Numerical evaluation of flow regime changes induced by the Three Gorges Dam in the Middle Yangtze

The waterbodies in the Middle Yangtze basin (here, referring to the stretch from Yichang to Datong) immediately downstream the TGD are highly influenced by the TGD. They include many shallow lakes and low-lying alluvial plains shaped by the interaction between the Yangtze River and large tributaries (Yin et al. 2004). China's two largest freshwater lakes, Poyang and Dongting, are located in the region, and naturally interact with the Yangtze River (Zhu & Zhang 1997; Dou & Jiang 2000). The Middle Yangtze, particularly two China's largest freshwater lakes, provides habitats for many important aquatic animals and plants, and migratory birds including a large number of endangered species (Kanai et al. 2002; Fang et al. 2006; Ji et al. 2007; Harris & Zhuang 2011; Fang et al. 2012a; Dong 2013). However, seasonal hydrological droughts, characterized by extremely low water levels, have been frequently occurring in this region, especially in Poyang Lake, since the TGD operation (Min & Zhan 2012). The extremely low water levels and apparent shrinkage of the great lakes were continuously observed in 2006, 2007, 2009, and 2011 (Lai et al. 2014a). The hydrological changes may break down the long-standing ecological balance and cause biodiversity decline (Milliman 1997; Cao & Fox 2009). Consequently, many efforts have been made to investigate flow regime changes in this region and also their causes (Feng et al. 2013; Gao et al. 2013; Liu et al. 2013, among others). Great attention has been paid to the TGD due to its huge regulating capacity of 22.15 km³. Despite no change in annual total discharge, the operation of the TGD has altered seasonal patterns of flow regimes in the Middle Yangtze by regulating natural discharge downstream. The potentially irreversible changes induced by the TGD may thus increase the risk of the utility of water resources and ecological degradation of wetlands and aquatic ecosystem in the Middle Yangtze.

Various flow regime changes in this region caused by the TGD were investigated, including the river discharge alteration in the mainstream of the Yangtze River (Dai et al. 2008; Gao et al. 2013), lake inundation area (Feng et al. 2013; Liu et al. 2013), and water level changes (Fang et al. 2012a, 2012b; Guo et al. 2012; Zhang et al. 2012a; Lai et al. 2013a; Nakayama & Shankman 2013) in Poyang and Dongting Lakes. These insightful studies make a strong case that the TGD has altered the river discharge in the mainstream of the Yangtze, and changed the Yangtze interaction with the connected lakes, and their water level changes. However, significant climate variability and other intensive human activities have influenced the Middle Yangtze simultaneously (Chen et al. 2001; Gemmer et al. 2008; Zhang et al. 2011, 2012b; Lai et al. 2014c). These various factors complicate the understanding of the TGD's impact on flow regimes. Meanwhile, this system is known for its complicated flow regimes induced by the complex river–lake interaction. Therefore, direct comparison of flow regimes before and after the TGD is inadequate to identify the contribution of the TGD to the flow changes from observed changes. For example, convincing evidence shows that the intensive sand mining along its outflow waterway significantly affects the Poyang Lake level (Lai et al. 2014c). Using a hydrodynamic model, the TGD's contribution to the recent extremely low water levels in the main stream was presented by Lai et al. (2014a). It clearly shows that the TGD contributes just part of the flow changes in the mainstream of the middle Yangtze River, even though such changes are dominant in some months. Dynamic mechanism for the hydrological responses of Dongting and Poyang Lakes to the TGD impoundment was also investigated (Lai et al. 2014b). However, all of these investigations have not demonstrated the spatial and temporal pattern of flow regimes caused by the TGD in the main waterbodies of the Middle Yangtze. It will influence the administration to take pointed countermeasures for alleviating the TGD's impact on water resources and ecology.

We employed a large-scale hydrodynamic model (Lai et al. 2014b) that can well reproduce flow regimes in this river–lake system to quantify the TGD's impact. Specific daily water level and discharge changes caused solely by the TGD based on its real operation during the years 2006–2011 were estimated and the distribution of the TGD's impact on flow regimes in time and space was derived. Our results revealed the contribution component of the TGD to the recent flow regime changes in the Middle Yangtze and identified the TGD's main action time periods and water bodies that have not been fully addressed by previous studies. Our comprehensive analysis of the TGD's influencing magnitude and pattern in time and space on flow regimes in the Middle Yangtze River will be useful for addressing the TGD-induced environmental issues, optimizing the TGD's operation, and generating adaptive management strategies for the complex river–lake ecosystem.

