Quantifying the effects of channel change on the discharge diversion of Jingjiang Three Outlets after the operation of the Three Gorges Dam

More than 47,000 large dams and 800,000 small dams have been constructed in river systems worldwide in the past few decades (Haghighi et al. 2014). Dam impoundment led to large amounts of sediment retention in reservoirs (Syvitski et al. 2005; Duan et al. 2015). The channels downstream of the dams also underwent erosion and morphological changes (Dai & Liu 2013; Csiki & Rhoads 2014). Channel changes can influence navigation management, ecosystems, and flood control and prevention. In recent years, the channel changes downstream of the dams and the homologous environmental effects have attracted extensive attention from the public and the science community (Hudson et al. 2008; Tealdi et al. 2011; Segura-Beltrán & Sanchis-Ibor 2013; Gao et al. 2015a).

Water level–discharge curves (rating curves) play an essential role in hydrology, i.e., in water resource management of river basins (Petaccia & Fenocchi 2015), as well as in flood risk and control assessment (Bormann et al. 2011). Rating curves are highly sensitive to channel changes (Jalbert et al. 2011), such that the rating curve shifts down or up, respectively, once the channel erodes or deposits. Long-term variations of the water level–discharge relationships can reflect the channel morphology changes (Zhang et al. 2015). Therefore, rating curves can be used to detect the channel changes and homologous environmental effects in a river. Numerous approaches for studying rating curves can be found in the literature. Regression methods based on data-driven, non-parametric, and non-linear models are popularly used for rating curve construction and water discharge forecasting (Duan et al. 2013; Valipour et al. 2013; Wolfs & Willems 2014; Valipour 2015a, 2016). Regression models can provide accurate water discharge forecasting results. However, information on the channel morphology changes cannot be obtained. In recent years, the most commonly used method of the rating curve as a power-law function has been applied to detect the channel morphology changes (Wang et al. 2013; Zhang et al. 2015).

The Yangtze River is the largest river in the Asian Monsoon region and suffers from a combination of highly intensified human activities (Chen et al. 2014) and natural flow regime variability (Zhang et al. 2007). The occurrence of severe floods and droughts in the middle–lower Yangtze River has accelerated at an increasing rate, during the past few decades (Nakayama & Shankman 2013; Gao et al. 2014). The Three Gorges Dam (TGD) on the Yangtze River is the largest hydroelectric project in the world. The TGD impoundment in 2003 intercepted 65–85% of upstream sediments (Yang et al. 2014), thereby leading to downstream channel erosion, which reached 979 million m³ in 2002–2010 (Lu et al. 2011). Long-distance channel incision occurred downstream (Dai & Liu 2013). The channel incision lessened the water level of the main river with the same discharge (Wang et al. 2013), and also directly altered complex river–lake relationships (e.g., Gao et al. 2014; Zhang et al. 2014a). The TGD operation decreased the magnitude of extreme flow in the summer season (Gao et al. 2013) and partially contributed to flood control. However, the frequency and timing of severe floods in the middle–lower Yangtze River are affected because of the channel incision caused by the TGD (Nakayama & Shankman 2013). The channel incision downstream of the TGD is mainly distributed along the main stream of the Yangtze River (Dai & Liu 2013). The lowering water level reduces the ability of the water discharge to transfer to the water diversion area. Channel changes caused a new water regime situation to appear in the middle–lower Yangtze River basin. Understanding the influences of channel change on the water regime situation is crucial for flood control, the management of the ecological environment and water resources in the Yangtze River basin. Therefore, quantitative assessments of the contribution of the channel changes to the water regime changes are urgently needed.

