Characteristics and causes of water level variations in the Chenglingji–Jiujiang reach of the Yangtze River following the operation of the Three Gorges Dam

Natural rivers often have a dynamic balance of erosion and deposition due to stable water and sediment conditions and riverbed boundaries over many years (Williams & Wolman 1984). However, the natural balance has been disrupted by increasing anthropogenic stresses on rivers, especially the construction of dams, with >45,000 large dams (heights >15 m) distributed across 140 countries (Ma et al. 2022; Syvitski et al. 2022). Therefore, geomorphological and hydrological changes and their influences on rivers under different conditions (e.g., natural status, anthropogenic stresses, climate change) have always been important topics in the fields of river geomorphology, river engineering, and river ecology (Schmidt & Wilcock 2008; Gao et al. 2021; Li et al. 2021).

The construction of large dams changes the spatial and temporal distributions of water and sediment downstream of dams, which decreases the original stability of the basin. Accordingly, the water level downstream of a dam changes, affecting the river regime, water supply, navigation, flood control, and ecology (Surian & Rinaldi 2003; Boulange et al. 2021; Hu et al. 2022b). However, the variations in the water level in the lower reaches of dams in different rivers greatly differ. For example, after the completion of the Aswan Dam of the Nile River (Saad 2002), the Glen Canyon Dam of the Colorado River (Shields et al. 2000), and the Fort Peck Dam of the Missouri River (Topping et al. 2003), the downstream channels generally experienced a decrease in the water level and a decrease in the water surface gradient due to cutting of the riverbed. After the Danjiangkou Reservoir of the Hanjiang River was impounded, the middle and low water levels at the downstream hydrological stations decreased, but the high water level increased by approximately 1.47 m from 2016 to 2017 due to the partial construction of the Huangzhuang–Datong Reach (Xiao et al. 2018; Huang et al. 2022). After the Xiaolangdi Reservoir of the Yellow River was impounded, the riverbed of the Huayuankou Reach was cut down by 3.44 m. However, coarsening of the riverbed promoted the development of larger sand dunes and thus increased the river resistance, resulting in an increase in the flood level when the discharge was greater than 6,100 m³/s (Ma et al. 2022). Therefore, it is necessary to study the mechanism of differential changes in water levels in large dammed rivers.

The CJR is dominated by bifurcated rivers, and large cities such as Wuhan and Jiujiang are distributed within this region (Figure 1). With abundant hydrological and topographic data, changes in the water level have always been the focus of research. Since the operation of the TGD, flood control in the lower reaches of the dam has significantly improved; however, the main object for flood control of the TGD is the Jingjiang Reach, and the flood threat downstream of Chenglingji is still large due to the impacts of extreme rainfall, tributary flow, and climate change (Zhu et al. 2011; Xia & Chen 2021; Zhou et al. 2021). For example, the largest flood occurred in the Yangtze River since the operation of the TGD in 2020. The joint operation of reservoirs in the upper and middle reaches of the Yangtze River reduced the maximum water level from Yichang to Shashi by 3.0–3.6 m, but the Hankou station still experienced the fourth highest peak water level in history (Chen 2020). In addition, there are a large number of water-taking and water-regulating projects along the CJR, so it is urgent to study the characteristics and causes of water level variations after the operation of the TGD.

Thus, taking the CJR as an example, we analyzed water level change trend by using rich hydrological data via the Mann‒Kendall method, and computed geomorphic change and river resistance by using 406 cross-sectional profiles via cross-sectional topography method and one-dimensional hydrodynamic mathematical model. The objectives of this study are as follows: (i) to identify the characteristics of water level variations under different specific discharges and clarify the contributions of river erosion and silting, downstream water level adjustment, and change in resistance to these variations and (ii) to identify the critical conversion discharge of water level variations first and quantitatively explain the causes of critical conversion discharge. This study can provide a reference for predicting trends in water level adjustment and formulating countermeasures, such as navigation and flood control, after the construction of dams in different rivers around the world.

