Discussion on the Dominant Factors Affecting the Main-Channel Morphological Evolution in the Wandering Reach of the Yellow River

Abstract

The wandering reach of the Yellow River has long been a pivotal area of research due to its drastic fluctuations in water-sediment dynamics, frequent shifts in the main channel, and complex river regime evolution. Studies on the main-channel morphological evolution in this reach have focused on the analysis of parameters related to the overall oscillation or have only analyzed a certain reach within the wandering reach, with a lack of detailed studies based on the different characteristics of each area. Therefore, taking the Xiaolangdi Reservoir–Gaocun reach as the research area, by constructing a two-dimensional water-sediment dynamic model, the erosion–deposition characteristics of different sub-reaches and the morphological evolution characteristics of key cross-sections were quantified and analyzed. Based on measured hydrological, sediment, and topographic data, the temporal and spatial changes in the bankfull area and fluvial facies coefficient of typical sections before and after the construction of Xiaolangdi Reservoir were analyzed. By interpreting remote sensing images, the spatio-temporal variation characteristics of the migration distance and bending coefficient of different reaches before and after the construction of Xiaolangdi Reservoir were calculated, and the key factors influencing the evolution of river morphology parameters were identified. The results showed that after the Xiaolangdi Reservoir operation, the overall erosion of the Huayuankou–Jiahetan reach is greater than the deposition, and the erosion is more obvious in dry years. The river course direction and control engineering play a significant role in controlling the morphological evolution of the main channel during the process, causing the R2 reach to significantly swing to the north bank and the R3 reach to the south bank. When the sediment transport coefficient values were between 0 and 0.005 kg.s.m⁻⁶, water-sediment had a positive effect on shaping and evolving the main-channel morphology. The long-term low-sand discharge of Xiaolangdi Reservoir and the continuous improvement of river regulation projects are the main reasons for the above changes. The results can provide support for controlling the evolution of the main channel and improving river regulation projects.

1. Introduction

The evolution of main-channel morphology is a dynamic adjustment process through which river systems respond to the dual drivers of natural and anthropogenic factors. Research findings show that the influence of anthropogenic factors on runoff and sediment transport in many rivers exceeds natural factors [1,2]. Damming, as one of the most significant anthropogenic factor disturbances to rivers, has played positive roles in water–sediment regulation, flood control, hydropower generation, and ecological improvement. However, it simultaneously cuts off river continuity, alters natural water–sediment characteristics [3,4], and affects the natural evolutionary trajectory of the river [5]. Some representative examples include the Missouri River in the United States [6] and China’s Yangtze River [7,8] and Yellow River [9,10]. Therefore, research on the impacts of damming on channel morphological evolution is crucial for deepening our understanding of river morphology evolution characteristics and guiding river regulation strategies.

As the most complex river pattern unit in river channel morphology evolution, wandering reach exhibits disordered morphology, frequent migration, and silt riverbeds [11,12]. Many studies, from riverbed erosion and incision [13,14], channel morphology reconfiguration [15,16], riverbank stability change [17,18], sandbank and branch channel evolution [19,20], and local geomorphic response [21,22], have been carried out on the impacts of damming on the channel morphology evolution in this reach. As Ma et al. discovered, there are seven types of distribution patterns of erosion and deposition along the course of the Yellow River’s lower reaches [23]. Among them, the wandering reach is characterized by an upward erosion and downward deposition pattern. Yan et al. found that after the Xiaolangdi Reservoir (XLDR) operation, the main stream amplitude gradually decreased, and the wandering nature weakened year by year. However, the Huayuankou (HYK)–Jiahetan (JHT) reach still exhibited a certain wandering pattern [24]. Jia et al. analyzed the current situation of the bank slope damage, the mechanism of bank instability, and its main influencing factors of the lower reaches of the Yellow River and discussed and sorted out some research directions for ecological slope protection in the lower reaches of the Yellow River in the future [25]. Kong et al. discovered that dam construction effectively reduced the flow during large flow events [26]. Sandbars were difficult to erode by frequent small flows, which made them gradually stabilize and be covered by vegetation. The main channel would gradually become blocked by the uneven distribution of flow by the branch channels, and some parts would become silted-up, forming a single river course. The above studies have examined the impact of dam construction on main-channel evolution in the wandering reach from multiple aspects and perspectives, but they mainly focus on a certain aspect. The influence on retrogressive deposition and progressive deposition cannot be distinguished. The research on scouring and oscillation still mainly relies on data analysis. The principle understanding of the migration patterns, response mechanisms, and future evolution trends of the wandering reach are lacking. The identification of the key influencing factors still requires further research based on the characteristics of water-sediment changes.

