 8

Double-mass curves for rainfall and runoff (a), evaporation and runoff (b), temperature and runoff (c)

Full size image

Except that the evaporation amount and the measured runoff showed significant mutations in 1983 (Fig. 3), the mutation characteristics of climatic factors were inconsistent with runoff depth. In the natural state, the weak growth of rainfall and the significant decline in evaporation (Table 2) will produce a significant increase in the inter-annual variation of runoff, but it is completely contrary to the actual downward trend of actual measured runoff. In addition, according to Fig. 6, the different sequences of natural and measured runoff depth in the Qinjiang River Basin were stepped upward, indicating that human activities have broken the balance of input and output among rainfall, evaporation and temperature with runoff in the natural water cycle system. This is the main factor that the rainfall showed a weak growth and the measured runoff depth decreased significantly.

Impacts of human activities on runoff change

Among the factors affecting the reduction in runoff in Qinjiang River Basin, human activities were the main one. Based on AR, BR and CR periods are the research periods. Without considering the effect of temperature, the contribution of human activities to runoff reduction was 39.61% during the BR period and 58.35% during the CR period in the basin above the Luwu station. When we consider it, the contribution of human activities to runoff reduction was 58.19% during the BR period and 69.50% during the CR period. The impact of human activities on runoff reduction is limited by temperature, indicating that the improved SCQCR is more scientific than the original method in studying the effects of climate change and human activities on runoff reduction. For the whole Qinjiang River Basin, the contribution of human activities to runoff reduction was 43.78% during the BR period and 61.17% during the CR period without considering the temperature effects, and it was 61.17% during the BR and 72.66% during the CR period when considering the temperature. The results show that the impact of human activities on runoff reduction is more severe in the lower reaches of the Luwu station. The impact of human activities on the runoff of Qinjiang is mainly impacted by the following aspects: water conservancy project construction, land use/vegetation cover and socioeconomic water use.

There are three types of water conservancy projects in Qinjiang River Basin (Table 6): reservoirs, sluices and water pumping. The reservoir project was mainly built before 1980, only one small-I type reservoir and three small-II type reservoirs were built after 1980. The sluice project was also mainly built before 1980, and only one small-II sluice was built after 1990. The water pumping project for municipal water was completed in 1970, and the irrigation pumping was mostly built after 1980. Therefore, the construction of water conservancy projects in Qinjiang River Basin has little impact on runoff which is not the main human activity factor that causes runoff reduction.

Table 6 Completion time of the main water conservancy projects in the Qinjiang River Basin
Full size table

Although the water conservancy project had little impact on runoff, the Qinjiang River Basin is located in the urban expansion zone and is an important source of water for production and living in Qinzhou City. Its land use structure is significantly affected by urbanization. The dynamic changes in land use in Qinjiang River Basin from 1990 to 2010 were mainly due to the continuous increase in cultivated land, construction land (road traffic, industrial and mining, and residential areas) and waters, while forest land and grassland were decreasing (Fig. 9). This type of land use transfer is intended to serve urban development and the quality of life of urban dwellers. In last 40 years, Qinzhou’s the Gross National Product has grown exponentially and the urban population has grown continuously (Fig. 10), and this leap-forward development requires a strong supply of water. According to The Water Resources Bulletin of Qinzhou City (2000–2015), the average annual water consumption of farmland irrigation in Qinjiang River Basin accounts for 73.7% of all water consumption, and the industrial water and urban domestic water consumption accounts for 12.8% and 9.0%, respectively. Other types of water consumption account for low volume ratio relatively. According to Fig. 11, farmland irrigation and urban domestic water consumption in Qinjiang River Basin showed a relatively steady upward trend, while industrial water consumption showed a significant increase. Among them, from 2002 to 2005 was the dry year of the Qinjiang River Basin, resulting in abnormal data.