As the third longest river in the world, the Yangtze River, with a total length of ∼6,300 km, is generally divided into upper, middle, and lower reaches according to different hydrological characteristics and geographical settings (Lyu et al. 2020). The ∼521 km long reach studied herein, i.e., the CJR, is located between Chenglingji and Jiujiang in the lower part of the Middle Yangtze River, ∼446 km downstream of the TGD. The Han River, the largest tributary of the Yangtze River, flows in the middle of the CJR. The Dongting and Poyang floodplain-type lakes are at the head and end of the river, respectively (Figure 1), and now function as tributaries to the Yangtze River and create backwater effects in the river channel (Hu et al. 2022b).

As shown in Figure 1 and Table 1, the CJR covers 406 specified cross-sections surveyed by the Changjiang Water Resources Commission (CWRC), with an average spacing between two successive sections of approximately 2 km. The CJR is a typical braided channel with some central bars in the middle of the river. Most riverbanks in this reach consist of loose sediments that are often stratified, and the riverbed consists mainly of fine sand, with median diameters of 0.12–0.20 mm (CWRC 2021). The three hydrometric stations, Luoshan, Hankou, and Jiujiang, are located approximately 488, 697, and 967 km downstream of the TGD, respectively (Figure 1), and these hydrometric stations are representative of the hydrological conditions in the CJR. The CJR is classified into two subreaches, the Chenglingji-Hankou Reach (CHR) and the Hankou-Jiujiang Reach (HJR).

As shown in Table 1, hydrological data (i.e., daily mean discharges and water levels at the Luoshan, Hankou, and Jiujiang stations from 1991 to 2022) and topographical data (i.e., one-dimensional cross-sectional profiles at 406 specific locations for 2004, 2013 and 2020) were collected from the CWRC. Because the TGD began to fill with water in 2003, 2003 was taken as the boundary year to study the change in the water level before and after the completion of the reservoir. Because the TGD began to be affected by the storage of cascade reservoirs in 2013 (Dong et al. 2019), 2013 was taken as an important time node in the following analysis. Therefore, we divided the study period into 1991–2002, 2003–2012, and 2013–2022, which are called period 1, period 2, and period 3, respectively.

Mann‒Kendall (M-K) analysis is an effective tool for analysis of the trends in hydrological elements, and its advantage is that samples do not need to follow a certain distribution (Hamed 2008; Li et al. 2022). M-K analysis was adopted to analyze the trends of the water level time series at the hydrological stations and to quantitatively determine the significance of the trends. The change in the trend is assessed by the statistical value Z. If Z > 0, the series shows an upward trend; otherwise, it shows a downward trend. Given the significance level α = 0.05 and corresponding Z_1-α/2 = ±1.96, if |Z| > 1.96, the hydrological series is considered to have a significant trend. The index used to measure the magnitude of the trend was also the Kendall inclination β, which indicates the amount of change per unit time; β > 0 indicates an increase, β < 0 indicates a decrease, and the magnitude of the β value represents the average rate of change.

We can conclude from Equation (7) that the variation in the water level is mainly the result of the combined effect of channel erosion and deposition, downstream water level variation, and resistance change. These three aspects are also the focus of the following analysis.

To explore the trends in the variations at different water levels and identify the critical conversion discharge, representative discharges need to be selected at each hydrological station. The specific discharges were selected based on the following principles: (a) discharges must cover both flood and dry events; and (b) long-term time series of water levels can be obtained (Yang et al. 2018; Chai et al. 2020; Hu et al. 2022b). According to these principles and relevant references (Zhang et al. 2020), five specific discharge events were selected as the characteristic discharge events: low, low–medium, high–medium, bankfull, and flood (Table 2).

To analyze the trend of variation in the water level in the CJR before and after the operation of the TGD and to identify the critical conversion discharge, five specific discharges were selected for M-K analysis at the Luoshan, Hankou, and Jiujiang stations (Table 3). Since multiple data correspond to different water levels at the same discharge in each year and because the selected discharge data are missing in some years, to ensure the accuracy and completeness of the corresponding water level of the selected discharge in each year, the average water level within the range of each specific discharge ±5% was used in the analysis of the trend for each year. Due to missing water level and discharge data in some years, water level–discharge relationship interpolation was used to determine the data.