The Jingjiang Three Outlets (JTO), which form the main entrance of the water discharge of the Yangtze River entering the Dongting Lake, directly play a dominant role in flood control in the middle–lower Yangtze River, Dongting Lake evolution (Ding & Li 2011), and ecological environmental changes (Hu et al. 2015). During the flood season, the JTO recharge 20–30% of the main river water discharge (Lu et al. 2012), thereby reducing the flood pressure on the lower reaches. Meanwhile, the diverted discharge can relieve the lake droughts during the other seasons. After 2003, the amount of discharge diversion at the JTO significantly decreased (Lu et al. 2012). Chang et al. (2010) concluded that the TGD channel erosion at the main stream and the JTO jointly caused the decrease in the discharge diversion ratio, and then resulted in the amount of discharge diversion decreasing. However, Zhu et al. (2015) believed that the decreased ratio and amount of discharge diversion were mainly caused by the hydrological variations of the main stream. Previous studies still disagree on the degree to which channel changes affect the discharge diversion of the JTO. Therefore, figuring out the channel change characteristics and the contributions of the channel changes to the discharge diversion of the JTO is necessary. Specifically, this study has the following objectives: (1) to reconstruct the water level–discharge rating curves for detecting the river channel change characteristics and quantify the effect of the river channel changes on lowering the water level; (2) to establish equations on the discharge diversion and level difference of the Yangtze River and the JTO, thereby estimating the discharge diversion under no channel changes; and (3) to analyze the influence of the channel changes on the discharge diversion.

The water level–discharge rating parameters can be related to the physical characteristics of the river channel. a is a scaling factor that encompasses the section width, local bottom slope, and Manning coefficient (Valipour 2012, 2015b; Khasraghia et al. 2015). b includes the river bank geometry, particularly the departure from the vertical banks (Leon et al. 2006). is connected to many factors that affect the rating curve (Jalbert et al. 2011).

Commonly, climatic and hydrologic series may generally display serial autocorrelation. Many studies have indicated that the serial autocorrelation may alter the MK and Pettitt test results. To eliminate the effect of a serial autocorrelation on the MK and Pettitt test, the ‘Trend-Free-Pre-Whitening’ procedure was applied in this study (Yue et al. 2002).

The amount of discharge diversion at the JTO is affected by many factors, such as the relative position between the diversion entrance and the main stream, river regime variations of the main stream, and scouring and silting changes of the diversion channel. Among these factors, the level differences between the water level at the main stream and the riverbed elevation at the JTO were closely related to the amount of discharge diversion (Lu et al. 2012).

The trend and abrupt changes of the water level–discharge rating parameters at Zhicheng during 1980–2014 were calculated by the MK and Pettitt tests (Table 1). and show increasing trends with the 99% confidence level, whereas b displays a decreasing trend at the same confidence level. The parameters ⁠, b and all display an abrupt change point in 1993 with the 95% confidence level. The parameter changes indicate that the river width–depth ratio dramatically decreased and the river channel depth increased during 1980–2014. The river channel morphology exhibits drastic changes.

Table 1

Results of the MK and Pettitt tests for the rating parameters

.	.	MK test .		Pettitt test .
Parameter .	Data periods .	Z .	Trend .	Change point .	p .
	1980–2014	5.04**	Increasing	1993	0.004**
	1980–2014	−5.01**	Increasing	1993	0.004**
	1980–2014	4.596**	Decreasing	1993	0.013*

.	.	MK test .		Pettitt test .
Parameter .	Data periods .	Z .	Trend .	Change point .	p .
	1980–2014	5.04**	Increasing	1993	0.004**
	1980–2014	−5.01**	Increasing	1993	0.004**
	1980–2014	4.596**	Decreasing	1993	0.013*

*Significant at p < 0.05.

**Significant at p < 0.01.

View Large

Normally, the relationship between the water level and the discharge gradually varies under natural conditions (Zhang et al. 2015). However, the parameter changes are remarkable after 2003. The rating curves move downward when the discharge is less than 30,000 m³/s (Figure 6(a)). The channel at the main stream significantly erodes after the TGD operation (Figure 6(b)). Most sediments are trapped in the TGD, and the downstream channel erosion is severe (Yang et al. 2014), particularly in the Yichang–Chenglingji reach, after the TGD operation. The average channel erosion depth reaches 2.1 m in the Yichang–Zhicheng reach and 1.1 m in the Jingjiang reach (Lu et al. 2011), thereby resulting in a significant water level decrease with the same discharge (Figure 6(a)). Not all of the channel changes of the post-TGD period can be attributed to the TGD. However, many studies have found that the TGD operation is the main driver for the channel change (Dai & Liu 2013). Accordingly, 65–85% of the changes are attributed to the TGD (Yang et al. 2014).

