The Three Gorges Dam (TGD) Project is the largest water conservation project in human history. It provides comprehensive benefits encompassing flood control, hydropower, shipping, water supply, energy conservation, and emissions reduction (Zheng 2016). The construction and operation of the Three Gorges Reservoir (TGR) has influenced the regulation of the Yangtze River and high/low water levels downstream and attracted the attention of researchers worldwide. From 1950 to 1998, the Yichang-Hankou reach of the Yangtze River mainly underwent a process of siltation. From 2003 to 2007, scouring dominated (Chen et al., 2010). Based on the comparison of data from 2009 to 2010 with that from 1998 to 2002, there is evidence of deep, one-dimensional undercutting in the 100 km downstream of the dam and alternating siltation and scouring further downstream (Yuan et al., 2012; Dai and Liu, 2013). The distribution of siltation and scouring along the course of the river evolved from “bed scouring and beach building” before water was stored at the TGR to “erosion of both bed and banks” afterwards (Xu et al., 2011, 2012; Dai and Lu, 2014). Scouring is concentrated in the low-water channel (Han et al., 2014) of the Yichang-Chenglingji reach of the river (Xu et al., 2011). Based on the comparison between 2012 and 2002, undercutting was the principal form of scouring in the middle and lower reaches of the Yangtze River. The river’s cross section became deeper and narrower (Xu et al., 2013). Existing studies on the channel morphology are mainly based on specific time periods after water was stored at the TGR. Previous studies did not provide a comprehensive analysis reflecting the differences in channel adjustments on the scale of individual reach of the river at different stages of the TGR’s development. Low water levels downstream of the dam are dropping (Sun et al., 2015). This effect was anticipated (Lu et al., 2002; Fang et al., 2012) because it occurs downstream of large reservoirs worldwide (Bormann et al., 2011). Worldwide, flood water levels downstream of large reservoirs generally increase slightly or remain unchanged (Bormann et al., 2011). Evaluations of the changes in flood water levels at given flow rates in the time since water has been stored at the TGR are inconclusive. Based on the results of different studies, the flood water levels have decreased (Jiang and Huang, 1997), remained basically unchanged (Li et al., 2009), slightly increased, or significantly increased (Mei et al., 2015; Zhang et al., 2016). Some research indicates that the flood water levels in the river channel downstream from the TGR have not decreased under the influence of clear water released from the reservoir. Instead, the capacity of the Jingjiang reach to divert floodwater to the Dongting Lake and the river’s overall flood discharge capacity have decreased (Mei et al., 2015). From 2003 to 2013, the water level corresponding to a flow rate of Q = 50,000 m³/s at Hankou Station has increased (Mei et al., 2015). Intense scouring of the riverbed is the main cause of the decrease in low water levels. Thickening beaches, narrowing of the flood channel, roughening of the riverbed, and increased riverbank vegetation are the main elements contributing to rises in flood water levels (Moshe et al., 2008; Greene and Knox, 2014). In the time since water has first been stored at the TGR, the channel morphology downstream has changed; the riverbed has become rougher; there are fewer flood days per year; and human activities have increased. The effects of these changes on flood water levels have not been thoroughly examined. Localized flooding occurred between June and August 2016 in the middle and lower reaches of the Yangtze River. Water levels at Luoshan and further downstream exceeded flood warning levels. However, the flow rate corresponding to flood warnings was lower than that of 1998 or 2010. This paper aims to explain the reasons for that.

In summary, there is a lack of research thoroughly describing the overall effects of siltation and scouring downstream of the TGR. How have they affected channel morphology and water levels in different reaches of the river over time? This paper clarifies the processes of riverbed morphology in individual reaches by examining water levels and channel morphology data from 1955 to 2016. It also combines analyses of riverbed and sidewall resistance and human activities that drive changes in fixed flow-rate water levels to evaluate how changing water levels in general affect the flood control situation and depth of the Yangtze River shipping channel.

