Research On Stage Forecasting Model In Tidal Reach

Aiming at the damages of flood in current estuarine area, the importance of the flood control, the key problems and the difficulties of real-time river stage forecasting in tidal reaches, the paper studied current main models for channel flood routing: The one-dimensional hydrodynamic model base on difference solution of St. Venant equations and BP Neural Networks. The advantages and shortages of these models are analyzed, and it is pointed out that one-dimensional hydrodynamic model is capable of simulating rather well the dynamic channel flow of flood, but is inapplicable to the regions of insufficient data;while the BP Neural Networks method suffers a slow speed of convergence and unsteadily calculation. So the most urgent problem in river stage forecasting of tidal reach is how to develop a river stage forecasting modal which is stable, simple, widely applicable and of certain accuracy. Muskingum is a widely-used method for river flood routing in hydrology, which is simple, and of good calculation accuracy, as verified by lots of hydrologists over a long time. The paper establishes a river stage routing method by modifying Muskingum into flood routing method for stage forecasting according to the basic principle of Muskingum and the stage-discharge relation formula. Based on the characteristics of water moving in tidal reaches, the flood hydrograph of tidal reach is considered as the overlay of bi-directional propagation of flood wave from upper reaches and tide waves from lower reaches, and the two propagation processes are calculated by stage routing method respectively. Accordingly, the stage forecasting model of the tidal reach is established, and named as bi-direction wave channel routing method.This model is then applied to tidal reaches of Yangtze River and Cao'e River , with a relatively good calculation accuracy at the tidal reach of Yangtze River but poorer at Cao'e River. Based upon the analysis on the basic assumption of model and comparison over geographic, hydrologic and channel cross-section characteristic between Cao'e River and Yangtze River, the proposed model is amended and modifiedt, and finally a model of more wide applicability is obtained, which behaves well in the simulation of some floods in Cao'e River.The application of the one-dimensional hydrodynamic model and BP Neural Networks in Cao'e River is analyzed, the results of which are then compared withthose of bi-directional stage routing model. The result shows that with insufficient original data, bi-directional stage routing model can adjust its parameters and getting a better calculation result, which proves that it have better flexibility and applicability, and be simple with clear physical concept. Therefore this model deserves a further research and the development.