Modeling of magnitude and frequency of floods on the Narmada River: India

The concept of magnitude–frequency was introduced by Fuller, and since then, it has been widely used, especially in Europe and Asia. Over the past, 9–11 decades, several probability distribution models have been developed and applied. The principal objective of the present study is to identify the best-fit magnitude–frequency model at various sites on the Narmada River amongst the Gumbel Extreme Value type I (GEVI) and Log Pearson type III (LP-III). Therefore, Kolmogorov–Smirnov (KS) and Anderson–Darling (AD) tests of goodness-of-fit were applied to find out best-fit model at each site on the river under review. The result shows that LP-III model is the best fit for Dindori, Manot, Barman, Sandia, Hoshangabad, Handia, and Mandleshwar, and GEVI model is the best fit for the Garudeshwar site. Accordingly, flood magnitudes for 2, 5, 10, 25, 50, 100, and 200 year return period were predicted. The analysis shows that the return period of largest peak flood on record (69400 m3/s) on the Narmada River at Garudeshwar is 96 years. These models demonstrate satisfactory results for predicting discharges and return periods. In addition to this, magnitude–frequency curves reveal that fitted lines are fairly close to the most of stream flow data points. Therefore, GEVI and LP-III are the best-fit distributions for modeling of magnitude and frequency of floods on the Narmada River.