Seasonal changes of sediment fluxes in the Yangtze River: roles of precipitation change, human conservation measures in sub-basins, and large dams

Sediment transport by rivers to the ocean is an important part of global geochemical cycle (Walling & Fang 2003). The total seaward sediment fluxes from global rivers are estimated to be 12.8–19.1 × 10⁹ t per annum (Li et al. 2020), exerting considerable consequences to ecology, environment, and social development. Seasonal variation of sediment load is a key factor controlling riverbed morphology, delta evolution, and floods (Yang et al. 2003). Therefore, changing patterns of riverine sediment load in flood and dry seasons should be analyzed, especially in the context of dramatic climate change and intensive human interferences of recent decades (Walling 2006).

One of the major climatic factors is precipitation change. It indirectly affects seasonal sediment load through runoff (Yang et al. 2014). A close correlation between the changes of precipitation and sediment fluxes has been observed worldwide during the past few decades, such as in the Mekong River Basin (Shrestha et al. 2013), the Thames River Basin (Bussi et al. 2016), and the Yellow River Basin (Wang et al. 2010). Human activities also interfere sediment yielding process of a basin in different seasons (Walling 2006). Major anthropogenic factors include dam interception and soil conservation (Peng et al. 2020). By trapping water in flood seasons and releasing it in dry seasons, a huge amount of sediment is deposited in reservoirs. For example, more than 50% of the sediment flux from the Mississippi River has been reduced since the early 1950s due to damming (Xu & Yang 2015). Sediment loads of the Nile became very close to zero after the Aswan High Dam was built (John & Robert 1983). Soil conservation measures also present different effects in dry and flood seasons due to the seasonal characteristics of land cover (Panagos et al. 2015). At present, human interference to river sediment fluxes are becoming increasingly important globally (Syvitski 2005).

With an area of 1.8 × 10⁶ km² and a length of 6,300 km, the Yangtze River is the largest river in China and the third longest in the world (Zhao et al. 2017). During the past 54 years, decreasing trends of sediment transports have been observed in the upper, middle, and low reaches by 67–90%, particularly after the Three Gorges Reservoir (TGR) became operational in 2003 (Li et al. 2009). Since 1950, more than 50,000 dams have been built within the Yangtze River Basin, including the TGR, which is the largest reservoir in the world (Yang et al. 2011). These reservoirs trap huge amount of sediment inside, reducing sediment load of the downstream reaches (Zhao et al. 2017). For instance, the TGR silted up 172 million tons of sediment per year during 2003–2008, with an average trapping efficiency of 75% (Li et al. 2009); 93% of the sediment flux from the Hanjiang River, a major tributary of the Yangtze River, was trapped by the Danjiangkou Reservoir (Dai & Lu 2014).

Previous research has mainly emphasized the impact of dam construction on changing sediment fluxes of the Yangtze River. Zhang et al. (2006) analyzed the trends of sediment loads from seven hydrology stations on both the main stream and the tributaries of the Yangtze River and pointed out that reservoirs had greater impacts on sediment fluxes than on water discharges. Yang et al. (2011) concluded that sediment reduction of the Yangtze River was mainly caused by the impoundment of more than 50,000 dams in the Yangtze River Basin, especially the TGR. Nevertheless, the role of precipitation changes and other types of human interference (e.g. soil conservation) have been less focused. Zhao et al. (2015) estimated that the climatic and anthropogenic factors contributed 14% and 86%, respectively, to the sediment load reduction of the Yangtze River from 1970 to 2010s. However, considering the seasonal characteristics of sediment fluxes, precipitation, and human activities (Chai et al. 2020a, 2020b), clear division of the contributions by both precipitation change and human interference to sediment flux change in dry and flood seasons of the Yangtze River Basin is needed.

