DATA AND METHODS

3.1 Data collection

With the operation of the XLD reservoir, the sediment entering the LYR has dramatically decreased, and the geomorphology of the LYR has changed considerably. To investigate the temporal-spatial distribution of sediment and the geomorphological evolution of the braided reach between HYK and GC, hydrological data from 1960-2017 and bathymetry data sets of various cross-sections from 2000 to 2017 were collected.
The hydrological data on this study reach, including the annual runoff and sediment concentration, were measured at the HYK and GC stations from 1960 to 2017. The cross-sections lie between opposite primary levees of the LYR, with the distance between neighbouring cross-sections ranging from 0.5-5 km along the channel direction. The Beijing 54 geodetic coordinate system was adopted for all the bathymetric data sets to obtain the elevation of the riverbed in reference to a fixed level across the whole region. The Yellow Sea 1985 datum was applied as the reference level in this research.

3.2 Methods

The spatial interpolation method has been an active research topic for many years in various fields. This method has been applied to examine changes in sediment deposition and erosion in different areas, such as the tidal reach of the Yangtze River (Yuan, Lin, & Sun, 2020) or the Yellow River Delta (Fan, 2019; Jiang, Pan, & Chen, 2017). Similarly, the numerical simulation of river hydrodynamics requires spatial interpolation of discrete bathymetric data to determine the elevations at the nodes of the computational grid. There are many spatial interpolation methods for river channel topography, such as the inverse distance weighted, natural neighbour, kriging, spline, and other methods, which have their own advantages, and a reasonable interpolation method should be selected according to different characteristics of the river channel topography.
The braided reach of the LYR contains a broad floodplain with secondary perched compound cross-sections. According to the distribution characteristics of the measured cross-section sets, the floodplain and main channel are treated separately to obtain high-resolution DEMs from the scattered data. Merwade et al. (Merwade, Maidment, & Goff, 2006) showed that in a flow-oriented coordinate system, anisotropic spatial interpolation methods are clearly better than isotropic methods for interpolating river channel bathymetry. In this research, the mesh generator of the MIKE software (Amidror, 2002) was employed for mesh generation and interpolation purposes. A curvilinear grid was constructed covering the main channel, and the distance along the channel direction s was divided by a factor larger than 1 (3 was adopted). An unstructured triangular mesh was applied in the floodplains with natural neighbour interpolation in regard to the bathymetry, as shown in Figure 4.