Study On Method Of Real-time Flood Forecasting Of Braided River And Its Application

Braided river is a common style of river. There are many braided rivers of two braches or many braches in our country, such as Yangzi River, Beijiang River, Xijiang River, Xinyi River in the north of Jiangsu Province etc. We can make use of the skill of current simulation of river networks, such as two, three and four-level unit solution, junctions group method etc. However, the braided river has so few junctions, so how to find a efficient and practical method is a very meaningful problem. The paper gives a new hydraulic model which is easier to apply in those braided rivers on the basis of Xijiang River and Beijiang River.Based on the Saint-venant equations describing the channel flow movements, the nonlinear algebraic equations derived by the use of Preissmann weighted implicit four-point scheme are solved with the Newton method. The precision about the result is improved greatly. A peculiar Guass elimination method with down dimension and compress-storage is used when solving the equations, which can help prevent the possibility of discontinuity caused by the very little denominator. At the same time, in order to improve the precision of forecasting result, according to the character of forecasting model of hydraulic, two different means of real-time correction were discussed and they are faded-memory least square method and Kalman filter method. Also a analysis and comparison was made in the thesis.The result produced by the real-time flood-forecasting model in the Xijiang river and Beijiang river proved the flood-forecasting in braided river was feasible and the model can be used for reference in resemble area.