Table 3 shows that the water level at the different discharge levels exhibited different trends of change before and after the operation of the TGD. Before the operation of the TGD (1991–2002), except for the low discharge water level at the Jiujiang station, which showed a significant downward trend, the absolute Z value of the three stations at each discharge level was less than 1.96, indicating a nonsignificant trend. After the operation of the TGD (2003–2022), under low discharge, low–medium discharge, and high–medium discharge conditions at the Luoshan station, the Z value was less than −1.96, and the β value was negative, so the water level exhibited a significant downward trend. As the discharge increased, the Z and β values were close to 0 at bankfull discharge. It can be preliminarily determined that ∼35,000 m³/s is the critical transition point for water level change at the Luoshan station. Under the low discharge and low–medium discharge conditions at the Hankou station, the Z value was less than −1.96, and the β value was negative, so the water level exhibited a significant downward trend. As the discharge increased, the Z and β values were close to 0 at high–medium discharge. Similarly, we concluded that ∼30,000 m³/s is the critical transition point at which the water level changes at the Hankou station. The trend of the water level at the Jiujiang station was the same as that before the operation of the TGD, and the water level showed a significant downward trend under low discharge. As the discharge increased, the Z and β values were close to 0 at high–medium discharge. Therefore, ∼30,000 m³/s is also the critical transition point for water level change at the Jiujiang station. Under flood discharge, the Z value was greater than 0 but less than 1.96, and the β value was positive at all stations, so the water level showed a nonsignificant upward trend.

The years 2004 and 2020 were selected as the initial and current representative years after the operation of the TGD, respectively. As shown in Table 4, the water level‒discharge relationships at the Luoshan, Hankou, and Jiujiang stations were fit to calculate the water level under a specific discharge. Under low discharge conditions, the water levels at the Luoshan, Hankou, and Jiujiang stations all decreased significantly (−1.11, −1.08, and −0.58 m, respectively), implying that the decrease in the water level at downstream hydrometric stations contributed to the decrease in water level at upstream hydrometric stations. Then, with increasing discharge, the influence of the downstream water level gradually weakened. Under flood discharge conditions, the water levels at the Luoshan, Hankou, and Jiujiang stations all obviously increased (0.45, 1.15, and 1.13 m), indicating that the increase at the downstream hydrometric station contributed to the increase in the water level at the upstream hydrometric station. The specific impact of the water level variation in the downstream hydrometric station under different specific discharge conditions on the water level of the upstream hydrometric station is specifically analyzed in Section 4.5.

Using the one-dimensional hydrodynamic model, Manning's roughness coefficient (n) of the river under five specific discharge levels was estimated based on the 406 observed fixed section profiles and the measured hydrological data at the Luoshan, Hankou, and Jiujiang stations in 2004, 2013, and 2020. As shown in Table 5, taking the increase in the value/absolute value of the roughness coefficient as a proportion of the change, the roughness changed little at each discharge level before 2013 and decreased slightly (−2.8 to −4.08%) in the CHR, while the roughness increased slightly (0–7.62%) in the HJR. After 2013, the roughness at each discharge level increased significantly, and the roughness in the CHR clearly increased (by more than 7%) after the discharge reached bankfull discharge. The roughness in the HJR increased more obviously (close to 20%) after the discharge reached the high–medium discharge. In general, the increase in the amplitude of roughness in the HJR was greater than that in the CHR.

Figure 6 clearly shows the causes of the water level variations at the different discharge levels. From 2004 to 2020, the measured water level changes under the five specific discharges (−1.11, −0.91, −0.47, −0.09, and 0.45 m at the Luoshan station, and −1.08, −0.66, −0.04, 0.21, and 1.15 m at the Hankou station; Table 5) were approximately equal to the linear superposition values of three factors for the water level (−0.99, −0.85, −0.41, −0.06, and 0.47 m at the Luoshan station, and −1.02, −0.63, −0.03, 0.31, and 1.07 m at the Hankou station; Figure 5). This further verifies the conclusion of Subsection 3.4, namely, that the change in the water level is mainly the result of the comprehensive action of channel erosion and deposition, downstream water level change, and resistance change. However, the weight of influence of each factor was different under different discharge levels.