The Yellow River is well known for its large amount of sediment. This large river has the largest sediment transport and the highest sediment content in the world [27]. After flowing through the Xiaolangdi Gorge into the middle and lower reaches, a typical wandering reach is formed between the XLDR and Gaocun (GC). This reach experiences a sudden drop in riverbed gradient and decreased flow velocity. As the current wanders across the floodplain area, most sediment is deposited in the lower reaches [28,29]. Since the XLDR operation in 2000, the Yellow River’s lower reaches have experienced significantly increased runoff and substantially reduced sediment transport, which leads to fundamental changes in water–sediment flux [30,31]. These hydrological alterations have profoundly impacted the erosion–deposition evolution of the wandering channel section [32,33], resulting in new sediment movement phenomena and the evolution of the downstream riverbed [34,35]. Some scholars have studied the evolution law of wandering riverbeds under changing environments from the perspective of macro-statistics [36,37]. The adjustment process of riverbed section shapes with a change in water-sediment has also been simulated and analyzed by establishing a quantitative relationship between the section shape variable and water-sediment factors [38,39]. However, the current research on channel evolution mainly focuses on the analysis of alluvial river channel morphology under equilibrium or quasi-equilibrium conditions, which is not applicable to non-equilibrium rivers that have been drastically adjusted under the influence of strong human activities. Meanwhile, research mainly focuses on the analysis of the overall swing parameters or only analyzes a certain part of the wandering reach due to a lack of data. Detailed studies of the sub-regional system according to the different characteristics of each region of the wandering reach are lacking. Accordingly, this study took the wandering reach of the Yellow River as the research object, and the erosion–accumulation process and migration oscillation pattern along the main channel were analyzed by dividing into short river reaches. The key factors that affect erosion–accumulation and main-channel characteristic parameter changes were identified by combining the morphological change characteristics of typical sections. The response relationship between characteristic parameters with water-sediment variations were established, and the threshold values of the key factors were explored. The results are conducive to comprehensively understanding the morphological evolution law of the wandering reach channel and provide scientific guidance for river regulation.

2. Overview of the Study Area

The reach from XLDR to GC in the middle and lower Yellow River is a typical wandering reach, approximately 310 km. This reach has a wide and shallow riverbed with dense dikes along both banks. The distance between the dikes ranges from 5 to 15 km, extending up to 20 km at the widest sections. The main channel varies between 1 and 3.5 km in width. Floodplain areas account for over 80% of the total channel area [40]. Four hydrological stations were established in the study area, named XLDR, HYK, JHT, and GC (Figure 1). Based on the wandering characteristics, the change characteristics of the river situation, and the river pattern, the study area is divided into four reaches: R1 (XLDR–Taohuayu (THY)), R2 (THY–Heigangkou (HGK)), R3 (HGK–JHT), and R4 (JHT–GC). The changes in parameters such as the main-current sway and the bending coefficient are analyzed. In the R1 reach, there are many high cliffs on both banks, and the water flow is relatively turbulent in the first half of the section flowing through the canyon, while it slows down after exiting the canyon in the second half, and the river situation is chaotic. In the R2 reach, the riverbed is wide and shallow, and it belongs to a micro-erosion zone, which is the most significant wandering area in the study area. The main channel is significantly constrained by the south bank embankment, and lots of control projects have been built on the north bank. The R3 reach belongs to a micro-sedimentation area, with a high channel and a low beach. The riverbank is easily brushed, and control projects are built at the bending points of the main channel to control the erosion of the riverbank. In the R4 reach, it becomes relatively straight after turning, and it belongs to a sedimentation area. There are many low beaches, and the river situation is relatively stable.

3. Materials and Methods

The morphological evolution of the river channel is closely linked to variations in water-sediment. Following the XLDR operation, the annual runoff and sediment discharge in the wandering reach have undergone considerable changes. Runoff initially decreased and then increased, whereas sediment discharge decreased dramatically (Figure 2b). Since the XLDR began impoundment in October 1999, the majority of medium- grained and coarse-grained sediments have been trapped within the reservoir, drastically reducing the sediment supply to the lower Yellow River. The release of clear water has triggered sustained scouring of the downstream channel, leading to considerable adjustments in riverbed morphology. The adjustment of riverbed morphology is also influenced by the main channel cross-section type. The main channel cross-section of the wandering reach is mostly composed of river channels and multiple terraces, with a large covering area. The marshlands are generally categorized into three tiers: high floodplain (gao tan), low floodplain (di tan), and young floodplain (nen tan) (Figure 2a) [41]. The high floodplain formed over a long geological history, is highly stable, and is rarely inundated. Its stability depends on the elevation difference between the floodplain and channel. Compared to the high floodplain, the low floodplain features a lower elevation and poorer stability. The young floodplain refers to areas that are frequently submerged during flood seasons but are exposed during dry periods. These zones are highly unstable and undergo continuous morphological changes [42]. The R1 reach is mostly constrained by high cliffs, and after a long period of evolution, the main channel mostly approaches the side of the high cliffs, with limited scouring and oscillation range. The R2 and R3 reaches have narrow and shallow riverbeds and have low terraces and young terraces, which are prone to scouring and oscillation, and the evolution of the main channel is mainly affected by the dike and control engineering. The R4 reach has a narrowed and relatively straight river channel, with a small scouring and oscillation amplitude. Therefore, different river reaches should be divided for analysis of erosion and oscillation.

3.1. Research Approach

• Based on the measured hydrological, sedimentary, and topographical data, a two-dimensional water–sediment dynamic model was constructed to simulate the changes in water-sediment in the study area over time and space and to obtain the hydrological and sedimentary data of any area. Using the water balance principle, the sedimentation and erosion amounts of different river sections were calculated, and the evolution laws of the main channel of typical sections were analyzed.
• By comprehensively selecting the parameters of river channel morphology, such as the bankfull area (A_bf) and the fluvial facies coefficient (ε), as well as the evolution parameters of the main channel, such as the migration distance and the bending coefficient (K_a), the evolution characteristics of the main-channel morphology in the wandering reach over a long time series were analyzed.
• Using the correlation analysis method, the relationships between the river morphology parameters and annual runoff volume, annual sediment discharge, flood season runoff volume, flood season sediment discharge, and sediment contribution coefficient were analyzed. Through multiple linear regression analysis, significant factors were further identified.