The length of the Yichang-Datong reach of the Yangtze River downstream of the TGR is 1183 km. The Yichang-Dabujie reach is a 116.4 km long sandy cobble reach. Downstream of Dabujie is a 1066.4 km long sandy reach (Figure 1). The main stream of the river in the research zone includes the following hydrological stations: Yichang, Zhicheng, Shashi, Jianli, Luoshan, Hankou, Jiujiang, and Datong. Flood diversion channels to the Dongting Lake include Songzikou, Taipingkou, and Ouchikou, which are known as the three outlets of the Dongting Lake. The Xiangjiang, Zishui, Yuanjiang, and Lishui rivers converge at the Dongting Lake. These rivers are known as the four waters of the Dongting Lake. The Chenglingji hydrological station controls the discharge from the Dongting Lake into the main stream of the Yangtze River. The Hanjiang River confluence is controlled by the Huangzhuang Station. The Hukou Station controls the discharge to the river from the Poyang Lake. The Poyang Lake is fed by the Xiushui, Ganjiang, Fuhe, Xinjiang, and Raohe rivers, known as the five rivers.

The hydrological stations involved in this study are Yichang, Zhicheng, Shashi, Jianli, Luoshan, Hankou, Jiujiang, and Datong Stations. Also involved are the three diversion channels to the Dongting Lake and the discharge channels from the Dongting and Poyang lakes that are controlled by the Chenglingji and Hukou stations, respectively. The runoff rate, flow rate, sediment transport, and water level information was collected from 1955 to 2016 at each of the hydrological stations. Deposition, erosion, and cross section data were collected from the Yichang-Hukou reach from 1987 to 2014. Water level data were collected along the course of the river in the Yichang-Hukou reach from 1981 to 2014. Information on the depth and width of the shipping channel from 2002 and 2015 was also collected. Data collection times vary, but they all end within the past three years, clearly revealing recent patterns. Table 1 shows the source of each type of data.

2.3.1 Runoff and sediment transport dynamics

From 1955 to 2015, the quantity of runoff at the Yichang and Datong Stations erratically increased and decreased without a clear overall trend. Between 2003 and 2015, the average runoff at the Yichang and Datong Stations was 7.7% and 5.5% lower, respectively, than the average runoff at those stations between 1955 and 2002 (Figure 2a), mainly due to climate change (Yang et al., 2015). The sediment transport showed a decreasing trend from 1955 to 2015, especially between 2003 and 2015 at the Yichang Station (Figure 2b). This is consistent with global trends downstream of large reservoirs (Maren et al., 2013).

2.3.2 Composition of runoff and sediment transport

The relative quantities of runoff at the Datong Station originating from the Yichang Station, Dongting Lake, Hanjiang River, Poyang Lake, and the region in between remained relatively unchanged after water was stored at the TGR (Figure 2c). The proportions of sediment transport varied, on the other hand. In the three time periods before water was stored at the TGR, the sediment at the Datong Station mainly came from the Yichang Station. The Dongting Lake accumulated sediment at a decreasing rate, the proportion of sediment from the Hanjiang River decreased, and the contribution from Poyang Lake stayed close to zero. The main stream of the river had begun to accumulate sediment. After water was stored at the TGR, 29.2% of the sediment at Datong Station originated from Yichang Station. This was a much smaller proportion than during any of the three previous periods. The Dongting Lake changed from sediment sink to sediment source. Its proportion was around 6.8%. The contribution of the Hanjiang River to the total proportion of sediment at Datong was around 4.4%, insignificantly different from the 1955 to 2002 proportion. The proportion from Poyang Lake rose to 10.2%, slightly higher than during all three periods of time before water was stored at the TGR. The main stream of the river overall changed from siltation to scouring. Scoured sediment from the main stream accounted for 49.5% of the sediment transport at Datong Station.

The years 2003, 2012, and 2016 were chosen for comparison (Figure 3). At a given flow rate, the water levels at the Yichang, Zhicheng, Shashi, Jianli, Luoshan, Hankou, and Jiujiang Stations first decreased and then increased. There are critical flow rates at which corresponding water levels began to increase or decrease over time. They are close to the flood stage flow rates corresponding to the reach of the river in which these stations are located. When the river was flooded, the water levels for a given flow rate did not decrease over time, as predicted; they increased. Based on the comparison of 2016 and 2012 with 2003, the difference in flow rates between low water and flood water levels decreased. This means that the magnitude of the flow rates required for flooding gradually decreased.

3.2.1 Minimum water levels

Before water was stored in the TGR (Figure 4), the minimum water levels at the Yichang, Zhicheng, and Shashi Stations fluctuated, trending downwards. After water was stored, the minimum water levels began to increase. At the Luoshan and Hankou Stations, the minimum water levels have been rising since water has first been stored in the TGR.

