Design rainfall estimation in Qatar : selection of the best-fit probability distribution

![CDATA[Design rainfall is an important input to hydrological studies needed for urban stormwater design and various other water resources design tasks. Design rainfall is generally expressed in the form of intensity-frequency-duration (IFD) curves. A number of steps are involved in the development of IFD curves. One of the most important steps in deriving IFD curves is the selection of a suitable probability distribution for at-site rainfall frequency analysis. When recorded rainfall data length is not long enough, which is a common problem in arid regions, the selection of appropriate probability distribution in frequency analysis becomes a difficult task. This paper focuses on the selection of a suitable probability distribution in rainfall frequency analysis in Qatar using twenty nine rainfall stations in Qatar. This is the largest rainfall dataset available in Qatar for the first time as a part of a national project on design rainfall estimation and development of national stormwater guideline. A number of goodness-of-fit tests are applied in this study, such as Kolmogorov-Smirnov test, Anderson-Darling test and Chi-squared test to select the best fit probability distribution for 24-hour annual maximum rainfall. A total of fourteen candidate distributions are considered. It has been found that Generalised Extreme Value distribution is the best fit distribution, followed by Log-logistic distribution. The findings of this research will assist in developing updated IFD curves for national application in Qatar.]]