This paper set up a framework to separate the specific influence of precipitation change and human activities of major sub-basins on sediment flux change of the Yangtze River in dry seasons (from May to October) and flood seasons (from November to April). Based on daily observations of water discharge and sediment loads from six gauging stations on the main channel of the Yangtze River during 1961–2014, suppositional sediment series without human interference are reconstructed using regressions between seasonal precipitation and seasonal runoff, and between seasonal runoff and seasonal sediment load. Comparing reconstructed sediment fluxes with actual sediment fluxes gives a better understanding of how precipitation and anthropogenic activities affect sediment fluxes.

The Yangtze River is the largest river in China, with a length of 6,300 km and a basin area of 1.8 × 10⁶ km². Divided by Yichang and Hukou gauging stations, the main channel consists of the upper, middle, and lower reaches, with lengths of 4,504, 955 and 938 km, respectively. Located just upstream of the Yichang station, the Three Gorges Reservoir (TGR) is the largest hydropower works in the world. This study involves six gauging stations on the mainstream, including Zhutuo station, Cuntan station, and Yichang station in the upper reach, Luoshan station and Hankou station in the middle reach, and Datong station in the lower reach (Figure 1(a)).

Therefore, the whole Yangtze River Basin is divided into six subregions by those stations, i.e., the subregion upstream of Zhutuo station, subregion between Zhutuo and Cuntan stations, subregion between Cuntan and Yichang stations, subregion between Yichang and Luoshan stations, subregion between Luoshan and Hankou stations, and subregion between Hankou and Datong stations (Figure 1(b)).

Daily precipitation of the 145 meteorological stations (Figure 1(c)) from 1961 to 2014 was provided by the Resource and Environment Data Cloud Platform (http://www.resdc.cn/UserReg.aspx). Daily water discharge and sediment fluxes of the six gauging stations from 1961 to 2014 (Table 1) and high-resolution underwater topographic data of the TGR (from Qingxichang to the dam) during 2003–2014 were provided by the Changjiang Water Resources Commission of the Ministry of Water Resources.

Given significance level to be α = 0.05, the confidence level is 1 − α (i.e., 95%) and the critical value Z_1−α/2 = ±1.96 can be obtained based on the standardized normal distribution. If Z ≥ |Z_1−α/2|, the null hypothesis is not acceptable, i.e., the sequence has an obvious increasing (with a positive Z) or decreasing (with a negative Z) trend.

From 1961 to 2014, the precipitation and runoff in the Yangtze River have changed with different directions and extents. To analyze the trends of these series, Z in Equation (6) is calculated. It shows that precipitation presented upward trends in both dry and flood seasons, especially at the Zhutuo and Cuntan stations in the upper reach of the Yangtze River, where the increasing trends were significant at 95% level (Figure 5). Figure 6 demonstrates that the runoff at the six hydrological gauging stations increased in dry seasons, but decreased in flood seasons, implying that the seasonal runoff has homogenized (i.e., more evenly distributed between dry and flood seasons) over most of the Yangtze River. However, the trends were only significant (95% level) at Cuntan, Yichang, and Luoshan in flood seasons, and at Hankou in dry seasons. Similarly, the trends of sediment fluxes during the impact period from 1981 to 2014 are analyzed. In Figure 7, sediment fluxes had very sharp downward trends at all the six stations in both seasons, except at the Zhutuo station in dry seasons. Table 2 further gives a declining rate of seasonal sediment fluxes in the impact period (1981–2014) compared with those in the natural period (1961–1980). The average annual declining rate of sediment fluxes in dry seasons varied from 6.8% at the Zhutuo station to 74.6% at the Yichang station, while that in flood seasons varied from 14.6% at the Zhutuo station to 38.7% at the Datong station.