3.2. Method for Calculating Sedimentation and Erosion of Water Flow

The study area has six hydrological stations: XLDR, HYK, JHT, GC, Wuzhi (WZ), and Heishiguan (HSG). The WZ and HSG hydrological stations are monitoring stations for the Qin River and the Yiluo River, which are tributaries that flow into the Yellow River. The distance between each hydrological station is 80–128 km. Based on the hydrological monitoring data, the overall situation of sedimentation and erosion of the entire river section can be evaluated; however, local cross-sectional topographic changes cannot be described. To calculate the sedimentation and erosion volumes over a short distance and to understand the sedimentation and erosion characteristics, it is necessary to obtain the water-sediment data at any cross-section in the study area as accurately as possible. To obtain the water-sediment data of the entire study area, the Danish Water Institute developed the professional engineering software package MIKE 21h (2014), which was used to construct a two-dimensional water-sediment dynamic model from XLDR to JHT. This model consists of a water dynamics module and a sediment transport module. The water dynamics module adopts the Navier–Stokes equations based on three-dimensional incompressibility and the Reynolds mean value. Additionally, it applies the two-dimensional non-stationary shallow water equations under the assumptions of hydrostatic pressure and Boussinesq. The sediment module is divided into suspended and bedload sediments. The suspended sediment adopts the sediment transport theory proposed by Galappatti (1985) [43], whereas the bedload sediment adopts the Engelund–Hansen equation. The hydrodynamic module is divided into 194,079 grids and 98,691 nodes. Each grid has an area of 11,000 m². The main parameters include the dry–wet zone, riverbed roughness, and eddy viscosity coefficient. The parameter values are 0.005 m for dry water depth, 0.05 m for submerged water depth, 0.1 m for wet water depth, a Manning number of 47 m^1/3/s, and an eddy viscosity coefficient of 0.28. The parameters of the sediment module mainly include the settling velocity coefficient, critical shear stress, and riverbed erosion coefficient, and differences in the flood and non-flood seasons are considered. The settling velocity coefficient is 2 m/s and the erosion coefficient is 3 × 10⁻⁵ kg/m²/s in both seasons, and the critical shear stress is 0.22 N/m² and 0.2 N/m², whereas the riverbed erosion coefficient is 3.5 and 4.3, respectively.

Using the daily measured flow and sediment concentration data from XLDR, WZ, and HSG, as well as the daily measured water level and sediment concentration data from JHT in 2019 as inputs, the simulation results were verified with the daily measured water level, flow, and sediment concentration data from HYK in 2019. The average relative errors of the water level, flow, flood-season sediment concentration, and non-flood-season sediment concentration were 0.15%, 10.87%, 39.58%, and 18.22% respectively. Overall, the error in the model calculation results was small, and the simulation accuracy met the requirements. The verification results of water level, flow, and sediment concentration at the HYK section are shown in Figure 3.

Based on the water balance principle, the sedimentation and erosion amount of a certain reach is calculated by subtracting the sediment discharge at the upper boundary section from that at the lower boundary section. The difference represents the sedimentation and erosion conditions in that section. The specific calculation formula is as follows:

Q_s = (Q₂ × S₂ – Q₁ × S₁) × t/1000

(1)

where Q_s represents the sediment discharge (t); Q₁ and Q₂ are the flow rates at the upper and lower boundaries of a certain section, respectively (m³/s); S₁ and S₂ are the sediment concentrations at the upper and lower boundaries of the same section, respectively (kg/m³); and t represents time (s).

3.3. Calculation of River Main-Channel Characteristic Parameters

The characteristic parameters of the main channel mainly include the channel morphology characteristic parameters and main-channel evolution characteristic parameters. The values of A_bf and ε of the channel morphology characteristic parameters were selected to reflect the section morphology characteristics. The values were calculated from the measured data. The migration distance and K_a were selected as evolution characteristic parameters, which were used to analyze the evolution trend of the main channel with the change in water–sediment. The values were extracted and calculated from remote sensing image interpretation. The calculation methods and significances of each parameter are shown in

Table 1. Calculation methods and significances of river main channel characteristic parameters.

Characteristic Parameter	Calculation Method	Parameter Significance	Unit
Abf	-	The section area where the water level of the section reaches the level of the flood plain. It has a corresponding relationship with the floodplain water level and the river width.	m2
ε	$\epsilon =\sqrt{B}/H$	Comprehensive index to characterize riverbed morphology, which is related to water depth and river width under bankfull water level.	m·0.5
Migration distance	$\overline{{{\displaystyle S}}_L}={\displaystyle \frac{{\displaystyle \sum _{i=1}^n{{\displaystyle S}}_{il}}}{n}}(i=1,2,3~n)$ $\overline{{{\displaystyle S}}_R}={\displaystyle \frac{{\displaystyle \sum _{i=1}^n{{\displaystyle S}}_{ir}}}{n}}(i=1,2,3~n)$	ArcGIS 10.8 is used to convert each river reach in 1986 and 2019 into centroid points. The direction and distance between centroid points are the migration direction and distance of the main channel. The migration of centroid points to the left bank is positive, and the migration to the right bank is negative.	m
Ka	${{\displaystyle K}}_a=L/I$	The ratio of the actual length to the straight-line length. The dividing line is generally 1.3, and when it is greater than 1.3, the river is usually a wandering river.	-

Note: B and H are the water surface width and average water depth under the bankfull water level, m; S_i, $\overline{S}$, and S_max are the swing area, average swing area, and maximum swing area of the main channel in each section, km²; S_il and S_ir are the migration distance to the left and right bank of the centroid point in each reach, m; and $\overline{S_L}$ and $\overline{S_R}$ are the average migration distances to the left and right bank of the centroid point in each section, m.