The TGD (30°49′N, 111°00′E) in the Yangtze River is located in Sandouping Town, Yichang City, Hubei Province of China. It is about 45 km upstream of Yichang hydrological station, a boundary between the upper and middle streams of the Yangtze River (Figure 1). The dam construction lasted 17 years from the preparatory work in 1993 to its completion in 2009 (http://www.ctgpc.com.cn/en/). The water was filled to 135 m above mean sea level in June 2003. The water level of the reservoir fluctuated from 135 m to 139 m and had a relative small alteration of downstream flow regimes. In 2006, the main concrete dam of 185 m high was completed and the reservoir level was then elevated to 156 m for initial pilot operation. From 2008, the pilot impoundment of 170 m was launched. The targeted level of 175 m was reached for the first time on 26 October, 2010. This marked the start of the TGD's operation at full capacity. The impoundment milestones of the TGD are listed in Table 1.

Such operation of the TGD has obviously altered the temporal allocation of river discharge, although the TGD operation maintains the annual balance of inflow and outflow. The river discharge has a significant reduction in the impoundment period but an obvious increase in the water releasing period. The most significant reduction of river discharge occurs in October, when the TGD starts to store water. The mean discharge reduction reached 3,602 m³/s during the years 2006–2011, only 34.9% of normal discharge in October. The most significant increase was observed in the water releasing period when the TGD dropped the water level to the flood restricted level of 145 m. The discharge in May had an increase of 1,093 m³/s, or 11.4% of normal discharge in that month.

Daily water level and/or discharge (2006–2011) at over 40 main gauging stations for providing initial and boundary conditions, modeling validation and impact analysis were collected from the Changjiang Water Resources Commission (http://www.cjh.com.cn) and other local hydrological agencies. Part of the water level data in 2011 is not available for modeling results validation, e.g., Xingzi in Poyang Lake. The reservoir water level of the TGD, inflow discharge into the TGD, and its outflow discharge data (2006–2011) were acquired from the online platform of the China Three Gorges Corporation (http://www.ctg.com.cn/inc/sqsk.php).

Various types of models based on physical mechanism or data are popularly used in the hydrological community. Although there are successful applications of data-based models to forecast streamflow (e.g., Chau & Wu 2010;,Taormina & Chau 2015; Li et al. 2016), hydrodynamic models have special advantages in reproducing physically reasonable dynamic processes of water movement. They can simulate hydrodynamics within a given water system with a high spatial resolution, seen in tens of meters. They are the preferred application for demonstrating the details of the TGD's impact over this complex river–lake system. Here, we selected a developed large-scale hydrodynamic model termed ‘CHAM’ to simulate the flow regimes in the Middle Yangtze with its complex river–lake interaction (Lai et al. 2013b). This model couples the one-dimensional (1D) hydrodynamic model for rivers and the two-dimensional (2D) hydrodynamic model for large lakes. The 1D model was constructed on the 1D Saint-Venant equations and solved by Preissmann scheme (Cunge et al. 1980) using a finite difference method. The 2D model was developed on the 2D shallow water equations and solved by approximate Riemann solver (Toro 2009) using an unstructured finite volume method. The 1D and 2D model components are dynamically coupled using a four-step procedure (Lai et al. 2013b) by appropriately defining overlapping zones in the computational domains.

Its implementation for the hydrodynamic modeling in the Middle Yangtze River (CHAM-Yangtze) considered the main rivers and the lakes in the middle-lower Yangtze River reach from Yichang to Datong (see the generalized water system in Figure S1, available with the online version of this paper). Particularly, the river–lake interactions are elaborately considered in this model, which allows us to make a robust hydrological analysis in this complex river–lake system. The previous study (Lai et al. 2013b) showed that the CHAM-Yangtze satisfactorily simulated the major hydrodynamic processes with significant seasonal changes and strong river–lake interactions in this water system. Main parameters of the model, such as terrain data, roughness coefficients, etc., can be referred to in the literature (Lai et al. 2013b).

