Comparing Extreme Values and Partial Time Series of Maximum Daily Precipitation to Identify the Best-fit Distribution (Case Study: Tehran Province)

The analysis of frequency and return period of rainfall and flood events is the most important step in water resources management and hydraulic structure designing. In this paper¡ the data of 24-hour maximum precipitation in five synoptic stations of Tehran province were provided to identify the best-fit distribution. Then¡ the results of nine probability distributions including Normal (N)¡ 2-parameter log-normal (LN2)¡ 3-parameter log-normal (LN3)¡ Pearson type 3 (P3)¡ Log-Pearson type 3 (LP3)¡ Gumbel (GUM)¡ Generalized extreme values (GEV)¡ Generalized logistic (GLOG)¡ Generalized Pareto (GPA) distributions were compared with the partial series fitted to exponential distribution using HYFA and MATLAB softwares. The RMSE ¡R2 ¡Bias ¡MAE and (Relative Root Mean Square Error) RRMSE statistics were used to identify the best-fit distribution at all stations. The results showed GPA at Abali and Mehrabad stations¡ GLOG at Doshantapeh and Northern Tehran stations and LN3 at Karaj station have better fit to data. Also¡ the Normal Exponential distributions were the most unsuitable ones. Finally¡ the results of RRMSE statistic represented that GEV distribution is the best-fit distribution and can be suggested for analyzing of maximum daily rainfall for the study area.