3.2.2 Maximum water levels

Before water was stored in the TGR (Figure 5), the yearly maximum water levels at the Yichang, Zhicheng, and Shashi Stations fluctuated, trending downwards. Since water has been stored, the maximum water levels at these stations have followed no clear pattern; the peaks have been below pre-storage peaks. The maximum water levels at the Luoshan and Hankou Stations slightly increased before water was stored in the TGR; however, no clear trend can be observed currently.

From 1981 to 2002, the Yichang to Hukou reaches of the Yangtze River were scoured at low water levels. Scouring of the Yichang to Zhicheng reaches and Upper Jingjiang reach was concentrated in the low-water channel. Some siltation occurred between the low water level and flood plain. Siltation was observed in the Lower Jingjiang reach and Chenglingji- Hankou and Hankou-Hukou reaches of the Yangtze River. The overall evolutionary trend was “erosion of the channel and deposition in the flood plain.” From 2003 to 2015 (Xu et al., 2011, 2012, 2013), the total quantities of deposition/erosion in the Yichang-Zhicheng reach (YZR), Upper Jingjiang reach (UJR), Lower Jingjiang reach (LJR), Chenglingji-Hankou reach (CHR), and Hankou-Hukou reach (HHR) were -144, -411, -280, -218, and -389 million m³, respectively (Figure 6. Note that data for the Yichang-Zhicheng and Hankou- Hukou reaches from October 2014 to October 2015 are missing). Figure 5 compares the deposition/erosion intensity patterns of different reach of the Yangtze River in the low-water channel, typical river channel, and flood plain channel during three periods (2003-2006, 2006-2008, and 2008-2015), each representing a different water storage level of the TGR. The scouring intensity of each part of the channel in the Yichang-Zhicheng reach decreased over time. In the Upper Jingjiang reach, the scouring intensity increased in low-water and typical river channels. In the flood plain, the scouring intensity first decreased and then increased. In the Lower Jingjiang reach, the scouring intensity first decreased and then increased. There was overall deposition in the flood plain channel between 2006 and 2008. In the Chenglingji-Hankou reach, the scouring intensity generally increased in the low-water and typical channels. In the flood plain, the channel scouring intensity was weak or negative at first but increased later. In the Hankou-Hukou reach, the scouring intensity increased in the low-water channel. In the typical channel and flood plain channel, the scouring first weakened and overall deposition occurred from 2006 to 2008; the scouring intensified later.

From 2003 to 2006, the Yichang-Zhicheng reach had the greatest rate of erosion per unit river length, followed by the Lower Jingjiang reach. The Chenglingji-Hankou reach had the smallest rate. From 2006 to 2008, the Yichang-Zhicheng reach still had the greatest amount of erosion per unit of river length, followed by the Upper Jingjiang reach. The Lower Jingjiang reach had the smallest rate of erosion and the Hankou-Hukou reach had overall deposition. From 2008 to 2015, the erosion was the greatest in the Upper Jingjiang reach, followed by the Hankou-Hukou reach. Therefore, the location of the greatest erosion per unit river length in 2008 to 2015 suddenly changed to the Upper Jingjiang reach after being the Yichang-Zhicheng reach from 2003 to 2008. The intensity of erosion in the Lower Jingjiang reach and further downstream intensified over time as the effects of the release of clear water gradually increased.

The low-water channel is defined as the deep channel. The low beach confines the channel between the low-water channel and typical channel. The high beach divides the typical channel and flood plain. Table 2 compares the proportional compositions of the total erosion/deposition of the flood plain. The components include the low-water channel, low beach, and high beach. The analysis indicates:

(1) Before water was stored in the TGR, the low-water channels in the Yichang-Zhicheng reach and Upper Jingjiang reach were eroding. The high and low beaches were undergoing slight deposition. The low-water channel in the Lower Jingjiang reach and the Chenglingji-Hankou and Hankou-Hukou reaches of the Yangtze River were eroding. The high and low beaches were experiencing heavy siltation. This indicates “deepening channel and thickening beaches” evolution.

(2) From 2003 to 2008, the low-water channel and high and low beaches of the Yichang-Zhicheng reach and the Upper and Lower Jingjiang reaches were eroding. The Chenglingji-Hankou reach was undergoing the process of “deepening channel and thickening beaches,” just as it was before water was stored in the TGR. The deposition was concentrated in the high beach. In the Hankou-Hukou reach, the low-water channel and low beach were eroding, while the high beach was experiencing slight deposition.

