Risk Analysis Of Flood Peak Reduction Based On Uncertainty Of Flood Forecast Model

For the reservoir with weak regulation performance,flood control is the main task and face greater pressure in flood season,many factors such as flood fluctuation and flow concentration time rapidly puts forward higher requirements for rainfall runoff forecasting and flood control.Rainfall runoff model plays an important role in flood forecasting and dispatching,however,the model parameters are uncertain,the optimal parameter obtained by the traditional optimization method is also not unique,the uncertainty of model parameters leads to uncertainty of forecasting process,flood control decision which is made based on the single prediction process of the traditional model has a certain risk,the scheduling decision made by the scheduling rules does not cut the peak of the flood process,and the adjustment of the scheme is mainly based on artificial experience,uncertainty has always existed from forecasting process to scheduling process.Aiming at the problem of uncertainty in forecasting and scheduling,this paper takes Shuangpai reservoir in Hunan Province as a case study,through the calculation and analysis,the rationality and practicability of the model,method and strategy are verified,the main research contents are as follows:(1)Based on the analysis of rainfall runoff data,combined with GLUE(Generalized Likelihood Uncertainty Estimation)method and Monte-Carlo random sampling method,Nash-Sutcliffe as the target,for a long time in different time step continuous forecasting simulation,through a lot of simulation,the probability distribution of the parameters of the Xin'anjiang model is obtained and the uncertainty of the parameters is analyzed,providing reference for reducing the parameter value range,laying the foundation for improving the prediction level.(2)Establishing optimal flood control model under flood peak reduction goal,according to the drawbacks of conventional flood control operation without completely cutting the peak of the flood,putting forward the strategy of gate operation under the condition of peak cutting,using POA two stage optimization method to reduce the peak of the storage process,Nash-Sutcliffe as the target,through the analysis of the status of the gate process uncertainty fitting peak cutting process,under the premise of ensuring the safety of reservoir downstream,searching for a variety of scheduling strategies for a certain flood,it has practical significance and application value to improve flood control safety.(3)The uncertainty of forecasting model parameters is applied to the risk analysis of scheduling decision,presenting an automatic adjustment strategy of scheduling scheme with considering risk under flood peak reduction goal.Due to the uncertainty of model parameters,even in the case of actual rainfall,it is possible to produce a large prediction error,therefore,there are certain risks in flood control decision making based on the forecast results.Using the multi group model for a flood of ensemble forecasting,then the simulation results are introduced into the scheduling decision to get the probability of risk events and the uncertainty of the forecasting process is transformed into the uncertainty of the scheduling process.Aiming at the disadvantages of conventional scheduling decision adjustment,combined with the state of gate operation under the condition of flood peak shaving,the automatic adjustment of decision with considering risk is realized,the results showing that the method is effective and practical,it has practical application value to improve flood control decision-making level.Finally,this paper summarizes the results of this study and further research prospects.