IDENTIFICATION OF THE UNDERLYING DISTRIBUTION FORM OF PRECIPITATION BY USING REGIONAL DATA

A method of identifying the underlying distribution form for precipitation is proposed in this paper. This method is different from the classical hypothesis-testing method and is based upon using regional data. Two fundamental assumptions are made: (1) the distribution forms of rainfall at all stations in a studied area are the same, and the third L moment ratios, tau3, at all stations are also the same; and (2) the unknown underlying distribution form is one among five commonly used distributions: i.e., generalized extreme value (GEV), gamma (GAM), lognormal (LON), generalized Pareto (PAR), and generalized logistic distribution (LGI). Using the proposed method, one can not only choose a distribution form for the given data, but also know the probability that the identification is correct. For a parameter within the range commonly occurring in practice, this probability is reasonably high. Therefore the proposed method would be valuable and informative.