Developing Stage-Discharge Relations using Optimization Techniques

Abstract

Accurate estimation of discharge in rivers forms the basis of efficient flood management and surface water planning. However, continuous measurement of discharge in a stream is impractical, therefore, discharge in natural streams is generally estimated using the stage-discharge relationship. The reliability of discharge predicted from rating curve depends on the accuracy of the method used in developing stage-discharge relationship. Generalized Reduced Gradient (GRG) solver has been reported as a powerful optimization tool for handling nonlinear optimization problems, therefore, present work is an implementation of Generalized Reduced Gradient technique to determine rating curve parameters. Spreadsheet based non-linear optimization has been applied to model the rating curve for two gauging sites viz. Walton site on Choctawhatchee river and Philadelphia site on Schuylkill river, USA. The stage-discharge relationships for the two sites were also developed by using Genetic Algorithm (GA) and classical rating curve approach rating curve. Comparison of results shows that rating curve equations developed by using GRG technique are as effective as those developed by GA and more accurate than classical rating curve approach. The goodness of fit criteria for both the sites shows that the rating curves obtained by using GRG technique fit the observed data better than classical rating curve approach.