Application of Power Law in Flood Frequency Analysis of Sarbaz River

Nowadays it has been proved that the frequencies of many natural phenomena (such as floods earthquakes, landslides, and forest fires) fit with the power law distribution. The power law distribution is extracted from the fractal concepts of earth science events such as rainfall and runoff. In the present study, this distribution and also other conventional probability distributions were used to analyze flood frequency in the Sarbaz River of Sistan and Balochestan Province. Since in the studied area at some cases the river flow amount was almost low or even zero and also in the wet years several floods occurred so use of the maximum annual floods series could not provide accurate results. Therefore, in this study partial duration series were used for flood frequency analysis. Results showed that the partial duration series on the contrary to the annual maximum series fitted with the power law distribution properly. Among all the probability distributions which were used for flood frequency analysis the power law distribution had the best performance so that the Relative Root Mean Square Errorindices for power law distribution compared to conventional probability distributions in Pirdan and Pishin sites reduced at least 35 and 40 percent, respectively.