Modelling hyperconcentrated floods in the Middle Yellow River using an improved river network model

Abstract

In the section of the Middle Yellow River from Longmen to the Sanmenxia Dam, the prediction to hyperconcentrated flood processes is difficult because of the abrupt change of channel geometry along the river and the high variability of the amount of flood water and sediment load from the main river and tributaries. A river network model based on a Godunov-type FVM scheme is proposed in order to investigate the flow dynamics, sediment transport and bed evolution during hyperconcentrated floods. A quasi 2D algorithm is adopted to calculate the velocity at the junction cell, which is then used for providing the interior boundary conditions. The computation for each reach can be executed simultaneously and the solution procedure is parallelized using OpenMP. The speedup factor compared with a serial program increases linearly in theory with the number of reaches in the river network. Two flood events, featuring significant erosion and deposition respectively, were simulated to validate the model. The calculated water levels, discharges and fractional concentrations of non-uniform suspended load were in agreement with the measurements. A comparison was made between the proposed model and an existing model neglecting the momentum conservation principle at the junction, with the former showing better prediction of the peak water levels near the confluence. Numerical experiments were conducted to analyze the influences of reservoir operation and discharge capacity of the dam on flood routing and bed evolution in the river network. It is demonstrated that decreasing pool level during flood season and increasing discharge capacity are both valid measures to alleviate sedimentation in the Sanmenxia Reservoir, and the former measure is a prerequisite. Therefore, the model presented here can help to enhance the understanding of river network system subject to high sediment input and inspire better strategies on sustainable reservoir management.