Multivariate Hydrological Analysis Based On Bayesian Model Averaging And Copula Function

Summary the development process of hydrological frequency calculation method. Analysis the limitations of traditional analysis methods, thus puts forward the research content of this article.Based on the Copula function of the basic theory, introduced a multiple model integrated simulation of Bayesian Model Averaging, using the expectation-maximization algorithm to calculate the marginal distribution of Copula function. Using the 126-year observed data (volume, peak and duration) of the Yangtze River, calculated the suitable Copula function, so as to analyze the variables between two joint probability distribution and the conditional probability distribution and the corresponding return period, for the flood control and disaster mitigation in the study area and provide scientific basis for water conservancy project management.Specific conclusions were given in the following:(1)Using the bayesian model averaging method can improve the frequency distribution of hydrological variables, make its can more accurately reflect the hydrological variable frequency. (2)Using bayesian model averaging method to calculate frequency distribution as marginal distribution of Copula function, and calculate three kinds of archimedean Copulas function.Rank correlation coefficient calculation, find out that peak and volume, volume and duration have a strong positive correlation, while peak and duration have a relatively weak positive correlation. (3)Based on three kinds of archimedean Copulas function evaluate goodness-of-fit and fitting test, Frank Copula function expression effect is superior to other two kinds of forms of expression.Graphical analysis method and relevant evaluation index calculation both proved this conclusion, the optimal function of expression through Anderson-Darling fitting test proved to be conform to the inspection standards. (4)According to the optimal function expression, building peak and duration, peak and volume, volume and duration of the joint probability between the two variables and the over probability, we mapped the distribution probability. According to the joint probability to calculate the joint return period, drawing contour map, to plan water conservancy project management, and to provide theoretical basis for flood control. (5)According to the optimal Copula function expression, respectively establish peak and duration, peak and volume, volume and duration of conditional probability between the two variables, under certain conditions the variables drawn into the return period of conditional probability of figure, for risk assessment and provide scientific analysis of construction management.