1Central Water and Power Research Station, Pune, Maharashtra, India
Online Published on 19 December, 2024.
Design rainfall depth is used as an input to predict the design flood, which forms as a basis for planning and design of civil and hydraulic structures. Design rainfall is the output of frequency analysis that is determined based on the 1-day maximum rainfall data corresponds to each year. This can be achieved by fitting probability distributions to the series of annual 1-day maximum rainfall (AMR). This paper presented a study on intercomparison of extreme value family of distributions (EVD) viz., Extreme Value Type-1 (EV1), Extreme Value Type-2, Generalized Extreme Value (GEV) and Generalized Pareto (GPA) for modelling the AMR of Afzalpur, Aland and Kalaburagi sites wherein the parameters of the distributions were determined by method of moments, maximum likelihood method and method of L-Moments (LMO). The performance of the EVD applied in modelling the AMR was examined through model performance indicators such as correlation coefficient (CC), Nash-Sutcliffe model efficiency (NSE) and root mean squared error (RMSE). The study showed that (i) the average of predicted AMRs by MoM, MLM and LMO of EV1, GEV and GPA are closer to the observed AMRs for Afzalpur, Aland and Kalaburagi; (ii) the difference between the average of observed and predicted AMRs by EV1 and GEV are found as minimum; (iii) there is a good correlation between the observed and predicted AMRs by three methods of EVD and the CC values vary between 0.937 and 0.996; and (iv) The NSE computed by LMO estimators of EVD applied in modelling the AMR vary from 96.4% to 98.7% for Afzalpur, 92.4% to 96.3% for Aland and 90.8% to 99.1% for Kalaburagi. Based on RMSE values, it was found that EV1 (LMO) is better suited distribution for modelling the AMR of Aland whereas GEV (LMO) for Afzalpur and Kalaburagi. The study suggested that the estimated 1-day maximum rainfall by EV1 (LMO) for Aland whereas GEV (LMO) for Afzalpur and Kalaburagi can be considered for the planning and design of water resources projects in the respective sites.
Correlation coefficient, Extreme Value Type-1, Generalized extreme value, Mean squared error, L-Moments, Model efficiency, Rainfall