The TGD-induced impact analysis was conducted based on its real operation during the years 2006–2011. Imposing measured discharge (2006–2011) data at all inflow and lateral boundaries, we ran the CHAM-Yangtze under the given rating curve at the outflow boundary at Datong and initial water level and discharge in the Middle Yangtze. The flow processes during 2006–2011 are achieved with a high time resolution (300 s in this computation). The daily water level and discharge are then averaged from these computed time-series data. The hydrograph of daily water level and discharge at selected controlling stations are compared with the measured values (Figures S2–S7, available with the online version of this paper). The well-consistent hydrographs showed that the model satisfactorily reproduced the hydrographs of discharge and water level during the years 2006–2011 in this complex river and lake system.

After model validation, the flow regime changes solely induced by the TGD's operation during 2006–2011 were quantified by subtracting the reproduced flow regime at that time in the Middle Yangtze from the restored one without the influence of the TGD. The restored flow regime was computed by running the same model with the previous model parameters, except the discharge boundary condition at Yichang, whose discharge was altered by the TGD's operation. The discharge (2006–2011) at Yichang without the TGD's operation was restored from the observed inflow discharge to the TGD.

The phase differences in days shown in the hydrographs of discharge changes induced by the TGD (Figure 3) also identify the propagation speed of its impact. On average, the TGD affected flow regimes with a mean lag-time of 3–4 days at Luoshan and of 8–10 days at Datong. The storage effects of flow also caused a longer influencing time of the TGD's regulation on downstream rivers.

Considerable increases of the averaged water level (above 0.18 m monthly increase at Datong) were observed from February to June, compared with slight increases in December, January, July, and August. TGD operation has caused two peaks of water level increase in early March and early June (Figure 4). The causes for those increases are different. The operation of the TGD resulted in the steady discharge of about 5,500 m³/s in early March after its normal operation in 2009, that is larger than the natural flow discharge (about 4,200 m³/s at Yichang averaged during 1956–2002 in January–March, the driest months in the Middle Yangtze). According to its operation guide, the TGD should release more water than the natural inflow to decrease the reservoir water level to the flood restricted level of 145 m from the normal water level of 175 m before June 10 of a year. The increased discharge caused a significant rise of the water level in late May and early June during recent years.

The most significant drops of water level were found in the impoundment period from September to November. The water levels decreased substantially during this period compared to the natural regime without the TGD's operation. In October when the TGD started to store water, the monthly water level of the main stream decreased by 1.44 (±0.71) m at Luoshan, 1.23 (±0.63) m at Hankou, and 0.81 (±0.4) m at Datong averaged during the years 2006–2011. The water levels at the outlets of Dongting and Poyang lakes decreased by 1.51 (±0.77) m (Chenglingji) and 1.03 (±0.57) m (Hukou), respectively. The decreased water levels resulted in an earlier start of the dry season in the Middle Yangtze.

The water level alteration along the Middle Yangtze caused by the TGD can help us to identify the main water areas that are significantly influenced by the TGD's operation. The identification of those areas has special implication for the management of Dongting and Poyang lakes regarding their ecological importance (Harris & Zhuang 2011). However, it has not been fully investigated as yet, except for Dongting Lake (see Lai et al. 2013a).

The TGD has more complex influences on the water levels of Dongting and Poyang lakes. Poyang Lake, the largest freshwater lake of China, connects to the Yangtze River via a single channel. The TGD-induced Yangtze River discharge changes have altered the interaction between Poyang Lake and the Yangtze River. The largest impacts of the TGD on Poyang Lake are observed at the lake mouth (Hukou), and the least impacts further to the south (Figures 4 and 5). The annual largest rise and fall in monthly water level averaged during 2006–2011 was 0.29 (±0.22) m in March and 1.03 (±0.57) m in October at Hukou, the lake mouth station; 0.14 (±0.11) m in June and 0.52 (±0.35) m in October at Duchang, the middle lake station; 0.07 (±0.09) m in June and 0.08 (±0.11) m in October at Kangshan, the southern lake station. The results clearly indicate that the TGD-induced water level variations weaken gradually along the reverse direction of flow from north to south. The TGD's impact on Poyang Lake level represents a spatial pattern with high impact in the north (close to the main channel) but low impact in the south.