(3) Based on the comparison of 2008 to 2015 with 2003 to 2008, erosion was mainly concentrated in the low-water channel during both periods in the Yichang-Zhicheng and Jingjiang reaches. Beach erosion was less intense or slightly negative. In the Chenglingji-Hankou reach, erosion was concentrated in the low-water channel. The low beach changed from siltation to erosion over time and the rate of siltation in the high beach decreased. In the Hankou-Hukou reach, erosion was concentrated in the low-water channel.

The low beach changed from erosion to deposition over time. There was continuous deposition in the high beach zone. Thus, there was a “channel erosion and thickening beaches” evolution.

An examination of the depth of the channel in the 410 km section of the river downstream of the TGR in October 2014 shows, on average, 1.50 m of undercutting compared with October 2002. Further downstream, alternating siltation and scouring occurred at different locations over the same period (Figure 7a). On average, the low water level of the waterway in the 240 km downstream of the TGR in 2003 to 2014 was 1.10 m lower than between 1981 and 2002. Further downstream, the low water levels rose on average. Undercutting was concentrated in the Yichang-Zhicheng reach and Upper and Lower Jingjiang reaches and the low water level decreases mainly occurred in the Yichang-Zhicheng and Upper Jingjiang reaches (Figure 7b). In the immediate future, the undercutting in the Lower Jingjiang reach should be controlled to prevent low water levels from dropping.

The analysis of dry-season water levels (measured at a flow rate of Q = 5600 m³/s at Yichang Station) in 2003, 2008, and 2014 indicates that the low water levels between Yichang and Mopanxi and between Chenerkou and Hankou generally dropped more between 2008 and 2014 than between 2003 and 2008. At the same time, the low water levels in the Yunchi to Zhicheng section dropped less between 2008 and 2014 than between 2003 and 2008 (Figure 8). Along the course of the river, the magnitude of the decrease in the low water levels first increases and then decreases. At Zhicheng Station, the magnitude of the decrease in the low water level hits a local minimum and the pattern repeats itself. This indicates that the Zhicheng reach had a node control effect on maintaining the upstream and downstream low water levels. Downstream of Zhicheng, the water levels dropped more and more rapidly. This trend tended to migrate upstream. The upstream effects of water level drops caused by erosion of the low-water channel downstream of Zhicheng Station should be carefully monitored.

Levels for various reach of the river and siltation/scouring of the channel are well correlated. The low water levels decrease as cumulative erosion increases (Figure 9). In the Yichang-Zhicheng and Upper Jingjiang reaches, the low water level drops less rapidly as the overall quantity of erosion increases, while in the Lower Jingjiang and Chenglingji-Hankou and Hankou-Hukou reaches, the low water levels drop more rapidly as the overall quantity of erosion increases. As the intensity of scouring increases along the Lower Jingjiang reach and downstream of Hukou, there is the possibility that the fixed flow rate of the low water levels will continue to drop.

Since water has been stored in the TGR, the depth of the waterway has mainly been controlled by erosion. To prevent adverse effects, such as shoreline collapse, beach and shoal atrophy, and reduction of flow in the main waterway during the dry season, the Yangtze River Waterway Bureau carried out systematic waterway remediation work in the middle and lower reaches of the Yangtze River from 2003 to 2015. As a result, the depth of the waterway increased by 0.50 to 0.60 m in the Yichang-Chenglingji, Chenglingji-Wuhan, and Wuhan-Anqing reaches, from 2.9 m, 3.2 m, and 4.0 m (2003) to 3.5 m, 3.7 m, and 4.5 m (2015), respectively (Figure 10). The width of the waterway also increased significantly. Thus, in 2015, the dimensions of the waterway had already increased to the target dimensions for 2020 (Cao et al., 2010; Maren et al., 2013; Yang et al., 2017a, 2017b).