Regression between seasonal precipitation and runoff (Equation (13)) at the six gauging stations in Figure 8 shows that all the linear relationships were significant at the level of p < 0.05, except one (Zhutuo station in dry seasons, p < 0.2). Likewise, all the linear regressions between seasonal runoff and seasonal sediment fluxes (Equation (14)) were significant at the level of p < 0.05, except that at the Hankou station in flood seasons (p < 0.07, see Figure 9). This implies that the regressions are generally reliable to reconstruct sediment series without human interference. Figures 6 and 7 show the regression equations and the linear coefficients demonstrating the direction and degree that precipitation affects seasonal runoff and sediment fluxes. The linear coefficients of all the two types of regressions are all positive, indicating that increasing precipitation will lead to an increasing sediment flux. It also shows that linear coefficients of one gauging station are generally larger in flood seasons than in dry seasons (except at the Hankou station), implying that precipitation in flood seasons could exert more crucial impact on runoff and sediment flux.

By comparing the average level of the reconstructed sediment series without human interference during the impact period and the mean value of the observed sediment series in the natural period (see Figure 3 and Figure 10), the impact of precipitation change is separated at the six gauging stations (Equation (16)). The results show that precipitation change generally leads to increase in sediment fluxes, which is opposite to the overall decreasing trends in seasonal sediment fluxes. The increment in the dry seasons varies from 0.72 to 4.22 t/km², which is 3.5–17.8% of the average level in the natural period (Table 3). In flood seasons, the increment is more considerable, varying from 4.95 to 73.32 t/km², which is 28–56 times as much as that in dry seasons. According to Borrelli et al. (2020), soil erosion is more closely related to the intensity of precipitation than the total amount. Since precipitation intensity can be reflected by the concentration of precipitation (Richardson et al. 1983), sediment yield generally increased more rapidly in flood seasons when over 70% of the annual precipitation was concentrated during 1961–2014. In the upper reach of the Yangtze River, percentages of precipitation in flood seasons (∼88–90%) were higher than in the middle and lower reaches (∼75–85%), which implies that soil erosion was more serious in the upper reach.

By comparing the mean values of the observed and reconstructed sediment series in the impact period (see Figures 3 and 10), the impact of human interference on seasonal sediment fluxes is separated (Equation (17)). It shows that the decreasing effect of dam construction and conservation measures by human beings overwhelmed the increasing effect of precipitation change on sediment series and was recognized as the primary factor leading to the sharp drop of seasonal sediment fluxes. From 1981 to 2014, in dry seasons, reduction of sediment yields varied from 1.25 t/km² (Zhutuo station) to 28.0 t/km² (Luoshan station); in flood seasons, sediment yields decreased by 98.9–215.7 t/km² (Table 3), and the reduction reached summit at the Yichang station (215.65 t/km²).

The contribution of human interference in the six subregions (ΔS_H-sub) can be further divided by subtracting ΔS_H (see Equation (17) and Figure 3) at the upper controlling station from the adjacent lower one. Table 3 shows the proportions of sediment flux reduction by the TGR and all these subregions. Before the TGR became operational, the subregion between Cuntan and Yichang stations (mainly locating the Wujiang sub-basin) took up most (46.8%) of the reduction of seaward sediment fluxes in dry seasons, whereas the subregion between Zhutuo and Cuntan stations (overlapping the Jialingjiang sub-basin) contributed more than half (52.5%) in flood seasons. After the impoundment of the TGR, the subregion between Yichang and Luoshan stations (containing the Dongting Lake sub-basin) was the largest contributor in dry seasons, while the subregion upstream of Zhutuo station led in the seaward sediment reduction (35.9%). It should be noted that the subregion between Yichang and Luoshan (mainly locating the Dongting Lake basin) gradually changed from depositional status to erosional status (Yang et al. 2014), which might lead to the underestimation of human conservation measures in the basin. According to Zhu et al. (2014), significant channel erosion and excessive sand mining occurred in the Poyang Lake (mainly overlapping the subregion between Hankou and Datong), which was not included in the ΔS_H of this subregion, especially in the dry seasons.

