Characteristics and attribution analysis of runoff and sediment evolution in the Wei River mainstream, China

River flooding and drought are common situations, and river management has been a great endeavor since ancient times in order to fight against nature (Song & Liu 2020; Brierley & Fryirs 2022). From simple containment in ancient times to concrete dams today, the impact of human activities on rivers has become increasingly prominent with the rapid development of society and technology (Liu et al. 2021a). Rivers evolve over a long period of time to develop unique regular characteristics, but due to strong disturbances from external factors, river conditions may change qualitatively at some point in time (Ye et al. 2020), such as runoff, sediment content, periodicity, and water quality conditions (Xu et al. 2011). According to statistics, the number of rivers with altered runoff worldwide is 24% of the total, and the number of rivers with altered sediment load is even higher, at around 40%. Few rivers in the world are in their natural state due to human disturbances (Li et al. 2020a). However, these river dynamics are important for understanding long-term water quality and ecological changes in the basin, so runoff and sediment changes have become an important topic of research (Mikhailov 2010; Li et al. 2020b).

Soil erosion is now one of the major environmental problems in the world, with the Yellow River basin bearing the brunt of the problem (Shi et al. 2010). The Loess Plateau in the basin has suffered serious ecological damage due to low vegetation cover and a high frequency of heavy rainfall coverage, and the sediment content in the river is high. These factors have caused serious sediment accumulation in the middle and lower reaches of the river, and the phenomenon of above-ground rivers is common, which seriously hinders the sustainable development of society (Schnitzer et al. 2013). In recent decades, human activities have greatly altered the underlying surface of the Loess Plateau, and the climate has changed significantly due to increased global warming from carbon dioxide, both of which have led to significant changes in the sedimentation of runoff from Loess Plateau rivers into the Yellow River (Tian et al. 2019). The Wei River is the first major tributary of the Yellow River, flowing through the Loess Plateau and converging in the middle reaches of the Yellow River, delivering 389 million tons of sediment and 7.45 billion m³ of water to the Yellow River each year, accounting for approximately 39.87 and 26.03% of the total sediment and water volume of the Yellow River, making it a relatively representative tributary (Chang et al. 2016). The Wei River is also an important source of water for humans in Shaanxi Province, supporting the operation of the grain-producing and industrial and commercial areas of the basin (Zhao et al. 2019). The study and analysis of runoff–sediment changes and their driving mechanisms in the changing environment of the Wei River basin (WRB) can help to analyze and predict the erosion and sediment production mechanisms and sediment load patterns in the Yellow River basin, which has important theoretical and practical significance for the prevention and control of soil erosion and also has significant strategic importance for the economic development of this basin (Deng et al. 2018; Zhang et al. 2022b).

The Wei River is located in the southeastern part of the Loess Plateau, which is an arid and semiarid region with a continental monsoon climate, so water resources are affected by the seasons. Droughts occur mainly in spring in the middle and upper reaches, are prone to occur in summer and autumn in the lower reaches, and occur mainly in winter in the tributaries, and there is a tendency for the severe development of droughts (Fan et al. 2022). Some studies have shown that this factor has a significant impact on runoff–sediment changes (Zhao et al. 2017). Secondly, the vegetation distribution in the watershed is discontinuous, and the ecological environment is fragile and extremely sensitive to human activities. People's excessive land reclamation and development have caused massive land erosion (Han et al. 2021; Zhang et al. 2022a). The sediment-producing areas are mainly located in the Jing and Beiluo River basins and the upper basin of the Wei River mainstream, where sediment accumulates in the lower reaches and is washed into the Yellow River, transporting large amounts of sediment to the Yellow River (Song et al. 2010). Huang et al. (2016) analyzed the contribution rate of climate and human activities to runoff reduction in different periods and found that human activities are the main driving factors of runoff reduction. Tian et al. (2022) analyzed the sediment load data of 15 hydrological stations in WRB and found that the sediment load in the basin decreased significantly, and rainfall, vegetation, and runoff all affected the sediment change. Most studies focus on the analysis of the variation trend and attribution of runoff and sediment load alone, but do not specifically study the internal relationship and correlation between runoff and sediment. This paper not only studies the interannual variation of runoff and sediment load in the mainstream of the Wei River but also studies the resonance cycle and interaction effect of runoff and sediment load based on cross-wavelet analysis and cumulative curve, and further expounds the relationship between sediment load and runoff. In addition, this paper also discusses in detail the impact of climate change and shadow activities on runoff and sediment changes, and emphatically analyzes the impact of terraced fields on sediment load, which provides more theoretical basis and ideas for ecological management.

