TR(BR)-133 : Regional flood frequency analysis for upper Narmada and Tapi sub zone 3C

Abstract

In this report, flood frequency analysis has been carried out for Upper Narmada and Tapi Subzone-3(c) using peak flood series data of 15 small and medium size catchments varying in size from 41 .80 sq. kms. to 21 10.85 sq. kms. The study involves application of Extreme Value Type-I(EV1), General Extreme Value(GEV) and Wake by distributions using ( 1 ) at site data (ii ) at site and regional data in combined form and (iii ) regional data alone. Statistical test based on U.S.G.S. method has been performed in order to test the homogeneity of the region.

Annual maximum peak flood series data of 13 bridge sites passing the U.S.G.S. homogeneity test have been considered for deriving the regional parameters. Whereas, the data of remaining two sites have been used for testing the methodology based on descriptive ability criteria. Synthetic flood series have been generated using the regional EV1(Case-1), GEV(Case-2) and Wake by (Case-3) parameters derived from the historical data of the catchments. Generated data sets of specific record lengths(same as the record length of historical data for respective gauging sites) have been considered for both the groups of the catchments for developing the flood frequency relationships. For the two independent sites, variable record lengths viz. 1 , 10, 20, 30 and 40 and 50 have been considered one at a time for computing the flood frequency estimates and the predictive ability criteria such as bias, root mean square error and coefficient of variation.

The above methodology has been applied to the generated data of different sample sizes for each population, considered as Case-1, Case-2, and Case-3 populations respectively for the two independent gauging sites. Performance of different methods has been evaluated based on predictive ability criteria viz. bias, coefficient of variation and root mean square error. it is seen that the method(SRWAKE) based on Wake by (PWM) approach using at site and regional data in combined form provides flood frequency estimates with computationally less bias, comparable root mean square error, coefficient of variation and coefficient of skewness for the two test catchments of the region.