Since the six subregions are all located over the mainstream, the hydrometeorology regime in the upper subregions might exert an influence on the sediment fluxes of the lower subregions, leading to uncertainties of estimating the contributions by human interference in the subregions. Considering that such influence is mainly reflected by erosion and deposition of the riverbed of the main channel, we summarized the annual riverbed erosion/deposition of different reaches in Table 4, which shows that the riverbed change is 8.2–35.2% of the total sediment yield of the corresponding subregion. However, the method of comparing the sediment fluxes from two adjacent stations does not distinguish the mainstream riverbed erosion/deposition from the total sediment change due to human activities in the subregions, i.e., erosion/deposition on the mainstream riverbed is mistakenly regarded as occurring in the sub-basin due to human activities by this method. Therefore, overestimation/underestimation of human conservation effect in the corresponding subregions in Table 4 is expected.

The Three Gorges Dam is the largest dam in the world with a storage capacity of 39.3 × 10⁹m³ (Zhang et al. 2006). Since the impoundment of TGR on 1 June 2003 (Chai et al. 2020a), the TGR contributed 3.2 and 23.9% to the reduction of seaward sediment fluxes in dry and flood seasons (Table 3), respectively, which is lower than expected. Figure 11 further gives the yearly sedimentation of the TGR from 2003 to 2014, and the contributions in dry and flood seasons. It shows that sediment deposition in flood seasons dominates the total annual sedimentation (116.5 Mt/yr in flood seasons vs. 1.7 Mt/yr in dry seasons). It is interesting that sedimentation is not proportional to sediment flux. The amount of sedimentation depends on the combined conditions of water and sediment inflows into the reservoir. Given the same sediment flux (sediment concentration multiplied by water discharge), sedimentation may be different if sediment concentration and water discharge are not completely the same. Since the TGR receives water and sediment from several large sub-basins which are different in water discharge, sediment concentration, and flood processes, water and sediment conditions of the inflows to the TGR may vary from time to time, resulting in the disproportional sedimentation to sediment flux. It is also found that changes in runoff and sediment fluxes affected the distribution of sediment deposition in the TGR. As the reduction of sediment fluxes, sedimentation in the TGR is greatly reduced. Meanwhile, the fluctuating backwater area was eroded, and the perennial backwater area was heavily deposited, especially the area away from the dam, according to the data provided by the Changjiang Water Resources Commission.

In addition to the TGR, more than 52,000 dams with a total capacity of 360 × 10⁹m³ have been built in the Yangtze River Basin since 1950, particularly located in the region upstream of the Zhutuo station which contributed 2.4 and 30.9% to the reduction of seaward sediment fluxes in dry and flood seasons, respectively (Table 3). During 1950–2015, there were 14,732 reservoirs in the upper reaches of the Yangtze River, with a total storage capacity of 167.3 × 10⁹m³. During 1956–2015, total sedimentation of the reservoirs in the upper region of the Yangtze River Basin was up to 6.925 × 10⁹ m³, and 77% of the reduction was contributed by large dams, according to the Changjiang Water Resources Commission. In the Jinshajiang reach, which is the upmost reach of the mainstream of the Yangtze River, the average annual sediment trapping of cascade reservoirs in the middle reach was up to 44 × 10⁶t in 2011–2016. In the Jialingjiang Basin (almost covers in the subregion between Zhutuo and Cuntan in Table 3), a major tributary in the upstream region of the Yangtze River Basin and the largest contributor (see Table 3) to the reduction of seaward sediment fluxes, a series of large dams have been constructed, such as the Bikou reservoir (521 × 10⁶m³, became operational in 1977), the Dongxiguan reservoir (165 × 10⁶m³, became operational in 1995), and the Baozhusi reservoir (2.55 × 10⁹m³, became operational in 1996). Besides, the impoundment of the Ertan Hydropower Station in the Yalongjiang River Basin and the Pubugou hydropower station in the Daduhe River began in 2000 and 2009, respectively. Similar to the TGR, sediment trapped in flood seasons by other reservoirs is more than that in dry seasons. That is because reservoirs usually impound water and sediment during the second half of the flood seasons (from August to November) and release them during the dry seasons (from January to February). According to Liu et al. (2019), the sediment releasing ratio (the ratio of the released sediment out of the reservoir to the sediment inflow of the same period) of the TGR increased from 17% in flood seasons to 25% in dry seasons, implying that sediment is mainly released in dry seasons.