Huaxian hydrological station, located in the lower reaches of the Wei River, is the control section where the Wei River flows into the Yellow River. It is more representative of runoff and sediment load and can well reflect the river state before the Wei River flows into the Yellow River. Therefore, the runoff and sediment load data from the Huaxian hydrological station from 1971a to 2020a were used to analyze the evolution of runoff and sediment in the Wei River and the interaction between the two, where the runoff data are measured runoff. Runoff and sediment load data for the Wei River Huaxian station from 1971a to 2020a were obtained from the China River Sediment Bulletin, and the precipitation data were obtained from the website of the China Meteorological Science Data Service Center (http://www.nmic.cn). Land use data come from the Resources and Environmental Science and Data Center of Chinese Academy of Sciences (http://www.resdc.cn).

The Mann–Kendall (M–K) method is a common method for analyzing serial trends and testing for mutations. Let the time series data be X₁, X₂, …, X_n, and construct the test statistic Z after defining the statistic S. At a given significance level, if |Z| > Z_1−α/2, the upward or downward trend of the series is significant at that significance level. When Z > 0, it indicates an upward trend, and vice versa, it indicates a downward trend.

The series of data was standardized according to the formula, and the calculated standard variation UF_k formed the UF curve. The UB_k and UB curves are obtained by the same method for its inverse series. The intersection of the curves UB and UF is the abrupt change points of the sequence. The specific methods and formulas for the calculation have been described in detail in other literature and will not be repeated here (Gocic & Trajkovic 2013; Ay & Kisi 2017). When UF> 0, it means that the sequence has an upward trend, and when UF < 0, the sequence has a downward trend.

The anthropogenic contribution C_H (%) is C_H = 100 − C_P. The contribution of the two factors to sediment load can be similarly found by replacing runoff with sediment load following the above steps (Kong et al. 2018).

The interannual variation graphs of runoff and sediment load at the Huaxian station are highly fluctuating and can only be analyzed qualitatively and with large errors by observation alone. In order to quantify the assessment, the M–K test was used to evaluate the trend of runoff and sediment load. The Z values of the test statistics for runoff and sediment load were calculated to be −0.44 and −4.99, and taking the significance level α = 0.05, then Z_1−α/2 = 1.96, |Z_runoff| = 0.44 < 1.96, |Z_sediment| = 4.99 > 1.96. This indicates that there is no significant downward trend in multiyear runoff and a significant downward trend in multiyear sediment load at the 95% confidence level for the Huaxian station, which is basically consistent with the results of the visual analysis above.

To determine the accuracy of the M–K method calculation results, the mean difference t-test was used to test whether the mutation year was reasonable. The significance level α is generally taken as 0.01 or 0.05, and to ensure the rigor of the test results, α = 0.01, which means t_α = 2.704. The calculated mutation index in annual runoff is 0.345, and the statistic t = 2.439 < 2.704, which indicates that at the significance level α = 0.01, no significant mutation occurred in runoff in 1991, so we consider that 1991 did not represent a significant abrupt change. Therefore, we believe that there was no abrupt change in 1991. The mutation index in annual sediment load is 1.054, with the statistic t = 4.824 > 2.704, which indicates that at the significance level α = 0.01, there is a mutation in sediment load in 2006.