Fig. 9
figure 9

Land use change in Qinjiang River from 1990 to 2010

Full size image
Fig. 10
figure 10

Gross national product and population change

Full size image
Fig. 11
figure 11

Changes in water consumption of different types in Qinjiang River

Full size image

In summary, the water conservancy project has a low impact on the runoff reduction in Qinjiang River Basin. Under the leaping development of Qinzhou City, the economy has grown exponentially since 1980, and the increasing population causes an increase in the area of cultivated land, irrigation water, industrial water and urban domestic water consumption, which is the most important factor influencing the reduction in human activities in the Qinjiang River runoff.

Summary and conclusion

In the background of the increasing conflicts between supply and demand of water resources in coastal cities of China, this study focused on the factors of rainfall, runoff, evaporation and air temperature of the Qinjiang River which is a typical seagoing river and located in a coastal city of southern China. According to data of monthly rainfall, runoff and evaporation of the Qinjiang River Basin from 1956 to 2013, the change trend of each factor was analyzed based on 3-year running mean method, linear regression method, Mann–Kendall test and R/S analysis method, the variation of each factor was identified based on Mann–Kendall test and the periodic changes were analyzed and the conversion of abundant and rare year was predicted based on Morlet continuous wavelet analysis. In this paper, the influences of climate change and human activities on the runoff of the Qinjiang River were qualitatively assessed from the aspects of change trend, variation and periodicity and the contributions of climate change and human activities to runoff reduction were quantitatively assessed using evaporation difference method and an improved comparative method of the slope changing ratio of cumulative quantity (SCRCQ). The following results were obtained:

  1. 1.

    In the period from 1956 to 2016, the air temperature in the Qinjiang River Basin increased by 1.513 °C and exhibited a significant increasing trend. The rainfall showed a weak increasing trend with 0.808 mm/a, whereas the runoff depth exhibited a significant decreasing trend with − 3.827 mm/a and the evaporation exhibited a significant decreasing trend with − 1.404 mm/a. The Hurst index of rainfall, runoff, evaporation and air temperature exhibited stronger persistence, weaker persistence, strong persistence and strong persistence, respectively. These results indicated that the rainfall and air temperature will keep increasing in the future, whereas the runoff and evaporation will keep decreasing in the future.

  2. 2.

    The runoff had no timescale 34a that was similar to rainfall. However, the runoff and rainfall exhibited good synchronization, whereas they exhibited significant difference to evaporation and air temperature. The changing trend of rainfall and runoff in the future is shown as follows: In the period from 2018 to 2021, there will be a fluctuation of the abundant and rare year; the period from 2022 to 2030 will be period of rare; the period from 2031 to will be period of abundant. The changing trend of evaporation is given as follows: The period from 2019 to 2021 will be period of rare; the period from 2022 to 2030 will have a conversion of the abundant and rare year; the year 2031 will be the year of abundant. The air temperature will be lower in the period from 2018 to 2031 and will be higher in the period from 2031 to 2049. However, it will present an overall significant increasing trend.

  3. 3.

    Rainfall increased weakly and evaporation decreased significantly in Qinjiang River Basin. Rainfall exhibited no significant variation, whereas there were two variation points (1986 and 2003) in the runoff, three variation points (1974, 1986 and 2011) in evaporation and one variation point (1996) in air temperature. These indicated that changes in rainfall, evaporation and air temperature were not the main factors for the runoff reduction. The natural and measured runoff depth difference series exhibited a significant increasing trend (Z = 5.397) with 5.096 mm/a; human activities contributed mainly to the runoff reduction. The contribution of human activities to runoff reduction increased from 43.78 to 61.17% in BR period and increased from 61.17 to 72.66% in CR period. This indicated that the contribution of human activities to runoff reduction increased continuously and it was significantly higher in the lower reaches than that in the upper reaches.

  4. 4.

    In a variety of human activities in the Qinjiang River Basin, the impact of water conservancy construction on runoff reduction is relatively low. The impact of human activities on the reduction in runoff in the Qinjiang River Basin is mainly due to the irrigation, industrial and urban residents’ water use, which is caused by the growth of population and the growth of economic index.