Siltation occurred in the Dongting Lake region from 1960 to 2006. The overall balance shifted to scouring from 2007 to 2015. In the Poyang Lake region, alternating siltation and scouring occurred from 1960 to 1999. From 2000 to 2015, a scouring trend was observed (Figure 10). Both lakes receive sediment scoured from rivers (Zhu et al., 2015). The scouring in the lake regions increased after water was stored in the TGR. This increased the lakes’ adjustable storage capacity to some extent. Scouring can reduce the lakes’ contribution to flooding of the main stream of the river and the flood control pressure on the main stream. The flow rates at the entrances to the river control the influence of the two lakes on the flooding of the main stream. They also determine if flooding in the middle and lower reaches of the river consists of upstream flood water or a combination of lake water and upstream flood water. However, the rates are not the main determinant of the fixed flow rate flood water levels.

Considering the characteristics of the river itself, the boundary obstructions downstream restrict the release of flood water from the river (Zhang et al., 2009). At flow rates of Q ≥ 50,000 m³/s in the Wuhan-Jiujiang reach, a bottleneck at Tianjiazhen delays the release of flood water by 2-3 days (Shi et al., 2007). It can also cause increased flood water levels upstream. Before water was stored in the TGR, the overall siltation of the main stream of the Yangtze River was the main driver of rising flood water levels at fixed flow rates (Yu et al., 2005; Tang et al., 2010). After water was stored in the TGR, the Yangtze River channel downstream of the TGR began to erode, especially the low-water channel (Figure 6 and Table 2).

Theoretically, this should not only cause decreases in low water levels but also in flood water levels. Scouring of the river channel seems inconsistent with rising flood water levels. Research about the causes of rising flood water levels at fixed flow rates should include changes in the hydraulic resistance and morphology of the riverbed caused by scouring and the effects of human activities.

After water was stored in the TGR, large-scale scouring and undercutting of the riverbed downstream occurred, but the flood water levels did not decrease as predicted. The adjustment of the water level in the reservoir can reduce the maximum flow rate and help ensure flood control. Soon, the TGR will begin an adjustment schedule aimed at reducing maximum flow rates in the flood season (Zheng 2015). Increasing flood water levels at the medium flood flow rate and the causes of these increases are the main points of concern. This section analyses the rising flood water levels observed at medium flood flow rates since the time at which water has first been stored in the TGR. First, the direct and indirect influences of a rougher riverbed and human activities, such as bank vegetation and port/channel morphology regulation, are analysed.

6.2.1 Effects of a rougher riverbed

A rougher riverbed increases a river’s hydraulic resistance and raises water levels. After water was stored in the TGR, the downstream riverbed became rougher (Yang et al., 2016; Zhang et al., 2017). The median grain size D₅₀ of the surface of the riverbed between Yichang and Zhicheng increased by a factor of 48, from 0.638 mm in December 2003 to 30.4 mm in October 2010. Between Zhicheng and Dabujie (Yangjianao), the median grain size D₅₀ increased by a factor of 20. Han (2014), Yang et al. (2016), and Zhang et al. (2017) calculated increases of 91% for the coarseness values of the riverbed downstream of the TGR, after water was stored (0-61 km, Yichang-Zhicheng reach), 65% (61-111 km, Zhicheng-Dabujie reach), 3% (111-319 km, UJR), and 2% (319-865 km, Chenglingji-Hukou reach; Table 3). Mathematical model calculations show the following (Han, 2014): In the Yichang-Dabujie reach of the river with pebble and gravel, the water levels corresponding to Yichang Station flow rates of 5000, 10,000, 23,000, and 35,000 m³/s increased by an average of 1.57, 2.04, 2.7, and 3.3 m, respectively, after coarsening of the riverbed. This is more than the actual drop in water level due to the increased channel depth corresponding to each flow rate increase. This explains how the coarsening of the riverbed effectively mitigated low water level drops in this pebble and gravel section of the river. In the sandy reaches downstream of Dabujie, the average water level rises corresponding to Yichang flow rates of 5000, 10,000, 23,000, and 35,000 m³/s were 0.13, 0.11, 0.16, and 0.16 m, respectively. The increases were limited relative to the effects of undercutting.

