Supria Paul, M. Alfi Hasan and Ahsan Azhar Shopan
Publication year: 2014

Daily rainfall distributions are used extensively for design of engineering works such as municipal drainage esystems, highway and railway culverts, and agricultural systems. It also plays an important role in a diverse range of nonstructural problems involving natural hazards associated with extreme events. In this study, attempts have been made to determine best fitted rainfall distribution and future projected rainfall using the best fitted distribution. For this purpose, observed rainfall data are analyzed using 57 different types of distributions considering three types of fitness co-efficient for a better understanding of rainfall extremities. Analysis has been made for the time period starting from 1971 to 2010. A combined ranking order has been applied considering Kolmogorov-Smirnov, Anderson darling and Chi-Squared goodness fit co-efficient for all the selected stations. Then, a statistical downscaling model, SDSM has been used to generate future projected rainfall for each station over Bangladesh. It has been found that, Burr type XII distribution, Birnbaum–Saunders distribution and Log Pearson type III distribution are the best fitted distributions for most of the rainfall stations in Bangladesh. A2 scenario cannot able to capture rainfall distribution of hilly area properly. But it can capture the distribution pattern of relatively flat area like Dhaka city.