The periodicity of sediment load in Figure 5 is relatively chaotic and less stable and is less periodic compared with runoff. With the 1990s as the dividing line, the periodicity changes greatly before and after. Before the 1990s, the periodicity of sediment load was relatively strong, with three main periodicities of 4–5 years, 7–8 years, and 18–24 years. However, after the 1990s, the periodicity gradually changed, and the 18–24 years periodicity gradually changed to 20–23 years, and the short periodicities of 4–5 years and 7–8 years gradually weakened and disappeared, as well as the emergence of a new periodicity of 12–14 years. The periodicity generally becomes longer and weaker in this period. From the wavelet variance diagram, it can be seen that the more obvious periodicities of sediment load are 3 years, 5 years, and 22 years, among which 5 years has the highest peak variance and is the dominant periodicity of sediment. Both runoff and sediment load changed around the 1990s, mainly because many water conservation projects were built in the Wei River in the 1970s and 1980s, and the runoff and sediment load tended to stabilize through runoff and sediment transfer, so the short periodicities of both began to disappear in the late 1980s, leaving only a few longer periodicities. Sediment load is more influenced by human activities, and the periodicity is relatively weak.

The power spectrum in Figure 6 shows a weak correlation between runoff and sediment in general, with few resonant periodicities. The two resonance periods that passed the test are all before 1995. The period 1971–1976 has a significant 0–2 years resonance periodicity between runoff and sediment, and since this time periodicity is partly in the region affected by edge effects, only the period 1974–1976 is considered, with a phase angle of 0°, indicating a positive correlation between runoff and sediment. The period 1980–1990 has a significant resonance periodicity of 3–5 years, with a phase angle of −45°, and the runoff is delayed by 1/8 period compared with the sediment load, which is roughly positively correlated. The resonance periodicity in the coalescence spectrum is mainly concentrated in the periodicities of 0–2 years from 1996 to 2005, 3–5 years from 2005 to 2014, and 9–12 years from 2005 to 2019. Since the 9–12a cycle is outside the influence cone, this cycle is not considered. Among them, the ones that passed the test are the 0–1.5 years periodicity from 2002 to 2005 with a phase angle of 0°, which is positively correlated, and the 3–5 years periodicity from 2005 to 2014 with a phase angle of 90°, where the runoff is 1/4 period ahead of the sediment load. In summary, the runoff–sediment resonance periodicity in the Wei River is small, and overall the interaction between the two is small and the consistency is weak. However, the resonance effect of runoff–sediment is better in some time periods, mainly concentrated in the 1980s and 2000s. Higher rainfall in the 1980s, and 1983 and 1984 were still the highest rainfall years, led to increased runoff and an impact on river siltation and a consequent increase in river sediment load. The precipitation periodicity of 3–5 years coincides with the greater significance of the 3–5 years periodicity of runoff and sediment between 1980 and 1990. The lower reaches of the Wei River experienced the ‘03·8’, ‘03·10’, ‘05·10’, and ‘11·09’ major floods, which were in a scouring state, coinciding with the significant periodicity of runoff and sediment during 2002–2014.

The amount of sediment load mainly depends on the soil condition and the construction of water conservancy projects in the basin, and precipitation can only affect it indirectly through runoff, so the linear correlation between sediment load and precipitation is very weak and not significant in most periods. Only in 1980–1994, the correlation between precipitation and sediment load is relatively high and p < 0.05, indicating that there is a significant correlation between precipitation and sediment during this period. This is mainly due to the high runoff volume in that period, which has an impact on river sediment and leads to higher sediment content in the river as it is washed downstream along with the river. Both runoff and sediment load were best correlated with precipitation in the time period 1980–1994. During this time period, precipitation has an important relationship with the change of runoff and sediment, giving some consistency in runoff and sediment load. Although the correlation between precipitation and sediment load is poor, some studies show that rainfall intensity has a great influence on sediment load. The Loess Plateau has loose soil, weak anti-erosion ability and is easily washed by rain. Therefore, the greater the rainfall intensity, the more serious the soil erosion, which leads to the increase of river sediment load. The WRB is located in the Loess Plateau, and its tributary Jing River is located in the sediment-producing area. Rainfall intensity will have a great impact on the sediment load of the Wei River. However, when the rainfall intensity exceeds a certain threshold, the sediment load will not increase with the increase of rainfall intensity. The vegetation coverage rate will affect the threshold, and the higher the vegetation coverage rate, the higher the threshold (Liu et al. 2020).

