Robust Real-time Flood Forecasting System

Robust real-time flood forecasting system is to introduce robust estimation theory into the real-time flood forecasting. The robust system can prevent abnormal factors from entering the flood system, so as to ensure the stability of the system and the accuracy of flood forecast. Firstly, the theory of robust estimation was introduced into every aspect relating to dealing with observed data of real-time flood forecasting, and robust estimation methods of single factor adapt to hydrological characteristic were presented. For rainfall error in remote system, three-stepwise robust correction method was put forward. Correction effect of method was improved by rainfall center considered, average rainfall graded and iterative computation. The other correction method, robust least square estimation, also was introduced. The impact of average rainfall graded and different initial value on correction results was study. The result shows that the effect of two robust estimation methods is better when error extreme is larger and the number of outliers is less. For measured outflow influenced by outliers, the robust least square method with some additional condition was employed for Muskingum parameters. The study of the effect of robust estimation relative to least square estimation shows the robust estimation can avoid the influent of outliers and obtain stable parameters. For outliers occurring in the measured data of flood discharge, the parameters of the AR model obtained by the robust recursive least square method are more robust than those obtained by the standard recursive least square method. The rectified results of parameters by the two approaches were compared, the result shows that the robust recursive least square method has potential to reduce estimation bias in the presence of noisy, and get parameters of less mean square error. Secondly, robust estimation methods of single factor and real-time flood forecasting system were integrated, and applicable and robust real-time flood robust forecasting system was recommended. The results of the system and standard real-time flood forecasting system for different basins were compared. The real-time flood robust forecasting system can resist outliers in rainfall and discharge, so as to ensure to gain stable and accurate forecast results. Lastly, by means of Mont-Carlomethod, the characteristic of risk and effect of robust estimation methods in flood forecasting was investigated. The result shows that risk is stable and effect is better when the same weight function constants are used and outliers is bigger. The risk and effect would change to different direction if constants are changed.