欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
本文将不完全数据分析方法与Copula函数相结合,提出基于Copula函数的不完全降水序列频率计算方法,并采用蒙特卡洛试验研究其统计性能。以陕西省6个测站的年降水序列为例,构建二维Frank Copula函数。运用多元Rosenbrock优选法则求解,得到基于Copula函数的不完全序列似然函数参数值,采用Copula函数模拟法和拟合标准差（SEF）进行拟合效果评定。结果表明：基于Copula函数的不完全降水序列频率计算方法具有良好的统计性能;Anderson-Darling检验统计量均小于？（28）0.05显著性水平下临界检验值。设计站在本文方法下所得SEF值均小于单变量分布下的SEF值,设计值估算也优于单变量计算方法。在设计站与参证站难以进行相关展延资料的情况下,文中方法是一种合理可行的频率计算方法,以期为提高降水序列频率计算精度提供新思路。
英文摘要：
In this study we have developed a frequency analysis method of incomplete precipitation data series using Copula functions and examined its statistical properties by conducting Monte Carlo simulation tests. Using this method, we have built bivariate Frank Copula functions for six annual precipitation series observed in Shaanxi province and obtained a solution of the new parameters by applying a multivariable-constrained Rosenbrock optimization algorithm. To verify the method, Copula function-based simulations were conducted and evaluated using the criterion of minimum standard error of fit （SEF）. The results show that the frequency of incomplete precipitation series calculated using the Copula functions has good statistical properties with all the statistics of simulation tests lower than critical values of the Anderson-Darling test at a significance level of 5%. The SEF of bivariate joint distribution is less than that of univariate distribution and the former＇s estimation of design values is also superior. Frank Copula functions are generally effective and acceptable in analysis of precipitation frequency, and they are able to describe precipitation series better than the univariate methods, particularly for those design stations of incomplete series with difficult data extension.
同期刊论文项目
同项目期刊论文
期刊信息
位置：