.

3.4. Data Acquisition and Statistical Analysis

The data used in this study included remote sensing data, measured topographic data of the cross-sections, and monitoring data from the hydrological stations. The remote sensing images from 1986, 1989, 1992, 1995, 1998, 2001, 2004, 2006, 2010, 2013, 2017, and 2019 were obtained via the “Geographic Spatial Data Cloud” of the Chinese Academy of Sciences by using the US Landsat 5 TM and Landsat 8 OLI satellites. Due to the long distance of the study area, a single image cannot cover the entire area. Therefore, multiple images needed to be combined. We selected remote sensing images with cloud cover of less than 30% during the period from October to December each year. ENVI 5.3 was used for data preprocessing through geometric correction, radiation correction, and atmospheric correction to eliminate various influences such as atmospheric, water vapor, and radiation. Then, the images were tailored and mosaicked to obtain a complete wandering river reach, and the MNDWI method was used to extract water information from the remote sensing images after concatenation.

The MNDWI index is calculated as follows:

MNDVI = (Green − SWIR)/(Green + SWIR)

(2)

where Green is the green light band and SWIR is a short infrared band corresponding to bands 2 and 5 of Landsat 5 TM and bands 3 and 6 of Landsat 8 OLI.

Topographic and hydro-sedimentary data were obtained from the Hydrological Yearbook of the Yellow River Basin and the Yellow River Water Conservancy Commission. The topographic data from 1960–2019 included all measured topographic data of the large sections set up by the Yellow River Water Conservancy Commission. The hydro-sedimentary data from 1960–2019 was the measured data of the six hydrological stations. Pearson’s correlation analysis was used to explore the correlation between the hydro-sedimentary factors and the main-channel characteristic parameters, and multiple linear regression analysis was used to test the significance of the correlation factors. Data preprocessing was carried out using the Excel 2019 software. Statistical analysis and plotting were conducted using Origin 2024 (OriginLab Co., One Roundhouse Plaza, Suite 303, Northampton, MA, USA), and the data after remote sensing interpretation were analyzed using ArcGIS 10.8.

4. Results

4.1. Analysis of Water-Sediment Erosion and Deposition Changes in the Study Area

The HYK–JHT reach is the most severely eroded and silted reach, and it is most severe in the overall formation and development of the suspended river. To analyze the erosion and siltation situation of this reach, the HYK–JHT reach was divided into 11 shorter reaches of approximately 10 km by Shuangkai (SK), Laitongzhai (LTZ), Zhaokou (ZK), Yueshi (YS), Doumen (DM), Yudian (YD), Liuyuankou (LYK), Peilou (PL), Caogang (CG), and Zhuzhai (ZZ) (Figure 1). The daily simulation results of the two-dimensional water-sediment dynamic model were used to extract the flow and sediment concentration data of the 10 cross-sections, and the sediment transport volumes of the different cross-sections were calculated. Using Formula (1), the erosion and siltation conditions of the different shorter reaches were calculated as examples for the median water year (2015, 50%) and the dry year (2016, 75%). The results are shown in Figure 4.

From the above figure, the HYK–JHT reach experienced erosion during the median water year and the dry year. Moreover, the erosion volume in the dry year (1.02 × 10⁸ t) was greater than that in the median water year (0.76 × 10⁸ t), while the deposition volume was not different (0.59 × 10⁸ t and 0.50 × 10⁸ t, respectively). In the median water year, the total length of deposition in the HYK–JHT reach was approximately 38.56 km, accounting for 38.44% of the total length. The total deposition volume was approximately 0.50 × 10⁸ t, mainly distributed in the ZK–YS and DM–YD reaches, with the deposition volumes of these two reaches being 0.19 × 10⁸ t and 0.25 × 10⁸ t, accounting for 38.00% and 50.00% of the total deposition volume, respectively. The total erosion length was approximately 52.07 km, accounting for 51.91% of the total length, and the total erosion volume was approximately 0.76 × 10⁸ t, with a relatively scattered erosion distribution. The HYK–SK, SK–LTZ, YS–DM, YD–LYK, CG–ZZ, and ZZ–JHT reaches experienced erosion. The erosion volume in the ZZ–JHT reach was 0.30 × 10⁸ t, accounting for 39.36% of the total erosion volume. In the dry year, the erosion situation was different from that in the median water year. The HYK–SK, YS–DM, and CG–ZZ reaches, which experienced erosion during the median water year, transformed into a deposition state in the dry year, with deposition volumes of 0.06 × 10⁸ t, 0.09 × 10⁸ t, and 0.01 × 10⁸ t, respectively. Together with the LTZ–ZK, ZK–YS, DM–YD, and PL–CG reaches, the total length of deposition in the dry year was approximately 65.56 km, accounting for 65.36% of the total length, an increase of 27.00 km compared to the median water year. The total deposition volume also increased by 0.09 × 10⁸ t compared to that in the median water year. The erosion length decreased by 17.36 km compared to that in the median water year, but the total erosion volume increased by 0.26 × 10⁸ t. The deposition was mainly concentrated in the LTZ–DM reach, with a deposition volume of 0.47 × 10⁸ t, which accounted for 79.66% of the total deposition volume. Erosion mainly occurred in the SK–LTZ, YD–LYK, LYK–PL, and ZZ–JHT reaches, with a total length of approximately 34.71 km, accounting for 34.61% of the total length of this reach, and the erosion volumes were 0.46 × 10⁸ t, 0.0008 × 10⁸ t, 0.05 × 10⁸ t, and 0.51 × 10⁸ t, respectively.

