Non-stationary Flood Frequency Analysis And Flood Control Risk Assessment Of Reservoir Under Changing Scenarios

In recent years,with the impact of climate change and underlying surface changes,extreme events such as floods and droughts have become more frequent,and the applicability of hydrological analysis methods based on consistency assumptions has been questioned.The application of hydrological frequency analysis is becoming more and more common.This paper takes Yichang Hydrological Station and Yangfanggou Reservoir as the research object,analyzes the non-uniform flood frequency of runoff and flood,and further analyzes the flood control risk of Yangfanggou Reservoir by considering the uncertainty of future rainfall.The main research contents and results are as follows:(1)Hydrological sequence non-uniformity diagnosis.The non-uniformity diagnosis of the runoff series of Yichang hydrological station and Yangfanggou hydropower station was carried out by using the more commonly used Pettitt and Mann-Kendall nonparametric test methods.Under the significance level of 5%,the hydrological sequence of Yichang hydrological station has significant jump.And the trend,the hydrological sequence of Yangfanggou Hydropower Station has little inter-annual variation and is still consistent.(2)Non-uniform frequency analysis.The maximum daily,1d,3d and 7d floods from Yichang Hydrological Station and the maximum 24 h,3d,7d and 15 d annual floods of the Yangfanggou Hydropower Station dam site,annual maximum flood peak flow,hydrological annual average flow,and average dry season The flow-running sequence is based on the GAMLSS model to select the optimal non-uniformity models of each runoff series for flood frequency analysis.(3)Consistent design flood calculation.From the baseline surveyed runoff sequence of the dam site of Yangfanggou Hydropower Station and the simulated runoff series of the coupled SWAT model of ASD and SDSM models in different scenarios in the future,the frequency curve is drawn by consistent frequency analysis,and the flood frequency curve of the baseline flood sequence consistency design is designed.Most of them are located between the ASD and SDSM models,and the design flood results of the flood sequence in the annual runoff sequence are smaller than those in different scenarios.(4)Uncertainty analysis of precipitation.From the measured precipitation and simulated precipitation in the reference period,the Bayesian model weighted average comprehensive forecast value is used,and the Monte Carlo combined sampling method is used to generate the uncertainty interval of the Bayesian model forecast value,and the forecast accuracy and uncertainty are obtained.The interval is evaluated,indicating that the Bayesian model reduces the relative error of the total amount of precipitation sequences,and combines the advantages of two downscaling models of ASD and SDSM.(5)Flood risk assessment.Based on the Monte Carlo sampling method,the optimal non-conformity model is used to calculate the design flood of Yangfanggou Reservoir,and the uncertainty of precipitation is transformed into the uncertainty of design flood.