The WSCP has been implemented in the Yangtze River Basin since 1988. By 2015, the area of water and soil conservation has reached 400 × 10³ km², leading to average annual sediment reduction in the Jialingjiang River Basin of 20.8–28.3 Mt/yr, which accounted for 16.9% of the total sediment reduction in this area. After the huge flood of the Yangtze River Basin in 1998, the Grain for Green Project was started. From then on, soil and water conservation projects gradually extended to the whole Yangtze River Basin. According to the Changjiang Water Resources Commission (CWRC 2015), soil erosion area in the Yangtze River Basin decreased to 384,600 km² due to effective conservation measures. With an average decreasing effect of 234 Mt/yr, soil and water conservation measures in the Yangtze River Basin accounted for 22.6% of the total sediment reduction 1989 to 2015 (Peng et al. 2020).

Extensive sand mining on the Yangtze Riverbed is another reason for the decrease in sediment fluxes (Yang et al. 2011). As the development of urbanization and construction industry, the amount of sand mining in the Yangtze River Basin has increased greatly. Sand mining in the middle and lower reaches of the Yangtze River reached 86 × 10⁶m³ during 2004–2011. After the implementation of the Sand mining Management Regulations by the CWRC, the total amount of sand mining was reduced to 41.5 Mt in 2015. However, the actual sand extraction might exceed 41.5 Mt/yr due to illegal mining (Wang et al. 2019). In the recent 2 years, the amount of sediment dredging of the Yangtze River reached 27.9 × 10⁶ m³ (2018) and 32.2 × 10⁶ m³ (2019), according to the Bulletin of Hydrological and Sediment of the Yangtze River (CWRC 2016).

This study aims to analyze the trend of seasonal sediment fluxes in the Yangtze River and the roles of major climatic and anthropogenic factors. Downward trends of sediment fluxes in both dry and flood seasons were found in upper, middle, and lower reaches of the Yangtze River during 1981–2014, while upward trends of seasonal precipitation and runoff were tested. Thus, the effect of increasing sediment yields by the precipitation change was found in both dry and flood seasons by 0.72–4.22 t/km² (i.e., increased by 3.5–17.8%) and 4.95–73.32 t/km² (i.e., increased by 1.9–25.5%). Due to human interference, sediment yield reduction was up to 0.07–20.74 t/km² (i.e., reduced by 0.9–64.6%) and 27.47–85.35 t/km² (i.e., reduced by 6.5–23.7%) in dry and flood seasons during 1980–2002, and further reduced 3.61–41.31 t/km² (i.e., decreased by 46.0–102.9%) and 175.63–471.52 t/km² (i.e., decreased by 59.6–126.2%) after the TGR became operational in 2003. The contributions to the reduction of seaward sediment fluxes were most notable in the subregion between Yichang and Luoshan in dry seasons, and in the subregion between Zhutuo and Cuntan in flood seasons. By comparison, the contribution of the TGR was only 3.2% in dry seasons and 23.9% in flood seasons, which is below expectation.

This work was supported by the National Key R&D Program of China (grant no. 2016YFA0600901) and the Youth Project of National Natural Science Foundation of China (grant no. 51609015). The hydrological data for this study were provided by the Bureau of Hydrology of the Changjiang Water Resources Commission and Yangtze River Waterway Bureau. Shitian Xu, Yuanfang Chai, and Yao, Yue conceived the study and wrote the draft of the manuscript. Xia Yan and Xiaofeng Zhang revised the manuscript and contributed valuable comments.

All relevant data are included in the paper or its Supplementary Information.