Overall, the net erosion volume in the HYK–JHT reach during the median water year and the dry year was 0.26 × 10⁸ t and 0.43 × 10⁸ t, respectively. The net erosion volume in the dry year was 0.17 × 10⁸ t greater than that in the flood year.

4.2. Analysis of the Evolution Law of the Main-Channel Morphology of Typical Cross-Sections in the Different Water Years

According to the characteristics of water-sediment scouring and silting in the HYK–JHT reach, as shown in Figure 4, the reaches that simultaneously experienced scouring in the median water and dry years were SK–LTZ and ZZ–JHT, whereas the reaches that simultaneously experienced siltation were ZK–YS and PL–CG. The reaches with opposite scouring and silting conditions were HYK–SK, YS–DM, LYK–PL, and CG–ZZ. To verify and illustrate the response of the cross-sectional morphology evolution to scouring, one cross-section was selected from each of the above reaches to analyze the evolution characteristics of the cross-sectional morphology between the median water and dry years. The selected cross-sections from the corresponding cross-sections were Shiqiao (SQ), Guantai (GT), Baochang (BC), Yuanfang (YF), Zhaolanzhuang (ZLZ), Heishi (HS), Wang’an (WA), and Qingheji (QHJ) (Figure 1). The evolution of the cross-sectional morphology is shown in Figure 5.

From the above figure, all eight cross-sections show different degrees of erosion and deposition. For the two cross-sections in the short-distance river section where erosion occurred, the main channel of the SQ cross-section eroded southward overall. In 2017, the southern part moved 69.75 m southward compared to 2015 and 39.42 m compared to 2016. Meanwhile, the deep-point elevation decreased 4.50 m and 2.87 m, respectively. At the same water level, the water area in 2017 and 2016 was 16.21 m² and 2274.75 m² more than that in the previous year, respectively. For the GT cross-section, owing to the existence of the river central island, the main channel was divided into two parts. The main channel was located in the narrower area on the southern side of the river central island. The deep-point elevation variation range of the main channel is within 0.50 m. The main channel is continuously being eroded towards the south. In 2017, the south bank moved 82.81 m southward compared to 2015 and 43.56 m southward compared to 2016. At the same water level, the water area in 2017 and 2016 was 153.49 m² and −70.35 m² larger than that in the previous year, respectively. For the two cross-sections where deposition occurred in the short-distance reaches, the BC cross-section was controlled by the guiding project, and the main-channel morphology was relatively stable, but the deep point and its surrounding areas changed significantly. After being eroded in 2015, the deep-point elevation in 2016 decreased by 2.22 m compared to 2015. In 2016, there was siltation, and in 2017, the deep-point elevation increased by 0.63 m compared to 2016. At the same water level, the water area in 2017 and 2016 was −87.70 m² and 213.19 m² larger than that in the previous year, respectively. For the YF cross-section, the position of the deep point completely changed from the south to the north side of the main channel in 2016 and 2017. However, the deep point elevation did not change much, remaining at around 71.00 m. The water area in 2017 and 2016 at the same water level was 318.68 m² and −352.40 m² larger than that in the previous year, respectively. For the four cross-sections where erosion and deposition occurred in the short-distance river section, the water area at ZLZ, WA, and QHJ continuously increased, and the deep-point position also kept shifting from 2015 to 2017. The water area at ZLZ increased by 1269.67 m² in 2017 and 965.17 m² in 2016 compared to the previous year. The water area at WA increased by −286.85 m² and 636.04 m², respectively, and that at QHJ increased by 22.01 m² and 476.24 m², respectively. For the HSG section, due to its proximity to the controlled engineering area (the distance from the bank is 2850 m), the water area did not change much, but the deep-point position kept decreasing. In 2017, it decreased by 0.90 m compared to 2016 and 1.11 m compared to 2015.

4.3. Analysis of Main-Channel Characteristic Parameter Evolution Based on Long-Term Sequences

4.3.1. Evolution of Typical Cross-Section River Morphological Characteristic Parameters

Selecting HYK, JHT, and GC as the typical cross-sections, by calculating A_bf and the ε from 1960 to 2019 based on the collected measured data, the evolution laws of the river morphology characteristic parameters of different sections were analyzed. The results are shown in Figure 6.

