Study On Frequency Distribution Of Short-duration Rainstorm Based On Joint Distribution In Wei Basin Of Shaanxi Province

Short-duration rainstorm can easily lead to the occurrence of urban waterlogging and mountain flood disaster,which poses a serious threat to people's life and property and damages the regional ecological environment.Study on the frequency distribution of short-duration rainstorm can provide reliable design results of rainstorm for urban drainage engineering design,mountain flood prevention and control in small watershed and comprehensive management of gully,which is the basis for rationally determining project scale and rainstorm disaster warning and prediction measures,and is of great significance for rainstorm flood prevention and control.Based on measured rainstorm data from 39 meteorological stations in Wei basin,Shaanxi over 1981 to 2017,this study inter-compared the commonly used 6 kinds of single variable frequency distribution curve,and three kinds of joint distribution parameter estimation method,and studied marginal distribution and the joint distribution of the maximum rainstorm series,and two variables of eight statistical time(10 min,20 min,30 min,45 min,60 min,90 min,120 min,180 min).The main research contents and results are as follows:1 Analysis of spatial and temporal evolution characteristics of rainstorm.The Mann-Kendall nonparametric test and sliding t test were used to analyze the time history evolution rules of the rainstorm series at each station.Resultes showed that: When the significance level ?=0.01,there is no trend or sudden change in the series of heavy rainfall in each period.By using the spatial interpolation method to analyze the statistical characteristics of the rainstorm,it was found that the mean rainstorm amount in each period increased gradually from the southwest to the northeast,the Coefficient of variation decreased gradually from the south to the north,and the difference of the Coefficient of variation in each station decreased gradually.2 Comparative analysis and study on optimal distribution of univariate frequency distribution parameter estimation methods.Using probability point moment correlation coefficient method,the proposed optimal rule of absolute value method,and graphical method,the moment method,maximum likelihood method,and linear moment method in three methods Pearson type III distribution(P-III)(EXP),exponential distribution,normal distribution(NORM),generalized extreme value distribution(GEV)and generalized pareto distribution(GP)and Gumbel distribution(GUM department),in total of 6 kinds of frequency distribution curves were compared.The results showed that: in the two angles of the rainstorm series in different periods of the same station and different stations in the same statistical period,the error between the calculated value and the measured value of the frequency distribution is smaller by using the linear moment method,showing a good adaptability;Generalized extreme value distribution and generalized pareto distribution in short duration storm frequency analysis also has a good adaptability in the statistical period less than or equal to 30 min;The generalized pareto distribution showed obvious advantages,generalized extreme value distribution during the statistical period more than 30 min of stronger adaptability,and generalized extreme value distribution in the study area storm sequence of optimal number and the site's optimal quantity two aspects showed obvious advantages,can be used as the frequency distribution of short duration storm frequency calculation theory.3 The parameter estimation method of bivariate joint distribution and the form of the optimal joint distribution function were determined.Kolmogorov-smirnov method,AIC information criterion method,bayesian information criterion method and deviation squared and minimum criterion method were used to compare and analyze the advantages and disadvantages of correlation index method and maximum likelihood method in the estimation of Copula function parameters in Clayton Copula,Frank Copula and Gumbel-Houggard Copula function 3.The results showed that the error between the theoretical joint frequency and the empirical joint frequency of each Copula function after the parameter estimation by the correlation index method is smaller,which shows a good adaptability.The functions of gumbel-houggard Copula and Clayton Copula have good applicability.Clayton Copula has obvious advantages in the analysis of the joint distribution of short-duration rainstorms within 90 minutes in each rainstorm combination,while gumbel-houggard Copula has more advantageous sites in the combined distribution of short-duration rainstorms outside the period of 90 minutes.Moreover,gumbel-houggard Copula has obvious advantages in both the optimal number of rainstorm combination sequences and the optimal number of stations in the study area,which can be used as the theoretical joint distribution for the joint analysis of short-duration rainstorm.4 The recurrence period of the bivariate joint distribution of short-duration rainstorm were analyzed.On the basis of determined correlation index method and Gumbel-Houggard copulas connect as optimal copulas connect function form,parameter estimation method and parameters for orders over return period when calculating the co-occurrence return period were greater than the joint return period,given a joint return period and the co-occurrence return period when the heavy rain with all design value for the joint design value is greater than design value now,from the principle of partial bad engineering design,recommend using joint return period as bivariate return period calculation standard.