This paper reviews the researches on the application of Kalman filter technique to the real time hydrologic forecasting. A method of multi-step (alternate) filtering correction on hydrodynamic model based on Kalman filtering is proposed.The feasibility of application of Kalman filtering to real time correction of hydrodynamic model is discussed in the paper and the state-space equation can be established using the discrete St. Venant equations by the four-point linear implicit scheme. The approach of multi-step (alternate) filtering real time correction on one dimensional hydrodynamic model of the single river is advanced. In this approach, three state-space equations that use water level, discharge and roughness as state vectors respectively are established and Kalman filtering is applied alternately to the three state-space equations, which achieve the correction of state variables and automatic adjustment of model parameter. Additionally, parameters of Kalman filter are estimated by trials and the estimation methods and the range of their values are put forward.The correction method this paper develops is applied to the reach from Qingxichang to Fengjie of Yangtse River. In this example, Kalman filtering is applied alternately to the state-space equations that use water level, discharge and roughness as state vectors respectively and the effect of the parameter estimation method using roughness as the state vector need further study in the example. Comparison between results of forecasting with correction and multi step extrapolation forecasting and results of forecasting without correction demonstrate that the forecasting error decreases and forecasting accuracy improves. The application of the correction approach proves reasonable and effective by this application.
|