Assessment of Goodness of Fit Methods in Determining the Best Regional Probability Distribution of Rainfall Data

Abstract

One of the most important problems in time series analysis of stream flow and rainfall data in an area is selecting the best probability distribution. Since the rainfall stations are associated and correlated with each other, so statistical analysis of the station data seamlessly are very important. Therefore, the first step in data analysis, is selecting the prevailing probability distribution in the region. Probability Plot Correlation Coefficient (PPCC) method is one of the most simple, efficient, and pointed approaches in this field which is used for selecting the best probability distribution at each station. Among these, the method of L-Moments is also developed and has the capacity of regional analysis to choose the best probability distribution. In this study, we examine monthly and annual rainfall data for a period of 50 years with both PPCC and L-moments methods based on different evaluations, in the North West of Iran. Although, results from the two methods were almost identical, but because of the simplicity of calculations and the use of a curve to determine the best statistical distribution, the use of L-moments was recommended.