From the above figure, before the XLDR operation, the A_bf value of the three typical cross-sections generally exhibited a declining trend. The HYK section experienced a 91.14% decrease compared to its peak value in 1996, whereas the JHT and GC sections decreased by 83.33% and 83.21%, respectively. The reason for this change is that before the XLDR operation, the sediment carried by the middle reaches continuously accumulated downstream, causing the riverbed to rise continuously and forming a secondary suspended river. Following the commissioning of XLDR, the A_bf value at these three typical cross-sections showed continuous growth. The JHT section demonstrated the most considerable increase, followed by the HYK and GC sections, with expansions of 551.16%, 380.65%, and 265.58% by 2019, respectively. After the XLDR operation, through water–sediment regulation and continuous discharge of clean water, the riverbed was constantly eroded and A_bf continuously expanded. Compared to the A_bf variations, ε exhibited more pronounced fluctuations. The JHT section displayed the most dramatic changes before and after XLDR operation. Notably, the fluctuation frequency decreased after reservoir operation compared to that in the pre-operation period. This might be due to the JHT cross-section being located near the area where the main channel changes its direction. Water-sediment have a significant impact on the main-channel width and depth, and ε changes dramatically. After the XLDR operation, the changes in ε were mainly related to the annual runoff volume. The changes in the water area and deep points of the WA, YF, and QHJ cross-sections, which belong to the R3 reach, were the same as the JHT cross-section in Figure 5, also confirming this change. The HYK and GC sections demonstrated consistent patterns of ε changes before and after XLDR operation, with oscillatory variations prior to operation transitioning to stabilization post-operation. After reservoir impoundment, the ε value at the HYK and GC sections decreased by approximately 50% compared to pre-operation levels [44], fluctuating within ranges of 24–29 and 6–8, respectively. Furthermore, the GC section exhibited a gradual decreasing trend over time. Before the XLDR operation, the water-sediment conditions at the HYK and GC cross-sections were highly variable, causing significant erosion and deposition of the riverbed. However, due to the different distances and riverbed widths, the degree of variation was not the same. After the XLDR operation, the average water surface widths at the bankfull levels over the years were not much different, being 2181.6 m and 2249.3 m, respectively, but the average water depths were quite different, being 1.46 m and 2.90 m, respectively, resulting in a higher ε at the HYK cross-section compared to the GC cross-section.

By comparing the results of A_bf and ε, when A_bf at the HYK and GC cross-sections showed increases after the year 2000, ε slightly decreased and tended to stabilize. This indicates that the average water depth under the bankfull water level varies greatly compared to the water surface width. The increase in A_bf is closely related to the increase in water depth.

4.3.2. Variations in Main-Channel Evolution Characteristic Parameters in Different Reaches

Based on the interpretation results of remote sensing images, the lateral cumulative migration distances and the K_a value of the R1, R2, R3, and R4 reaches were calculated. The results are shown in Figure 7.

From the above figure, there are differences in the cumulative migration distances along the main channels in the four reaches. R1, R3, and R4 exhibited bidirectional lateral migration (both banks) with an overall south bank tendency, and their maximum south bank migration distances reached 1450, 1221, and 1360 m, respectively, with corresponding average values of 823 m, 925 m, and 750 m, respectively. The maximum north bank migration distances were 1100, 530, and 593, averaging 784, 495, and 456 m, respectively. In contrast, the R2 reach demonstrated exclusive north bank migration, with maximum and average migration distances of 3100 m and 1304 m, respectively. Based on the cumulative migration distance statistics, the calculated cumulative swing areas toward both banks during 1986–2019 (

Table 2. The area swing to left and right banks of the main channel in each region. (km²).

Region	R1	R2	R3	R4	Total
Swing to left bank	15.4	28.6	7.9	7.4	59.9
Swing to right bank	17.9	6.2	19	10.2	53.3

) revealed disparities between the north and south bank swing areas in the R2 and R3 reaches, whereas the R1 and R4 reaches showed comparable swing magnitudes on both banks. According to Figure 1, the main channel of the R2 reach is mainly located near the southern bank and is significantly constrained by the dike. While the main channel of the R3 reach is mainly located near the northern bank, and there are numerous control and regulation projects distributed there, effectively restricting the migration of the main channel. For K_a, before XLDR operation, K_a initially decreased then increased, rising from 1.186 (R1), 1.140 (R2), 1.141 (R3), and 1.251 (R4) in 1989 to 1.433, 1.269, 1.332, and 1.465 in 2001, with increments of 20.83%, 11.32%, 16.74%, and 6.47%, respectively. After XLDR operation, the K_a value of the R1 and R4 reaches decreased gradually and showed a different trend than that of the R2 and R3 reaches. R1 and R4 decreased from 1.433 and 1.465 in 2001 to 1.327 and 1.365 in 2019, decreasing by 7.40% and 6.83%, respectively. R2 and R3 increased from 1.269 and 1.332 to 1.314 and 1.349, showing modest gains of 3.55% and 1.28%, respectively. Based on the relationship that the bay radius is directly proportional to the flow, the smaller the flow, the smaller the bay radius and the greater the bending. After the XLDR operation, the bending no longer changed significantly with the drastic variations in the flow. This indicates that under the control and regulation of the engineering, the bay radius changed slowly, and the river became straight and stable.

From the above analysis, we can see before the XLDR operation, the wandering reach exhibited disordered morphology, frequent migration, and a silt riverbed. After the XLDR operation, the downstream river gradually changed from being wide and shallow to deep and narrow. The scouring effect of the downstream water flow on the main channel gradually changed from lateral scouring to longitudinal scouring, and the closer it was to the XLDR, the more significant the scouring effect became.

4.4. Identification of Influencing Factors and Construction of Response Relationships for Main-Channel Morphological Characteristic Parameters

As demonstrated in this study, the XLDR operation has not only altered the original water–sediment regime but also modified the evolutionary trajectory of the main-channel morphology, imparting systematic regularity to these morphological adjustments. The continuous reconfiguration of the main-channel morphology results from the dynamic mutual adjustment between sediment-laden flow and the riverbed. There exist quantitative relationships between changes in channel morphological characteristic parameters and water–sediment conditions. To elucidate these relationships, a correlation analysis was conducted to quantify the relationships between key parameters (A_bf and ε) and annual runoff (W_s), annual sediment load (Q_s), flood-season runoff (W_f), flood-season sediment load (Q_f), and sediment coefficient (ξ) before and after the XLDR operation. The changes in the key influencing factors before and after the XLDR operation were analyzed. The results are shown in Figure 8.

