Open AccessASSESSMENT OF THE BEST PROBABILITY DISTRIBUTION METHOD IN RAINFALL FREQUENCY ANALYSIS FOR A TROPICAL REGION
Author(s) -
Stephen Luo Sheng Yong,
Jing Lin Ng,
Yuk Feng Huang,
Chun Kit Ang
Publication year - 2021
Language(s) - English
Resource type - Journals
ISSN - 1823-7843
DOI - 10.11113/mjce.v33.16253
Subject(s) - gumbel distribution , frequency distribution , frequency analysis , flood myth , goodness of fit , mathematics , generalized extreme value distribution , statistics , return period , anderson–darling test , null hypothesis , distribution (mathematics) , statistical analysis , kolmogorov–smirnov test , environmental science , statistical hypothesis testing , extreme value theory , geography , mathematical analysis , archaeology
The Intensity-Duration-Frequency (IDF) curve defines the relationship between rainfall intensities at certain durations and with the frequencies. The IDF Curve is extensively used in many applications such as flood modelling and peak discharge estimation. Over the years, the frequent occurrence of flood has become a great challenge in Kelantan river basin. Herein, IDF curves using frequency analyses based on different distributions were developed and compared. The historical rainfall data at eight rainfall stations for the period of 1985-2019 were selected for the assessment purpose. The Gumbel, Normal and Log Pearson Type III distributions were fitted into the annual maximum rainfall series for durations varying from 30 minutes to 24 hours. The goodness of fit tests were then used to evaluate the performances of each frequency distribution. It was found that the Gumbel distribution gave the highest passing rate followed by the Log Pearson Type III and then the Normal distributions. The Gumbel distribution resulted in respective 86% and 75% passing rate since most of the p-values generated by both the K-S and the Mann-Whitney test were greater than 5% of significance level leading to the acceptance of the null hypothesis. Thus, the Gumbel distribution is suggested for the frequency analyses in this study.