基于线型选择不确定性的水文频率分析——以黄河某站洪水资料为例

根据贝叶斯理论和贝叶斯因子法，计算样本隶属于某种线型的后验概率；然后，根据全概率公式，以样本隶属于某种线型的后验概率为权重，对各线型的概率密度函数进行加权平均，得到综合各线型的概率密度函数，进而给出各设计频率的设计值。最后，采用皮尔逊Ⅲ型和对数皮尔逊Ⅲ型作为备选线型对黄河某站洪水资料进行水文频率分析，结果表明：①从线型后验概率估计分析，对研究的样本系列，皮尔逊Ⅲ型稍优于对数皮尔逊Ⅲ型；②可以给出考虑线型选择不确定性的设计值，以此减少线型选择的不确定性。

Hydrological frequency analysis based on uncertainty of linear selection——A case study of flood data from a station on the Yellow River

According to Bayesian theory and Bayes Factor, the posterior probability of the calculation sample belonging to a model is calculated. This paper gives the design probability values by weighing individual probability density function based on the total probability formula. Finally, a case study of flood data from a certain station on the Yellow River is discussed, and the result indicates that P-Ⅲ distribution is slightly superior to LP-Ⅲ distribution; and the design values based on uncertainty of linear selection may be given.