From the above figure, the three typical cross-sections exhibit distinct correlations between A_bf, ε, and hydrological drivers, with shifts in the dominant influencing factors before and after XLDR operation. For the HYK section, A_bf showed significant positive correlations with W_s, Q_s, W_f, and Q_f, and ε correlated positively with Q_s and Q_f before XLDR operation, showing a positive correlation. ξ became the sole significant factor after XLDR operation, showing a positive correlation with A_bf (p < 0.05) and a negative correlation with ε (p < 0.01) (Figure 8a). For the JHT section, before XLDR operation, A_bf correlated positively with W_s and W_f and negatively with ξ. ε exhibited negative correlations with W_s, Q_s, and W_f. After XLDR operation, A_bf retained correlations with W_s (p < 0.05) and ξ (p < 0.01), and the correlation remained unchanged compared with that before XLDR operation, whereas no factors significantly influenced ε (Figure 8b). For the GC section, before XLDR operation, A_bf correlated positively with W_s and W_f and negatively with ξ, and no significant correlations with ε were observed. After XLDR operation, A_bf maintained correlations with W_s (p < 0.05) and ξ (p < 0.01), and the correlation remained unchanged compared with that before XLDR operation, and ε correlated negatively with W_s (p < 0.01) and W_f (p < 0.05) and positively with ξ (p < 0.01) (Figure 8c). Through multiple linear regression analysis, after XLDR operation, the significant correlation factors with A_bf and ε in the HYK section all passed the 0.05 significance test, with p-values of 0.016 and 0.006, respectively. In JHT and GC sections, ξ was the only factor that significantly correlated with A_bf and ε, passing the significance test of 0.05 with p-values of 0.000.

Based on the above analysis, the changing trend diagrams of A_bf and ε in relation to ξ of the three typical cross-sections after the XLDR operation were established (Figure 9). It can be observed that A_bf at the HYK cross-section gradually increased and then decreased as ξ increased. ε exhibited an opposite trend. The proportion of the relationship points of A_bf and ε, with ξ scattered within the range of 0–0.005 kg.s.m⁻⁶, was 85%. A_bf at the JHT cross-section gradually decreased as ξ increased. The proportion of the relationship points of A_bf and ξ scattered within the range of 0–0.005 kg.s.m⁻⁶ was 65%. At the GC cross-section, A_bf and ε exhibited opposite changing trends as ξ increased. A_bf gradually decreased with an increase in ξ, whereas ε gradually increased. For all cases, 70% of the relationship points of A_bf and ε with ξ were scattered within the range of 0–0.005 kg.s.m⁻⁶.

5. Discussion

5.1. Influence of Water-Sediment Condition Changes on the Main-Channel Morphology Evolution

The continuous adjustment of river cross-sectional geometry is shaped by the mutual adaptation between sediment-laden flow and the riverbed, with runoff playing a dominant role in channel formation, particularly during flood seasons [45,46]. Before the XLDR operation, the lower Yellow River experienced flood-season erosion and non-flood-season deposition, with overall deposition exceeding erosion. The riverbed elevations at the HYK, JHT, and GC sections all exhibited varying degrees of elevation, accompanied by continuous shifts in deep points. During this period, A_bf decreased, whereas ε fluctuated (Figure 6) [47,48]. After the XLDR operation, the annual average runoff in the lower Yellow River gradually increased, whereas the annual sediment load decreased dramatically (Figure 2b) compared to that before XLDR operation. The highly unsaturated discharged flow induced continuous channel scouring downstream of the dam, leading to considerable cross-sectional morphological changes. However, distinct adjustment patterns emerged across different cross-sections [49]. After nearly 20 years of XLDR operation, the sediment accumulated in the wandering reach has been significantly washed away, and the riverbed elevation has dropped. Under the current conditions where the river channel improvement project is relatively complete and the riverbed structure is relatively stable, the composition of water-sediment has become the key influencing factor affecting the main-channel morphology evolution.

According to Figure 8, the A_bf and ε values of the HYK, JHT, and GC cross-sections all show significant correlations with ξ. Some cross-sections also show significant correlations with the runoff volume. ξ is determined by the ratio of sediment concentration to runoff, and if only the influence of runoff on the main-channel morphology evolution is considered, it cannot objectively reflect the influence of dam construction on the main-channel morphology evolution. Comprehensively considering the overall changes in water-sediment can better reflect the key role of water-sediment in the main-channel morphology evolution. By comparing the main-channel evolutionary characteristics in 1996 and 2016 under similar water discharge conditions but with considerably different sediment regimes, the annual average sediment concentrations at the HYK and GC stations were 18.03 kg/m³ and 13.40 kg/m³ in 1996, respectively. During this year, channel aggradation occurred with notable riverbed elevation, increasingly wider and shallower cross-sections prone to morphological changes compared with 1986, and channel pattern alterations. Sediment concentrations dramatically decreased to 0.28 kg/m³ and 0.96 kg/m³ in 2016, respectively, and the main wetted cross section morphology remained stable (Figure 10a,c), indicating that substantial sediment reduction promoted channel stability [44]. This transition reflects the shift from persistent deposition before XLDR operation to persistent erosion after XLDR operation in the HYK-GC reach, with more pronounced scouring effects observed closer to the dam [46]. Analysis of parameter correlations revealed a negative correlation between A_bf and ξ, whereas ε showed a positive correlation with ξ (Figure 9). Since 2017, the ξ value of the three cross-sections has increased (Figure 10d), suggesting potential future aggradation that is particularly more pronounced at the HYK section than at the JHT and GC sections. Regarding the intra-annual variation in ξ, the XLDR maintains clear water discharge during the non-flood seasons [50], and the A_bf value of the three sections maintains a relatively high level. During the flood seasons, higher ξ corresponds to poorer water–sediment relationships, resulting in weakened scouring, enhanced deposition, and inevitable reduction in A_bf.

