Flood Frequency Analysis for Kosi River Basin, Bihar, India Using Statistical Methods

Abstract

Flood peak estimation provides assistance in water resources management by offering sufficient information regarding possible flood risk.In the present analysis, flood peaks are estimated for various return periods using the probabilistic model.Six statistical methods namely Normal, Gumbel, log Normal, General Extreme Value (GEV), Pearson III and log Pearson III are used to forecast the flood discharge of Kosi river which is responsible for inundating a large area of North Bihar plain despite various flood management activities.The annual flow data for a period of 33 years (1981 to 2013) at Birpur gauge station are used in the study.The flood peak magnitudes are computed for the return period of 5, 10, 20, 50, 100, 200, 500 and 1000 years.The Generalised extreme value method provided the higher values of predicted flood magnitude.The Goodness of fit test for six distributions is assessed using Kolmogorov Smirnov (KS), Chi-Squared (CS) and Anderson Darling (AD) tests.The tests of Goodness of fit show that Normal distribution followed by Generalised extreme value distribution provides the best results for Kosi river basin.The predicted flood peak for different return periods is of greater importance and may be utilised in designing important hydraulic structures along the river, constructing bridges, developing flood inundation zones and flood management activities.