A dual-layer equivalent channel method based on nonlinear Muskingum flood forecasting method is proposed to address the new challenges posed by the construction of national digital twin watersheds for predicting river water level and flow velocity in flood forecasting. This method uses the Muskingum parameters K and X to derive the formula for the dual-layer equivalent channel section, thereby achieving synchronous simulation of water level and flow velocity through hydrodynamic methods based on traditional flow calculation methods. Using the flood process from Chenggouwan Station to Linqing Station on the South Canal as an example for verification. The results show that compared with traditional method, the dual-layer equivalent channel method can not only forecast flow in areas lacking large cross-section data through hydrodynamic methods, but also simulate water depth and flow velocity information at any time and any cross-section of the river channel, achieving the expansion of large cross-section data of the river channel. The determination coefficients of the simulation results of the flow rate of Linqing Section using parabolic and rectangular dual-layer equivalent channel methods are both above 0.98, with peak error percentages of 0.3% and 0.17%, respectively, and root mean square errors of less than 20 m³/s. In terms of simulating the flow velocity of the Linqing Section, both the parabolic and rectangular dual-layer equivalent channel methods accurately simulated the time when the maximum flow velocity occurred.

To address the issues of the large amount of data required and the high difficulty in obtaining data during the modeling process of hydrodynamic water quality models, a method based on hydrological approaches for solving the basic equations of one-dimensional water quality models was proposed. This method substituted river topography and flow velocity data with Muskingum method parameters K and X for river confluence in hydrology. Moreover, coupled with the basic equations of the one-dimensional water quality model, a river pollution confluence model based on the Muskingum method was developed. Since taking a constant value for K is unreasonable for calculating pollutant propagation time in the developed model, the functions of K and X relative to flow rate Q were derived to solve this problem, and the nonlinear Muskingum model was coupled with the one-dimensional water quality model to build a river pollution confluence model based on the nonlinear Muskingum method. The verification results of the Dapuling-Changtaiguan section of the upper reaches of the Huai River show that the river pollution confluence model based on the nonlinear Muskingum method has ensured accurate prediction of peak time compared with the river pollution confluence model based on the Muskingum method. At the same time, the deterministic coefficient of the pollutant mass concentration prediction results has been increased by 0.04, the root mean square error has been reduced by 0.07 mg/L, and the average absolute error has been decreased by 0.022 mg/L.

Based on the water cycle model of distributed architecture, the Chuhe River Basin was divided into hydrological characteristic units such as the slope of the hilly area, the slope of the plain area, the river channel of the hilly area, the river channel of the plain area, the polder area, and the sluice dam project. The Xin’anjiang hydrological model was constructed in the hilly area; the hydrological model and the one-dimensional hydrodynamic model were constructed in the plain area; the flood drainage model was built in the polder area, and the project dispatching model was constructed for the sluice dam project. All models were coupled into an integrated flood control and drainage model that could fully generalize each natural entity element in the Chuhe River Basin. The accuracy and reliability of the model were verified through the flood processes of typical stations in the hilly and plain areas during the flood years (2016 and 2020). The results show that the integrated model can reproduce the flood processes in the basin in 2016 and 2020 quite well, and both the calibration and verification results reach Grade A accuracy.