6.2.2 The effects of beach vegetation on flood water levels

Beach vegetation can help to prevent erosion of riverbanks and maintain riverbed stability (Chen et al., 2012). It also increases the flow resistance and decreases the speed of the flow. This increases the water level and affects flood prevention to a certain extent (Heidi and Nepf, 2012). The Mississippi River in the United States flooded in 2011. The flow rates during that flood were less than those in the floods of 1928 and 1973 (Day et al., 2016). Lush vegetation at higher elevations (corresponding to high beach areas) caused backwater effects that further increased the flood water level (Carle et al., 2015). This led to localized “medium flood flow rate and high flood water level” conditions. Figure 10 compares the number of days per year during which the flow rate exceeded 30,000 m³/s between 2009 and 2015 and in 2016 after water was stored in the TGR with that from 2003 to 2008. From 2009 to 2015, the number slightly decreased. In 2016, the number significantly increased in Luoshan and further downstream (Figure 11). River management and waterway remediation projects in the middle and lower reaches of the Yangtze River included large-scale ecological engineering projects to protect riverbanks and slopes. The riverbank vegetation increased the sidewall resistance of the flow. In 2016, the middle (Luoshan) and lower reaches of the Yangtze River were flooded. The number of days during which the river was in a flood stage increased. There were fewer days when the river overflowed its banks in the preceding average or dry years; therefore, the vegetation in the flood plain became relatively lush. This increased the degree of obstruction of flood water and affected the river’s ability to release flood water. Under the combined effect of the TGR peak flood reduction scheme and Dongting Lake’s three flood diversion channels, the river almost never entered the flood stage in the Yichang-Chenglingji reach. This ensured flood security in the Jingjiang reach. The flow rate in the Hankou reach was high in 2016 due to flows from the tributaries of the Daohe, Juhe, Bahe, and Xihe rivers. These tributaries caused the flow rate in the main stream to rise by as much as 24,800 m³/s (measured on 30 June 2016, as the difference between the flow rates at the Jiujiang and Hankou Stations). This partially explains why the flood water level at Hankou was high in 2016.

6.2.3 Effects of human activities on flood water levels downstream of the TGR

The Jingjiang reach in the middle reaches of the Yangtze River is a major flood control region. It is said that the Jingjiang reach is the most dangerous part of the river. There are ports, wharfs, bridges, and scenic works along the banks of the middle and lower reaches of the Yangtze River. These occupy portions of the flood control zone and narrow the flood channel. Single projects have limited effects on the overall flood control. However, the combined effect of all bridges and wharfs has a significant impact on flood water levels and adversely affects the flood control in the channel (Zhang et al., 2011). Since the major floods of 1998, the Yangtze River Water Resources Commission has been strengthening and elevating the levees along the middle and lower reaches of the river to decrease the risk to the embankments associated with rising flood waters at given flow rates.

The middle and lower reaches of the Yangtze River are known as “golden waterway.” From 2003 to 2015, the Yangtze River Waterway Bureau carried out beach remediation projects aimed at sections of beach that interfered with shipping. The work in gravel and pebble sections included bottom protection aimed at stabilizing the channel to prevent further erosion from lowering the water level in the waterway. In the sandy reaches at Shashi and further downstream, the aim was to protect and manage the beaches and sandbars. The coastline was also fortified or protected to maintain the stability of waterway boundaries and to increase the hydraulic resistance. A June 2015 report entitled “Research on the Effects of Increasing Shipping Standards on River Management and Flood Prevention in the Yichang-Anqing Reach of the Yangtze River” showed that ten sections of beach had a maximum hydraulic resistance of 5%-12.2% after remediation of 29 shipping obstructions and 19 sections of beach had a hydraulic resistance of 0%-5%. In seven sections of beach, the maximum flood water levels were elevated by 5-12.4 cm and in 22 sections of beach, the maximum flood water levels were elevated by 1-5 cm (CRSRI, 2015). Waterway remediation projects mainly involved the remediation of the low-water channel. The effect on flood water levels was limited. The impact of the channel remediation project on flood water levels was minimized during the optimization portion of the planning process.