Dongting Lake, immediately downstream the TGD, receives water from the Yangtze River via the Sankou Distributary Rivers, but drains water into the Yangtze River via a single outlet at Chenglingji (Figure 1). In the northern area of Dongting Lake, water level variations were gradually weakened from the diversion outlets of the Yangtze River to Dongting Lake due to the direct discharge modification. Mean reductions of water levels in October were 1.11 (±0.88) m at Xinjiangkou, 0.79 (±0.53) m at Shiguishan, and 0.51 (±0.32) m at Nanzui. In the eastern area of Dongting Lake, water level variations were also diminished gradually from the lake mouth to the central lake and the low-lying land of the inflow rivers. Mean reductions of the water levels in October were 1.51 (±0.77) m at Chenglingji, the lake mouth station; 1.33 (±0.66) m at Lujiao, the eastern Dongting lake station; 1.13 (±0.57) m at Yintian, the southern Dongting Lake station; 0.48 (±0.29) m at Changsha, the Xiangjiang River station; 0.23 (±0.18) m at Shatou, the Zishui River station. The south-western area of Dongting Lake showed little impact from the TGD. The mean water level reduction of 0.15 (±0.11) m was observed in October at Zhouwenmiao station. Thus, the TGD's operation can directly change the water discharge into Dongting Lake, as well as reverse propagation of the TGD's impact against flow direction from the lake mouth. This resulted in a complex spatial pattern of the TGD's impacts on Dongting Lake that can be generalized as strong impact in the north and east, and weak impact in the south and west.

The acceleration of the Yangtze River recession means that the Middle Yangtze went into the low water season earlier than the natural regime, causing a prolonged dry season in the Middle Yangtze. Using the mean water levels in November as an example, the dry season started nearly half a month earlier in the lower reach of the Middle Yangtze. The low water levels appeared 16 d earlier at Hankou, 12 d at Datong, 34 d at Chenglingji, and 13 d at Hukou (Table 2). This earlier start of the dry season may cause various ecological consequences in this region sensitive to flow regimes because October is the month of importance for wetland vegetation and late rice growth.

The TGD's impact on the complex flow regimes in the Middle Yangtze was estimated by employing a large-scale hydrodynamic model that considered river–lake interaction. We successfully identified the whole pattern of such influences in time and space over this large-scale river–lake system, which cannot be observed or derived from measured hydrographs. The seasonality of the TGD's flow regulation determined the seasonal pattern of the flow regime modification. The operation of the TGD increased the river discharge and the lake water levels from December to August, but decreased them from September to November. The most significant flow regime changes induced by the TGD were observed in October, the main time period of the impoundment. An accelerated river recession rate and a marked decline in magnitude in October produced an earlier and prolonged dry season in the Middle Yangtze. High heterogeneous but regular spatial distribution of the river discharge alterations and of the lake water level fluctuations were mainly attributed to the complex river system and river–lake interaction. The main areas influenced by the TGD included the main stream of the Yangtze River, Sankou Distributary Rivers that convey water into the Dongting Lake, the east and north of Dongting Lake, and the north of Poyang Lake. In the south-west of Dongting Lake and the south of Poyang Lake, the TGD has very limited effects due to its weak hydraulic connectivity with the Yangtze River. Our study on the TGD's influencing magnitude and pattern clearly indicated the role of the TGD in the recent seasonal droughts, namely, alleviating the droughts in the spring but aggravating them in the autumn.

The operation of the TGD has altered the natural regime of river discharge in the Upper Yangtze River. The altered flow regime has affected the Middle Yangtze, resulting in a series of environmental changes, especially the wetlands highly controlled by water level fluctuation. This study provided essential flow regime data for the TGD-induced environmental impact assessment. Our results may facilitate administration to optimize the operation of the TGD (cf. Yang & Lu 2013) and adjust adaptive management strategy for downstream lakes with significant alterations in their flow regimes. For example, based on the identified main influenced areas and action time, some environmentally oriented operation guides can be suggested for a compromise in the TGD's operation rule.

Dams can produce adjustments in alluvial channels due to sediment trapping and flow regime change. The channel adjustments like channel downcut in return can affect water levels (Lu et al. 2011). The flow–sediment interactions for the TGD's impact are not considered in the present modeling. Considering the facts of large amounts of sediment trapped by the TGD's reservoir and currently observed channel incision, the water recharge effect of the TGD for alleviating droughts in the dry season might be balanced out. Thus, fluvial processes in the Middle Yangtze should also be paid great attention for improving the understanding of long-term impacts of the TGD's operation on downstream rivers and lakes.

This work is supported by National Key Basic Research Program of China (973 Program) (2012CB417000) and the National Natural Science Foundation of China (No. 41071021).