Under low discharge conditions, the influence values of topographic adjustment on the water level at the Luoshan station and the Hankou station were −1.17 and −1.83 m, respectively; the influence values of downstream water level change on the water level at the Luoshan station and the Hankou station were −0.06 and −0.05 m, respectively; and the influence values of comprehensive roughness change on the water level at the Luoshan station and the Hankou station were 0.25 and 0.92 m, respectively. Therefore, the main factor influencing the water level reduction at the Luoshan station and the Hankou station was channel erosion.

Under low–medium discharge conditions, the influence values of topographic adjustment on the water level at the Luoshan station and the Hankou station were −0.89 and −1.35 m, the influence values of downstream water level change on the water level at the Luoshan station and the Hankou station were −0.14 and −0.03 m, and the influence values of comprehensive roughness change on the water level at the Luoshan station and the Hankou station were 0.18 and 0.63 m, respectively. The main factor influencing the water level reduction at the Luoshan station and the Hankou station was channel erosion.

Under high–medium discharge conditions, the influence values of topographic adjustment on the water level at the Luoshan station and the Hankou station were −0.77 and −1.21 m, the influence values of downstream water level change on the water level at the Luoshan station and the Hankou station were 0.02 and 0.09 m, and the influence values of comprehensive roughness change on the water level at the Luoshan station and the Hankou station were 0.34 and 1.15 m, respectively. The main factor influencing the water level reduction at the Luoshan station was channel erosion, and the main reason for the lack of significant change in the water level at the Hankou station was that increasing resistance nearly counteracted the effect of channel erosion. In addition, the measured water level change and linear superposition effect of the three factors on the water level were close to 0 at the Hankou station. This further indicates that the critical conversion discharge of the water level change in the HJR was approximately 30,000 m³/s.

Under bankfull discharge, the influence values of topographic adjustment on the water level at the Luoshan station and the Hankou station were −0.74 and −1.17 m, the influence values of downstream water level change on the water level at the Luoshan station and the Hankou station were 0.1 and 0.09 m, and the influence values of comprehensive roughness change on the water level at the Luoshan station and the Hankou station were 0.58 and 1.39 m, respectively. The main cause of the nonsignificant variations in the water level at the Luoshan station was that increasing resistance nearly counteracted the effect of channel erosion. At the Hankou station, the main factor driving the increase in water level was that the effect of increasing resistance surpasses the effect of channel erosion. In addition, the measured water level change and linear superposition effect of the three factors on the water level were close to 0 at the Luoshan station. This further indicates that the critical conversion discharge of the water level change in the CHR was approximately 35,000 m³/s.

Under flood discharge, the influence values of topographic adjustment on the water level at the Luoshan station and the Hankou station were −0.67 and −1.19 m, the influence values of downstream water level change on the water level at the Luoshan station and the Hankou station were 0.29 and 0.33 m, and the influence values of comprehensive roughness change on the water level at the Luoshan station and the Hankou station were 0.85 and 1.72 m, respectively. The main reason for the increase in the water level at the Luoshan and Hankou stations was that the effect of increasing resistance was greater than that of channel erosion, and the increase in the downstream water level contributed to the increase in the water level.

A decrease in the downstream water level can be transmitted upstream to cause a decrease in the upstream water level. On the other hand, the increase in the downstream water level can increase the upstream water level. The effect of water level transmission varies greatly in different rivers. For the CJR, Section 4.5 quantifies the amplitude of the influence of downstream water level changes on the upstream water level at different discharge levels, but the underlying mechanism needs to be further clarified. Moreover, although the critical conversion discharge has been identified, why it exists and why the critical conversion discharge differs between the CHR and HJR remain to be determined.

As shown in Table 4 and Figure 6, under low discharge conditions, the water level at the Hankou station decreased by 1.08 m, resulting in a decrease in the water level at the Luoshan station of 0.06 m, and the water level at the Jiujiang station decreased by 0.58 m, resulting in a decrease in the water level at the Hankou station of 0.05 m. In contrast, under flood discharge conditions, the water level at the Hankou station increased by 1.15 m, resulting in an increase in the water level at the Luoshan station of 0.29 m, and the water level at the Jiujiang station increased by 1.13 m, resulting in an increase in the water level at the Hankou station of 0.33 m. This shows that under low discharge, the water level change at the downstream hydrometric station has a small influence on the upstream hydrometric station. In contrast, under flood discharge, the water level change at the downstream hydrometric station has a greater influence on the upstream hydrometric station. To explain the cause of this phenomenon, we conducted the following numerical sensitivity test.

