A Study On Selection Of Probability Distributions Of Extreme Hydrologic Parameters For The Peripheral River System Of Dhaka City
Use of appropriate probability distribution function is one of the most crucial steps of frequencyanalysis. There exists large number of probability distribution functions and hydrologists follow different methods to select the best one. In the current research, annual maximum water level and discharge of five peripheral rivers, namely Buriganga, Turag, Tongi, Balu and Lakhya around Dhaka city have been analyzed to compute the basic statistics and fit them with sixty two probability density functions (PDF). Three goodness-of-fit (GoF) statistics, namely Chi-square, Kolmogorov–Smirnov and Anderson Darling were used to rank each of the distribution. Moreover, ranks obtained from three GoF were used to compute overall rank of all distributions for each hydrologic parameter. The study result shows that, two different distributions were found best fit for two parameters. Dagum (4P) fit best for annual maximum water level, whereas Cauchy was found best for annual maximum discharge. Ranks of frequently used distributions, namely General Extreme Value (GEV), Log-Pearson III (LP3), Log-normal (LN) and Gumbels were compared with the best fit distributions and did not give satisfactory results. The method used in this study would be helpful for flood frequency analysis of other rivers of Bangladesh. This may also be used for evaluation of best fit distribution of river system for other countries as well.
Keywords- Probability Distribution, Rank, Water Level, Discharge, Dhaka City.