Selection of a basin-scale model for flood frequency analysis in Mahanadi river basin, India

The present study provides an insight into a systematic evaluation of probability distributions using some statistical measures along with a few relevant catchment and flow properties to select a basin-scale model for flood frequency analysis (FFA) of Mahanadi river basin, India. A comprehensive analysis identified generalized extreme value (GEV), Pearson type 3, generalized Pareto, and Gumbel as the best-fit candidates for FFA of the watershed. GEV was selected as the basin-scale model based on a descriptive statistical ranking method followed by its test for predictive ability through bootstrap sampling. The distribution parameters were correlated with a few hydrological and physiographic characteristics of the watershed through regression analysis. The predictive capability of the regressed equations was assessed by comparing the observed mean annual flood (MAF) with the anticipated MAF derived from the expected value of GEV density function. Various return period quantiles were estimated using the parameters obtained from these equations and compared with the observed values, which confirmed the robustness of the physically based GEV model over the entire watershed. Flood flow values estimated at the gauging sites considering the site-wise best distribution and the basin-scale standard model were compared. The marginal difference of error between them further supported the application of a standard model for the entire basin despite site-wise different models.