Identification and selection of best fitting candidate distribution for rainfall frequency analysis in Cameron Highlands

Abstract

In frequency analysis based on analytical method, there are quite a number of probability distributions to be used for quantile estimation. The selection of inappropriate one will lead to either overestimation or underestimation of the quantiles. Thus the identification and selection of the best fitting probability distribution should be given emphasis. The L-moment method offers advantages over the conventional method of moment and thus is more reliable in the distribution identification. The focus of this study is on the identification and selection of best fitting probability distribution, based on L-moment ratio parameters and L-moment ratio diagram. The results show that the GEV (Generalized Extreme Value) distribution fits quite well to data series at most of the homogeneous regions and rainfall intervals.