Abstract:
The log Pearson type 3 distribution has been recommended in the U.S. and Australia for representing annual flood flow series. The method of moments is one of the popular methods used for fitting this distribution. This method has been applied to (1) the sample of observed (raw) flood values, (2) the
sample of logarithmically transformed values and (3) a combination of raw and logarithmically transformed values. Some investigators have felt unconfortable about: (1) the use of transformed flood data in flood frequency estimation and (2) the use of high moments of the sample (moments of order 3 or more). We present a method which applies directly to the observed flood sample but which uses moments of low order, namely the harmonic mean (moment of order-1), geometric mean (moment of order "quasi zero") and arithmetic mean (moment of order 1). We call this method the "Sundry Averages Method" (S.A.M.).
