引言正如 Beard(2)所证明的邯样,由数学拟合洪水频率曲线所获得的超过给定洪水大小的概率,为一有偏倚的洪水风险的估计值。对于年洪峰的总体值假定为正态或对数正态的地点,Beard 曾估算过这种偏倚,它在很大程度上是由T年洪峰的时间抽样误差所造成的,因为这一T年洪峰是从年洪峰的一个有限系列估算出的。在后来的一篇报告中,Beard(3)指出,相同的偏倚量可用于偏差系数较小的 Pearson—Ⅲ型分布曲线。隶属水资源委员会的水文 INTRODUCTION As we can see from Beard (2), the probability that a given flood is mathematically derived from fitting a flood frequency curve is an estimate of the risk of a biased flood. Beard had estimated this bias for locations where annual population peaks were assumed to be normal or lognormal, largely due to time-sampling errors in the T-year flood peak, since this T year The peak is estimated from a finite series of years of peak. In a later report, Beard (3) pointed out that the same bias can be applied to Pearson-Ⅲ distribution curves with small coefficient of variation. Hydrology under the Water Commission