The city of Wuhan is a focal point of flood control in the middle reaches of the Yangtze River. Between 2003 and 2016, the water level at Hankou Station exceeded the flood warning level only in 2010 and 2016. The water level exceeded the warning level only by 1 cm in 2010. On 7 July 2016, the water level exceeded the warning level by 107 cm. This was the 5^th highest water level recorded since 1870. Figure 12a shows the calculated flow rates corresponding to flood water warning levels in different years at the Yichang, Luoshan, and Hankou Stations. Overall, the flow rates corresponding to flood warning water levels have gradually decreased. The rates had decreased by 10,600 m³/s, 10,700 m³/s, and 8500 m³/s, respectively, in 2016 compared with 1998. The 2016 rates are also lower than the 2003 rates (Figure 12b). The Yichang Station is close to the TGR. Due to the peak flood clipping effect of the reservoir, the maximum flow rate and number of days of flood warnings per year decreased significantly. There were no flood warnings in many years. The flood control situation has been greatly eased. Before water was stored in the TGR, siltation was the main cause of rising flood water levels at the Luoshan Station (Zheng, 2015). Since water has been stored in the TGR, the Luoshan-Hankou reach has been scoured. However, there has been a clear backwater effect caused by increased riverbed and sidewall resistance. These are the reasons for the rising flood water levels at the Luoshan Station since water has first been stored in the TGR. Severe localized flooding occurred in the middle and lower reaches of the Yangtze River in July 2016. Combined adjustments made in the TGR and reservoirs upstream blocked 227 billion m³ of flood water. This lowered the water levels of the Jingjiang reach,

Chenglingji region, and Wuhan and downstream reaches by 0.8-1.7 m, 0.7-1.3 m, and 0.2-0.4 m, respectively. It decreased the length of the channel under flood warning by 250 km, effectively decreased the flood control pressure on the Chenglingji section and Dongting Lake region in the middle reaches of the Yangtze River, prevented the Jingjiang reach from exceeding warning levels and flooding of the Chenglingji region, protected the lives and safety of the people living in the Jingjiang reach, and guaranteed the security of the main stream levees and important infrastructure along the Yangtze River.

According to the rules for adjusting the water level of the TGR, the reservoir is intended to act as a medium to small capacity flood regulator and assist in the regulation of the flow rate (Zheng, 2015). It effectively eases the flooding situation in the Yichang-Chenglingji reach. After upgrading the flood control standard of the levees of the Jingjiang reach, the flooding pressure in this section was greatly reduced. The river reach downstream of Luoshan is greatly affected by flood water from lakes and tributaries. Under the combined effect of the main stream, lakes, tributaries, and rain, the river reach downstream of Luoshan is still a focal point for flood control. After observing the effects of medium- and small-scale flooding, it can be predicted that extremely large flood events at the TGR will decrease its flood control capacity (Li et al., 2009). It is still difficult to determine if the currently observed “high flood water levels with medium flood flow rates” in the middle and lower reaches of the Yangtze River represent a long-term trend or short-term adjustment, mainly because random factors affecting the flow rate and water levels were not accounted for in the study. Since water has been stored in the TGR, large-scale flooding of the Yangtze River, such as in 1954 and 1988, has not recurred. However, the flow rates corresponding to flood warning water levels have decreased. Medium flood flow rates are now capable of producing high flood water levels. Flood control managers should therefore take note.

The TGR has been in operation for 13 years. Experimental storage of 175 m of water has already lasted seven years. This has had profound effects on the water levels and channel morphology downstream. After examining water level and channel morphology data from the last 60 years, the major conclusions in this study are:

(1) The low water level for a given flow rate downstream of the TGR is decreasing, but flood water levels associated with medium flood flows are increasing. Water released from the TGR during the dry season has raised the minimum water levels downstream and peak flood adjustments in the TGR have lowered the maximum flood water levels downstream.

(2) The low water level for a given flow rate has dropped. The superimposition of the effects of waterway remediation projects and scouring and undercutting downstream of the TGR shows that the absolute low water level for a given flow rate has dropped less than the absolute altitude of the riverbed, indicating that the waterway has deepened.

(3) Scouring is the main factor affecting low water level decrease. In the Upper Jingjiang reach and further upstream, the decreases in low water levels are stabilizing. In the Lower Jingjiang reach and further downstream, the decreases in low water levels are accelerating. Waterway managers should take note.

(4) Since water has been stored in the TGR, the combined effects of roughening of the riverbed, vegetation cultivation on the riverbanks, and human activities, have caused fixed flow rate flood water levels to rise. At the same time, the maximum flow reduction in the TGR during the flood season has significantly decreased the number of flood warning days per year in the Yichang-Jianli reach of the river, often to zero, greatly easing the flood control situation. The medium flood flow rates are causing high flood water levels. This is a cause of concern. The TGR can effectively improve flood mitigation in the Yangtze River near the dam. It cannot control flooding caused by tributaries further downstream. Therefore, the risk of flooding in the lower reaches of the Yangtze River downstream of the Luoshan Station is still rather high.