The Pearson correlation coefficient method was used to judge the impact of land types on runoff and sediment load, and their correlation was further analyzed. The area of unused land is less than 0.4%, which has negligible influence, and the error in the calculation is large, so it is not substituted into the calculation. Land use data for 1980, 1990, 1995, 2000, 2005, 2010, 2015, and 2020 and runoff and sediment load for corresponding years were used to calculate correlation coefficients. The calculation results are shown in Table 2. The correlation is strongest when the absolute value of the correlation coefficient tends to 1, and weakest when it tends to 0. The correlation between grassland and runoff–sediment is extremely small and the effect is negligible. As a whole, the correlation between sediment load and land type is above 0.5 in absolute value and is significant except for grassland and water body. Hence, the influence of sediment load from subsurface changes is greater, which is consistent with the above contribution. The correlation between runoff and all land use types is not significant. The underlying surface does not directly determine the runoff volume but only indirectly influences the runoff variation from the side. The correlation coefficient of forest land is the highest, which is 0.618, because forest land can promote the conversion of groundwater to rivers and has a recharge effect on runoff (Li et al. 2018). The ground permeability of built-up land is poor, resulting in less rainwater infiltration and increased ground flow production capacity, so it has a positive correlation with runoff, with a correlation coefficient of 0.345, which has a weak effect on runoff. The large reduction in cultivated land is converted to forested land and built-up land that will recharge runoff, so cultivated land has an indirect negative correlation with runoff, with a correlation coefficient of −0.493.

There is a significant negative correlation between sediment load and forest and building land. Forests reduce rainfall intensity through their canopy, thus slowing down the impact of rainfall on the ground, and the root system also consolidates the soil (Han et al. 2021). Buildup sites generally cover the land with construction materials such as cement, which prevents the erosion of the land by foreign objects and has the highest correlation with the change of sediment load. Both of the above reduce the amount of sediment load by the river from the source, with correlation coefficients of −0.815 and −0.860. The correlation between cultivated land and sediment load is positive, mainly because the root system of crops in cultivated land is shallow, which cannot consolidate the soil well, and it is easy to cause soil erosion. In addition, it is indirectly affected by the conversion of forest land, water area, and construction land, and the correlation coefficient is 0.810.

It is worth mentioning that terraced fields have a great impact on sediment load. In the document on soil and water conservation in the Yellow River Basin issued by the government, it is proposed that northern Shaanxi and other places should vigorously carry out terraced fields to reduce soil erosion. From 1980 to 2020, the terraced fields in WRB increased from 192 to 315 km², and the correlation coefficient with sediment load was −0.683, showing a significant negative correlation. Since the 1970s, the construction of terraced fields has reduced the sediment deposition in the upper and middle reaches of the Wei River by 101 million tons and 66 million tons (Liu et al. 2021b). From 2000 to 2015, the control efficiency of terraced fields on soil erosion increased from 0.263 to 0.365 (Wang et al. 2019). Compared with terraced fields on runoff, terraced fields have a greater impact on sediment load and remain stable all the time. When the land is seriously degraded or in the initial stage of restoration, terraced fields have a more significant impact on soil and water conservation (Wang & Yao 2019). Although terraced fields are a kind of cultivated land, their overall influence is opposite to that of cultivated land. This is because the stepped construction style can effectively slow down the velocity and kinetic energy of slope runoff by shortening the slope length, and the sediment can be intercepted to a certain extent (Chen & Yang 2011). However, terraced fields occupy a small area in cultivated land, accounting for only about 0.5%, so there is a positive correlation between cultivated land and sediment load as a whole.