Critical conversion discharge occurs widely in the lower reaches of dammed rivers. For example, after the construction of the Xiaolangdi Reservoir in the Yellow River in China, bed incision was accompanied by sediment coarsening, which facilitated the development of large dunes that increased flow resistance and reduced velocity relative to pre-dam conditions, resulting in the amplification of the downstream flood stage when the discharge was greater than 6,100 m³/s (Ma et al. 2022). Critical conversion discharge also occurs in the Yangtze River, and we propose three main reasons for the occurrence of critical conversion discharge. First, the water level at the critical conversion discharge was relatively high and close to the bankfull water level, and the channel shape was mainly a ‘V’ or ‘U’, and the channel width was significantly enlarged above this discharge level (Figure 4). Moreover, the extent of section expansion was mainly concentrated in the low water channel, and the water level decrease caused by erosion significantly decreased near this discharge. Second, reductions in the number of days of high flows caused by the TGD contributed significantly to the downstream growth of floodplain vegetation, and vegetation expansion increased floodplain resistance, which is thought to be the main factor driving the observed increase in water levels during high-flow inundation (stages at a given large discharge) in the MYR (Hu et al. 2022a, 2022b, 2023). Third, in recent years, the number of wading projects, such as bridges, wharves, and regulation projects, has increased intensively, and the pile foundations of wading projects have been mainly concentrated above low water channels. This increases the local resistance of the riverbed and the morphological resistance, and causes the water level and velocity field to experience superimposed impacts (Zhang et al. 2011; Yang et al. 2018).

As shown in Table 5, the river resistance of the CHR increased significantly when the discharge reached bankfull discharge, while the river resistance of the HJR increased significantly when the discharge reached high–medium discharge. This is why the critical conversion discharge of the HJR was lower than that of the CHR. We analyzed this difference in the composition of the river resistance. River resistance is mainly composed of grain resistance, dune resistance, bar resistance, morphological resistance, and additional resistance (Qian & Wan 1983). For the large alluvial river in the middle reaches of the Yangtze River, the bar resistance is mainly related to vegetation development, the morphological resistance is closely related to the plane shape and section shape of the channel, and the additional resistance is closely related to the density of wading projects, such as waterway regulation projects, bridges, and wharfs. At low water levels, the river resistance is mainly grain resistance, dune resistance, and morphological resistance (Hu et al. 2022a, 2024). With increasing discharge and water level, the influence of the bar resistance and additional resistance becomes increasingly serious.

The drainage capacity along the Yangtze River has significantly increased, which is also one of the important reasons for the increase in water level under the same discharge conditions. With the development of society and the economy and the acceleration of urbanization, the drainage capacity of the Yangtze River (especially in cities) has significantly increased. During heavy rainfall, rain-floodwater can be quickly pumped directly into the Yangtze River. Although the urban waterlogging situation has obviously improved, rivers and lakes have also filled rapidly, so the water level has risen sharply. In the middle and lower reaches of the Yangtze River, 55 medium-sized and above-scale pumping stations have an installed capacity of 318,000 kW, and the total drainage capacity of 57 culverts (with a designed flow of 50 m³/s or more) reaches 20,364 m³/s. In Hubei Province, which lies in the MYR, for example, 41 new pump (slug) stations have been built since 2016, increasing drainage discharge by 1,823 m³/s, and Wuhan's drainage capacity has increased from 980 to 1,960 m³/s. The waterlogging situation in the urban areas along the Yangtze River in 2020 was significantly better than that in 2016, but the floodwater level in the Yangtze River was also significantly greater than that in 2016 and was close to the level in 1998 (Chen 2020; Xia & Chen 2021). Therefore, according to the level of economic and social development and the importance of protection objects, urban and rural flood control and drainage standards should be scientifically established; the upstream and downstream regions of the river basin and the relationships between the left and right banks should be coordinated; the unified operation of large reservoirs, large sluices, and large pumping stations should be subordinated to overall interests; the comprehensive goal of control should be minimum comprehensive loss; and the relationships between external floods and waterlogging should be balanced.

