水文频率分析中的多项式正态变换方法研究

采用多项式正态变换(聊盯)方法，把偏态分布转换为正态分布，推求指定频率下正态分布的设计值(分位数)，然后通过一一变换转换成原偏态分布对应频率的设计值．以P．Ⅲ型分布为例，用Monte-Carlo法进行模拟分析，并对PNT法的统计性能与线性矩(L广M)法的统计性能进行了比较．结果表明，PNT法的精度略差于L-M,g-ej精度．由于PNT法本质上是非参数估计方法，不需要假设水文变量的分布线型，因此该法理论上是一种较为可行的水文极值分析方法．

Polynomial normal transforms in flood frequency analysis

The polynomial normal transforms method (PNT) was applied to convert the skew distribution into the normal distribution. By derivation of the design value for the normal distribution under certain frequency, the value of the skew distribution under corresponding frequency was obtained through one-to-one transform. With the P-III distribution taken as an example, the statistical characteristics of the PNT method were investigated by means of the Monte-Carlo simulation. The comparison of the PNT method with the L-M method shows that the precision of the PNT method is compatible to that of the L-M method. Owing to the fact that the PNT method is a non-parametric estimation method in essence without requirement for an assumption of a statistical distribution for hydrological variables, the method is a potential method for flood frequency analysis when there is not adequate knowledge on the selection of the statistical model for the extreme value.