| Urban rainstorm flood is the important factor restricting the stable development of urban economy and society.Analyzing the urban rainstorm flood evolution law,identifying the key driver for the urban rainstorm flood evolution and evaluating the risk of urban rainstorm flood are helpful to scientifically understand the evolution mechanism of urban rainstorm flood in the changing environment,so as to effectively mitigating and avoiding the urban rainstorm flood.Taking the Jialu River Basin in Zhengzhou City as the study area,the rainfall data of three precipitation stations and the runoff data of Zhongmu hydrological station in the basin from 1983 to 2018 are collected and collated,and 50 rainstorm floods are selected,then the runoff generation and concentration characteristics such as rainfall,flood volume and flood peak discharge are analyzed by M-K trend test,accumulated distance equal method and other methods.According to the theory and method of hydrological variation diagnosis,the mutation characteristics of rainstorm flood are analyzed.Combining with the statistical data and remote sensing images,the impacts of driving factors such as rainfall elements,land use/cover changes and urban rainwater pipe network on urban rainstorm flood are explored,then the key driver of urban rainstorm flood and the main influencing factors of flood volume are determined.The wavelet analysis method is used to decompose the rainfall and flood volume,and the multiscale attribution analysis of flood volume is carried out.The multiscale and multivariate prediction model of urban rainstorm flood is constructed to simulate and predict the flood volume of rainstorm flood by using the multiple linear regression model and ridge regression model,then the model performance is tested.According to the urban development plan and the forecast of meteorological and hydrological conditions,the different future scenarios are set up and the multi-dimensional joint distribution model between urban rainstorm flood and its driving factors is established by using Copula function,the combination risks of flood volume and its influencing factors are evaluated.The main research results are as follows:(1)The runoff generation and concentration characteristics of 50 rainstorm floods show different evolution trends.The rainfall decreases slightly without mutation;the flood volume and flood peak discharge show an increasing trend,the two variables decrease suddenly in 1994 and increase suddenly in 2005;the flood peak lag time shows an increasing trend,increases suddenly in 2005 and decreases suddenly in 2015;the flood duration shows an increasing trend,decreases suddenly in 1998 and 2015,increases suddenly in 2005.(2)The runoff generation and concentration relationship of Jialu River Basin in Zhengzhou City changes significantly in 1994 and 2005.In the three periods,the rainfall-flood volume relationship in the first and third periods is basically the same,and the statistical characteristics of the two variables have little difference;the synchronous probability between rainfall and flood volume are both greater than 80%,and the probability of the same wet is slightly greater than the probability of the same dry;the two variables are dominated by the sameness on each periodic scale,and the difference is smaller.In the second period,the statistical characteristics of the two variables and the rainfall-flood volume relationship are significantly different from those in the other two periods,where the mean value and dispersion degree of the two variables are significantly reduced;the synchronous probability is 59.43%and the probability of the same dry is the greatest,and various asynchronous encounter probability all increase significantly compared with the other two periods;the two are dominated by the sameness on the medium-long and long periodic scales,and the differences is smaller;the two are dominated by the sameness on the short periodic scale,and the difference also dominates;the two are dominated by the difference on the medium periodic scale,and the dominance of the sameness remains unchanged.(3)The underlying surface change is the key driver for the rainstorm flood evolution in Zhengzhou City,the rainfall and the underlying surface condition are the main influencing factors of flood volume.On the short,medium,medium-long periodic scales,the flood volume and rainfall basically maintain a constant linear relationship,the impact of the underlying surface change on the flood volume is negligible;the impact of the underlying surface on the flood volume is mainly reflected on the long periodic scale.The multiscale-multivariate prediction model of urban rainstorm flood has a good simulation performance on the flood volume of the first 45 rainstorm floods,where NSE,R~2,RMSE,R_e are 0.957,0.958,547.31 thousand m~3 and 13.47%,respectively.Moreover,the model predicts the flood volume of the 46th to 50th rainstorm floods well,and the relative errors between the predicted value and the actual value are all less than 20%.(4)Using the joint distribution model of rainfall P,impervious area S and flood volume R,the three-dimensional encounter probability,the two-dimensional conditional probability when S or P are of different magnitude,and the one-dimensional conditional probability for nine combinations of different magnitudes of S and P are analyzed.The results show that the synchronous encounter risk is 24.72%and the synchronous wet risk is the biggest.With the gradual increase of S or P,the synchronism of the other two variables becomes better and better,where the synchronous wet risk increases gradually,the synchronous dry risk decreases gradually,and the encounter risk of two levels of difference between the two also gradually decreases.The risk occurrence probability of the flood volume exceeding 7.413 million m~3 in the nine encounter situations is 88.34%,64.50%,51.03%,37.10%,30.03%,23.88%,17.35%,7.89%,6.83%,respectively,and the impact of rainfall on this risk is greater than the impervious area. |
PDF Full Text Request