The various outlets present distinct reductions in discharge diversion (Figure 9) because of the channel morphology at the different outlets, which exhibits different change patterns (Figure 2). The water level with the same discharge at Xinjiangkou decreases from 2003 to 2012, which indicates channel erosion (Figure 2). By contrast, the rating curves remain relatively stable at Shadaoguan, which suggests no obvious channel changes. The rating curves for Mituosi, Ouchi (kang), and Ouchi (guan) dramatically shift upward because of channel deposition. The results suggest that the deposition of the diversion channels under the same main stream conditions aggravates the reduction of the water-level differences between the main stream and the JTO and leads to a decrease in discharge diversion. The erosion relieves the reduction of the water-level differences and increases the discharge diversion.

At the same time, the amount of discharge diversion accounts for approximately 31% of the total discharge into the Dongting Lake (Liu et al. 2011), which is closely related to the water level and area variations of the lake. Thus, the changes in the amount of discharge diversion would significantly influence the lake ecosystem. Therefore, further work should be conducted to detect the corresponding influences of hydrological alteration on the lake ecosystem.

Table 2 shows the discharge diversion ratio of the main stream at Zhicheng to the JTO from 1994 to 2014. The discharge diversion ratio at the JTO can reflect the interactive strength between the Yangtze River and the Dongting Lake. The observed total discharge diversion ratio is 17.31% and 15.11% in 1994–2002 and 2003–2014, respectively. The estimated total discharge diversion ratio under no channel changes is 16.79% in 2003–2014. The observed total discharge diversion ratio in 2003–2014 decreases by approximately 2.20% compared with that in 1994–2002. The channel changes reduce the total decrease of the discharge diversion ratio to 1.68% (Table 2), which is nearly 76.36% of the total decrease of the discharge diversion ratio in 2003–2014. This result reveals that the channel changes are the major factor that affects the decrease of the discharge diversion ratio after the TGD operation. This result is also consistent with the perspectives obtained from several previous studies (Chang et al. 2010; Lu et al. 2012).

The changes in the discharge diversion ratio at different control hydrological stations display different amplitudes (Table 2). The channel changes in 2003–2014 result in average reductions of discharge diversion ratios of 2.58%, 4.76%, 23.84%, 69.23%, and 20.06% in comparison with the observed discharge diversion ratio.

Overall, the analysis shows that the channel changes have not yet significantly affected the discharge diversion ratio. However, channel erosion and incision downstream of the TGD are expected to continue to occur in the foreseeable future (Gao et al. 2015b, 2015c), thereby further lowering the water level with the same discharge. The channel deposition at the JTO entrance will also continue to occur (Lu et al. 2012). In the future, the discharge diversion ratio may inevitably decrease, which could increase the flood pressure of the lower reaches during the flood season. Thus, the flood diversion program downstream of the Yangtze River should be considered.

This study aims to detect the responses of the discharge diversion at the JTO to the significant channel changes that occurred at the main stream and the JTO after the TGD operation. The major findings are as follows:

• (1) The channel morphology changed tremendously based on the rating curves and cross-sectional channel profile analysis after the TGD operation. The channel incision at Zhicheng resulted in a significant decline of water level with the same discharge. By contrast, the channel deposition at the JTO caused the water level to rise with the same discharge.

• (2) Only 37.74% of the amount of discharge diversion decrease was attributed to the channel changes in 2003–2014. The hydrological variations of the main stream are the dominant factor in decreasing the amount of discharge diversion after the TGD operation.

• (3) The discharge diversion ratio decreased by 2.20%, nearly 76.36% of which was attributed to the channel changes after the TGD operation. The channel changes were the primary factor in facilitating the discharge diversion ratio decrease. The discharge diversion ratio will inevitably further reduce in the near future with the continuous channel erosion at the main stream and the channel deposition at the JTO. This result will potentially increase the flood pressure for the lower reaches in the flood season. Hence, the flood diversion program downstream of the Yangtze River should be considered.

This work is supported by the National Key Basic Research Program of China (973 Program) (2012CB417006) and the National Natural Science Foundation of China (No. 41571107 and No. 41601041). We thank Prof. J. H. Gao for his detailed and helpful comments on the manuscript. We also gratefully acknowledge the constructive and valuable comments of two anonymous reviewers who have helped to improve the quality of this manuscript.