It should be noted that increasing temperatures and human interventions have added stress to the region's hydrological sensitivity and have increased the risk of major flood events (Magnan et al. 2021; Shukla & Sen 2021; Rahmani & Fattahi 2024). Moreover, the strong rainstorm cycle in China has good consistency with the El Nino cycle (Kong 2020; Emberson et al. 2021; Zhou et al. 2021). It is necessary to develop better flood and drought risk management practices suitable for climate change and increased human activities.

We summarize our findings hereinafter. Decrease of the water level in the CJR has developed due to erosion of the channel bed downstream of the TGD. Increasing of the river resistance amplifies the water depth for a given water discharge. Especially, the river resistance increased by 7–20% from 2013 to 2020 when the discharge exceeded the bankfull discharge. Under the combined action of channel erosion and increased river resistance, the decreasing trend of the low water level shifts at a certain magnitude of discharge. Through our research, the critical conversion discharges of the water level changes in the CHR and the HJR are approximately 35,000 and 30,000 m³/s, respectively. Furthermore, we have clarified the underlying mechanism of why the amplitude of the influence of downstream water level changes on the upstream water level at different discharge levels is different. There are a number of obvious drops in the channel of the CJR, which form ‘overwater weirs’. Under low discharge, the throat effect blocks or weakens the propagation of the decreasing water level downstream to the upper reaches, while the effect is not obvious under flood discharge conditions. Therefore, the obvious increase in the floodwater level in individual years was due to the more obvious increase in the downstream water level.

In this article, the M-K method is used to analyze the trend of the water level variation in the CJR before and after the operation of the TGD. The influence of channel erosion and deposition, downstream water level adjustment, and river resistance change on water level change at different discharge levels is determined by using the separate variable method and a one-dimensional hydrodynamic model. Then, the critical conversion discharge of the water level change is identified first, and its causes are explained. The main conclusions are as follows:

• (1) Before and after the operation of the TGD, the water level trend of the variation in the CJR exhibited different characteristics. Before the operation of the TGD, except for the low discharge water level at the Jiujiang station, which exhibited a significant decreasing trend, the water level at all three stations exhibited a nonsignificant trend. After the operation of the TGD, the trend of the water level at the Jiujiang station was the same as before. The water level at the Luoshan station at low, low–medium, and high–medium discharge levels exhibited a significant decreasing trend. The water level at the Hankou station at low and low–medium discharge levels exhibited a significant decreasing trend. When the discharge exceeded the bankfull discharge, the water levels at the Hankou and Luoshan stations showed a nonsignificant increasing trend. The floodwater level increased obviously in some years.

• (2) The critical conversion discharges of the water level changes in the CHR and the HJR are approximately 35,000 and 30,000 m³/s, respectively. The significant decrease in the water level below the critical conversion discharge is mainly the result of the combined effect of channel erosion and increased resistance, and the dominant effect is channel erosion. The water level near the critical conversion discharge does not change significantly because the effect of increased resistance nearly offsets the influence of channel erosion, and the nonsignificant increase in the water level above the critical conversion discharge occurs because of the greater increase in resistance.

• (3) There are a number of obvious drops in the channel of the CJR, which form ‘overwater weirs’. Under low discharge, the throat effect blocks or weakens the propagation of the decreasing water level downstream to the upper reaches, while the effect is not obvious under flood discharge conditions. Therefore, the obvious increase in the floodwater level in individual years was due to the more obvious increase in the downstream water level. The contribution of tributary water input to flood-stage amplification and the associated risk factors require further studies.

• (4) More small flood discharges with higher water levels may become the new normal. In the face of great changes in flood control, we should provide enough storage space for floods and change from flood-fighting and emergency rescue modes to flood-related scientific management modes. On the other hand, overall planning for river basins is necessary to coordinate the flood control of major rivers and drainage along rivers and maintain peaceful, green, beautiful, and harmonious access to the Yangtze River.

This work was supported by the Yangtze River Water Science Research joint Fund of China (No. U2340217) and the Youth Project of National Natural Science Foundation China (No. 42301018). We would like to thank the Changjiang Water Resources Commission for providing hydrological and morphological data. We are grateful to the editor and manuscript reviewers for providing valuable comments on the revision of this paper. We would also like to thank the authors mentioned in all the references.

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

The authors declare there is no conflict.