Parameter Optimization And Real-time Updating Procedure Of The Flood Forecasting Model

Flood forecast technology is an important and non engineering tool for preventing and mitigating the floods, and provides scientific and accurate decision support for flood control and early warning in the basin. In this paper, based on the Shuangchao hydrological model and Nash unit hydrograph routing method, we established a flood forecast model for the upper basin of the Fenhe River. To avoid the defects of single objective optimization method, the multi-objective parameter optimization approach based on the MOSCDE(multi-objective shuffled complex differential evolution) and SCE-UA(shuffled complex evolution method developed at The University of Arzona) algorithm respectively was proposed. A real time forecasting procedure based on the Kalman filter was used to update the flood forecasting results. This process provides a higher precision of forecasting scheme for the upper basin of the Fenhe River, the main work is as follows:(1) With depth study in semi arid area Shuangchao model production convergence principle, compared with other runoff model, Shuangchao model based on the characteristics of semi arid region hydrology, improved Philip infiltration equation, in the process of the runoff calculation yield. The concept of water supply degree was introduced as the surface runoff control factor, it divide the runoff process into the infiltration excess runoff and interflow of chaochi, the method effectivly simulation the over river basin climate production flow mechanism.(2) The traditional optimization algorithm can only control one single flood characteristics optimize in the parameters of the model, so the MOSCDE and SCE-UA algorithm was used to establish Shuangchao model parameters optimization multiobjective parameter optimization. According to the characteristics of single objective SCE-UA,which used the linear weighted method to weight each objective function, transforme multi-objective into single objective. MOSCDE through non inferior solution set of multi-objective optimization, and in the evolution introducing the DE algorithm as the core of the algorithm to improve the convergence speed and the calculation efficiency of the algorithm. This paper mainly focuses on the comparison of the two algorithms' affect for flash flood forecasting accuracy. The results show that screenings of flood peak by the MOSCDE the multi-objective optimization method for the calculation of the qualified rate and peak time qualified rate of two indicators have been improved significantly. Therefore, multi-objective MOSCDE optimization method for runoff prediction in the upper reaches of Fenhe River has a better adaptability and can provide support for regional flood forecast in parameter optimization.(3) Based on the basic idea of Kalman filter, the sigma point Kalman filtering method was used to correcte the flash flood forecasting results in real time, and the precision of flood forecast was further improved. After the real-time correction, backcourt flood evaluation indexes such as the qualified rate of flood peak and flood the determination coefficient have been obvious improve.So the introduction of real-time correction technology for flood forecast of flash floods can provide reliable basis for flood timely and accurate warning.