基于R软件的Bayesian MCMC法水文频率线型选择不确定性研究

Research on Uncertainty of Hydrological Frequency Line Selection Based on R Software with Bayesian MCMC

线型假定使传统水文频率分析方法具有一定的不确定性,为快速有效评价水文频率分析线型选取带来的不确定性,以嫩江大赉水文站年最大洪峰流量序列为例,采用Kolmogorov-Smirnov、Anderson-Darling、Chi-Squared 3种拟合检验方法,分析了水文频率分析中较为常用的6种分布线型在嫩江大赉水文站的适用性,优选出最适合于嫩江大赉水文站的3种分布线型。基于优选结果,采用Bayesian MCMC法,应用R软件编程,对优选出的分布线型频率分析不确定性进行研究。结果表明,优选出的三种分布线型中,采用对数P-Ⅲ型分布获得的洪峰流量设计值不确定性最小,P-Ⅲ型分布次之,对数正态分布最大。采用R软件分析水文频率线型选择不确定性既便捷又有效。

There is certain uncertainty in the traditional hydrological frequency analysis method due to linear assumption.In order to evaluate the uncertainty caused by line selection of hydrological frequency analysis effectively and rapidly,the annual maximum flood peak series of Dalai hydrological station in Nenjiang was taken as an example.Three fitting test methods including Kolmogorov-Smirnov,Anderson-Darling and Chi-Squared were used to analyze the applicability of six distribution lines commonly used in hydrological frequency analysis in Dalai of Nenjiang River,and three distribution lines which were most suitable for Dalai of Nenjiang River were selected.The research on uncertainty of flood frequency line selection based on R software with Bayesian MCMC was developed.The results show that the uncertainty of design value obtained by the lognormal P-Ⅲdistribution is the smallest,followed by the P-Ⅲdistribution and lognormal distribution.It is convenient and effective to use R software for studying on hydrological frequency analysis and uncertainty research of line-type selection.