Regional flood frequency modeling for a large basin in India

The computation of flood magnitude and its likely occurrence to design different hydraulic structures are major challenges to the research community. The present study has been carried out to identify the homogeneous regions in the Mahanadi basin in Chhattisgarh part (data from 26 gauge/discharge sites) of India using conventional and clustering-based homogeneity tests and then computation and identification of probability-weighted moment and L-moment-based best regional distributions for different regions. Different simple to complex distributions like Extreme Value-I, Generalized Extreme Value, Logistic, Generalized Logistic, Generalized Pareto, Normal and Log-normal, Wakeby-4, and Wakeby-5 was used in the analysis through standardizing procedure to compute regional distributions. The best-fit distribution selected by simulating several series and compute L-kurtosis along with the L-moment ratio diagram. The homogeneity analysis confirmed that this basin can broadly be divided into two different homogeneous regions with 15 and 11 stations in the first (Region-1) and second (Region-2) regions, respectively. The GEV distribution was found best suited for Region-1 while the Generalized Pareto worked well for Region-2. To make results more convenient for field application, catchment area-based equations were converted in the form of Dicken’s or Ryve’s formulae for these regions to estimate flood quantiles of any return period.