Hydraulic projects are the main means by which people regulate rivers and have a direct impact on runoff–sediment changes. A large number of hydraulic projects have been built and put into operation since the 1970s, and as of 2015, 1,635 diversion projects and more than 700 large check dams have been built in the WRB, most of which were constructed after 2000 (Chang et al. 2016). The Dongfanghong Irrigation Project, the Jiaokou Pumping Irrigation Project, the Baojixia Main Canal, and the expanded Luohui Canal that operated in the 1970s, together with some small- and medium-sized irrigation projects, brought the total irrigated area to 5,900 km². The significant water withdrawal from the river has resulted in a 34.5 m³/s average annual runoff decrease, or 28% of the overall runoff reduction. In addition to transferring and storing water, Fengjiashan Reservoir and Shitouhe Reservoir in the 1980s also intercepted a large amount of sediment and reduced sediment by an average of 6.78 million tons per year in the 1980s compared with the 1970s. Reservoir storage increases the water area, resulting in greater evaporation, which also has an impact on runoff. In addition to this, studies have shown that the Sanmenxia Reservoir, although not located in the WRB, has had a serious impact on the runoff and sedimentation of the Wei River (Zhang et al. 2020). The Sanmenxia Reservoir raised the water level by more than 30 m due to power generation, slowing down the flow of the Wei River as it merges into the Yellow River, resulting in the lower reaches of the Wei River becoming seriously silted up, which raised the elevation of Tongguan by 4.7 m in only two years of operation. The allowable flow rate decreased from 5,500 m³/s before the construction of the reservoir to 1,000 m³/s after 1990, which was one of the main reasons for the major flood disaster of the Wei River in 2003 (Liu et al. 2022). To alleviate sedimentation in the lower reaches of the Wei River, the construction of the Dongzhuang Reservoir began in 2018 in the lower reaches of the Jing River, which is expected to intercept 750 million tons of sediment from tributaries to alleviate the burden on the lower reaches of the mainstream and is a key water conservation project in Shaanxi in recent years.

After the reform and opening up, the economy of WRB has increased rapidly, and with the continuous growth of population, the industrial water consumption, agricultural water consumption, and domestic water consumption have increased dramatically. Human water consumption becomes an important factor in the reduction of runoff in the Wei River, and some sediment will be taken away when taking water, which indirectly affects sediment load. According to statistics, the average water consumption of WRB was 2.793 billion m³ before the 1990s, and the average water consumption increased to 4.263 billion m³ in the 1990s, with an increase of 52.6%. The increase of water consumption is mainly manifested in the over-exploitation of groundwater, which affects the surface runoff. The increase of water consumption leads to the decrease of 864 million m³ of the Yellow River inflow in WRB, accounting for 21% of the actual decrease of the Yellow River inflow in the 1990s. Water consumption for soil and water conservation increased by 184 million m³ in the 1990s, accounting for 4% of the actual decrease of water inflow into the Yellow River in the 1990s (Xu et al. 2023).

• Runoff and sediment trends and abrupt change years in the Wei River mainstream basin are significantly different. There is no obvious trend change and abrupt change in runoff. While the sediment delivery shows a significant decreasing trend, the abrupt change in sediment delivery occurs around 2006. The runoff volume is dominated by long periodicity, with the main periodicities of 10a, 12–13a, and 29a, and 12–13a is the dominant periodicity. The periodicity of sediment load does not have obvious consistency with the runoff periodicities, mostly with shorter periodicity. The main periodicities are 3a, 5a, and 22a, with 5a as the dominant periodicity.

• In the high-energy region, there is a significant resonance periodicity of 3–5a in 1980–1990, which is roughly positively correlated. In the low-energy region, the main resonance periodicities are 0–1.5a in 2002–2005, which are positively correlated, and 3–5a in 2005–2014. The runoff–sediment sequence can be divided into four stages: 1971–1979, 1980–1994, 1995–2003, and 2004–2020, and the runoff–sediment relationship changes significantly at the time points of 1979, 1994, and 2003.

• The contribution of precipitation and human activities to runoff is 21.2 and 78.8%, and the contribution to sediment load is 4.7 and 95.3%, so human activities are the main factors of runoff and sediment change.

W.G. contributed to funding acquisition; project administration; resources; investigation; supervision. Y.S. contributed to conceptualization; data curation; formal analysis; investigation; methodology; resources; software; validation; visualization; writing – original draft, writing – review and editing. J.H. contributed to investigation; formal analysis; methodology; validation; visualization. W.W. contributed to visualization; investigation; formal analysis. H.W. contributed to funding acquisition; project administration.

This study was supported by the National Natural Science Fund of China (51779094), the 2016 Henan University Science and Technology Innovation Talent Support Plan (16HASTIT024), the Guizhou Provincial Water Resources Department 2020 Water Conservancy Science and Technology Project (KT202008), and the Basic Research Project of Key Scientific Research Projects of Colleges and Universities of Henan Province (23ZX012).

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.