Abstract
At-site Rainfall Frequency Analysis (RFA) is a crucial tool for designing of water related infrastructures. Partial Duration Series (PDS) and Annual Maximum Series (AMS) are the most popular techniques for RFA. PDS is capable of including more extreme events as compared to the AMS. Keeping in view the importance of at site RFA, in this study we identify suitable statistical model that best represent PDS and AMS extracted from daily rainfall records of 25 years in the vastly growing city, Rawalpindi, Pakistan. The most commonly used statistical distributions in RFA such as Generalized Pareto (GPA), Generalized Extreme Value (GEV), Generalized Logistic (GLO), and Pearson Type-3 (PE-3) distributions are used in this study. The parameters of these distributions are estimated by method of linear moments (LM). Anderson-Darling (A-D) testing criterion along with L-moment ratio diagram have been used to determine the best fit probability distribution. The findings of our study suggest that GEV as the best fit statistical distribution for PDS while GLO for AMS. Moreover, in this study we have also estimated quartiles for the best fitted probability distributions. The results revealed that PDS sample outperforms than AMS for various return periods. These estimates can be used in designing of water-related infrastructures such as culverts, bridges and other hydrological structures in the city.