Floodplains are an important part of the LYR, and the reasons for floodplain erosion can be summarized as follows. (a) Human activities, which affect fluvial systems in many ways (Xu, 2003). i) In recent decades, with increasing population and rapid urbanization, economic construction and land use have entered a period of rapid development, including agricultural production and building construction. ii) The residents of the floodplain domain have constructed many farm dykes on both sides of the main channel to allow agricultural production and to prevent small to moderate floods. iii) The construction of reservoirs has played an important role in blocking sediment, especially the operation of the XLD reservoir since 2000, and the sediment entering the LYR has dramatically decreased with an average annual sediment load of 1.03×10⁸ t/yr during 2000-2017, which is 83% lower than that from 1986 to 1999 (see Figure 3). Additionally, the XLD reservoir has efficiently controlled large-scale floods, which has reduced the risk of flooding in the floodplains. (b) Climate changes. A change in the rainfall regime may change the erosion and transport capability of storm runoff (Xu, 2003), especially extreme rainfall events, which are a determining factor in hydrological soil erosion (Nunes, Seixas, Keizer, & Ferreira, 2010). Owing to the thick layer and short depositional time of sediments in the floodplains of the LYR, loose sediment particles are vulnerable to erosion by rainfall and runoff, resulting in a decrease in the surface elevation.

The morphology of natural river channels can be divided into the cross-sectional form, longitudinal profile and plan form. The quantitative characteristics of the cross-sectional form include the width, depth, area, wetted perimeter and width-to-depth ratio. The longitudinal profile is mainly reflected in the changes in the riverbed slope along the channel, while the quantitative features of the plan form include the bending coefficient, curvature radius, bending distance and swing amplitude. In this research, we mainly consider adjustment regulation of the riverbed morphology from three aspects: the longitudinal profile, the horizontal migration of the thalweg and the cross-sectional forms.

The longitudinal profile, which is a graph of height (H ) against distance downstream (L ) expressed by H = f (L ), is an important element of river geomorphology. From 1986 to 1999, the longitudinal profile of the LYR displayed a trend of continuously rising riverbed elevation, in which the average riverbed elevation of the HYK-JHT reach increased by approximately 2.6 m, and the average riverbed elevation of the JHT-GC reach increased by approximately 3.0 m (Shen, Zhang, Li, Shang, & Pan, 2000). Since 2000, the river channel in the LYR has experienced continuous scouring. Figure 9(a) shows the longitudinal profiles of the riverbed in the HYK-GC reach for different years (2000, 2005, 2011 and 2017) by using the mean riverbed elevation points of the channel. The average riverbed gradient basically remained at approximately 0.18‰ during 2000-2017, which indicates that the longitudinal profile of the riverbed underwent parallel downcutting during scouring. The average bed elevation during the pre-flood season in 2017 was 2.78 m lower than that during the pre-flood season in 2000.

The reach between HYK and GC in the LYR is a typical braided reach, with complex changes in the river regime, and it is easily affected by the water discharge and sediment load from upstream. In addition, the main flows swing frequently and continually erode the floodplains, which results in frequent shifts between the main channel and the floodplain and places certain pressure on flood control works (Hu, 2003). Here, selecting the thalweg points during the pre-flood season in May 2000 as reference positions, the horizontal migration distances of the thalweg points in different years are calculated (see Figure 9b). A negative value indicates that thalweg points moved towards the left bank, and a positive value indicates that thalweg points moved towards the right bank. From a comparison of the different periods, large variations are observed to occur in the thalweg migration amplitude in the different years. Among them, the average annual migration rates along the three hydrological sections are 1,480 m, 1,300 m and 343 m. Moreover, we find that the average annual migration rates of thalweg points in the HYK and JHT cross-sections are four times higher than that in the GC cross-section. The amplitude of thalweg migration gradually decreases in the downstream direction, and the limiting effect of the channel form on thalweg migration is strengthened.