5.2. Influence of the Main Flow Direction and the Distribution of Control Engineering Structures on the Lateral Evolution of the Main-Channel Morphology

After the XLDR operation, the main channel of the wandering reach of the Yellow River continued to undergo siltation and erosion, and the river channel morphology changed considerably [51,52]. The main-channel cross-sectional morphology adjustment mainly reflects the lateral and vertical changes in river width and riverbed elevation, which is essentially the result of the nonlinear coupling of multiple factors. In addition to the influence of water-sediment conditions, the river course direction and distribution of control and guiding projects have a considerable impact on the main-channel morphology evolution [53]. According to Figure 5, the SQ, GT, HS (non-control area), and WA sections have varying degrees of erosion toward the south bank, whereas YF, WA, and QHJ have varying degrees of erosion towards the north bank. BC, ZLZ, and HS (control area) show no obvious erosion. Through further analysis of the river course direction and distribution of control and diversion projects, it was found that the BC, ZLZ, and HS sections (the control area) are directly affected by the control and diversion projects. The lateral erosion close to the control engineering is not obvious, but the longitudinal erosion is quite significant. Sections far from the control and diversion projects showed different erosion and scouring states. The BC and HS sections showed no considerable changes, whereas ZLZ showed obvious lateral erosion and longitudinal scouring. This might be related to the different distributions of the main channel line of the river course [54,55]. The main channel lines of BC and HS are distributed close to the control and diversion projects, whereas the main channel line of ZLZ deviates toward the side away from the control and diversion projects. For sections SQ, GT, HS (non-control area), and WA that were not affected by the control and diversion projects, erosion mainly occurred on the side close to the main flow line, especially in SQ and WA. The water continuously scoured the banks, causing them to collapse, which is the main cause of lateral erosion of the river course. For sections YF and QHJ, which are outside the control and diversion projects, erosion mainly occurred on the side outside the control and diversion projects, whereas the other side experienced different degrees of siltation. After the water leaves the control and diversion projects and is no longer constrained, there is a considerable outward expansion trend, causing the riverbank to erode toward the side close to the control and diversion projects and shifting the main flow line. The side far from the control and diversion projects experiences siltation due to a reduction in flow velocity.

6. Conclusions

(1)

After the XLDR operation, the HYK–JHT reach experienced erosion in the median water and dry years. The erosion volume in the dry year was greater than that in the median water year. The length of erosion decreased from 51.91% in the median water year to 34.61% in the dry year, whereas the length of sedimentation increased from 38.44% in the median water year to 65.36% in the dry year. The erosion volume was large and concentrated in the dry year, whereas it was smaller but more dispersed in the median water year.

(2)

The river channel cross-sectional shape evolution is influenced by changes in water-sediment conditions and also constrained by the main river flow direction and regulating engineering. In the area without the constraint of regulating engineering, the main channel of the river undergoes longitudinal and transverse erosion in the direction of the main river flow, such as cross-sections like SQ, GT, and QHJ. In the area with the constraint of regulating engineering, the main channel of the river undergoes longitudinal erosion within the area constrained by regulating engineering, such as cross-sections like BC, HS, and ZLZ.

(3)

The construction of the XLDR has had a considerable impact on the evolution of the main-channel morphology parameters. The A_bf and ε values of the typical cross-section showed different changing trends before and after the XLDR operation. Constrained by the dike and control projects, the R1, R3, and R4 reaches all shifted towards the south bank. The shift areas were increased by 2.5 km², 11.1 km², and 2.8 km² compared to the shift in the north bank. The R2 reach moved towards the north bank, and the shift area was 22.4 km² larger than that on the south bank.

(4)

After the XLDR was completed, the main factors influencing the changes in the main-channel parameters of the river underwent considerable changes. The main influencing factors of the main-channel parameters shifted from the joint effect of multiple factors to the prominent effect of a single factor. The three typical sections showed strong consistency. When the sediment input coefficient was between 0–0.005 kg.s.m⁻⁶, water-sediment had a positive effect on shaping and evolving the main-channel morphology of the river.

(5)

From a global perspective, the river channel morphology of the wandering reach of the Yellow River has reached a relatively stable state, but in some small areas, there are still phenomena of river channel oscillation and bank erosion. Future research on the river channel morphology evolution should be precisely aimed toward smaller areas and shorter distances. This places higher demands on the acquisition of data such as water level, flow rate, and sediment concentration. How to obtain precise data on water level, flow rate, and sediment concentration in any area of the wandering reach through existing means and measured data from key hydrological stations is a difficult point that must be overcome. The resolution of these issues and the resulting research will provide support for accelerating the governance of the Yellow River and strengthening its stability.