Optimization methods are used to estimate parameters required for routing floods through open compound channels. Besides initial and boundary flow conditions, data required especially include, cross-sectional area (A) of flow and conveyance (K) as functions of flow depth (y) for a representative cross-section of the study reach. Thus, instead of optimizing upon channel's geometric and hydraulic parameters, optimization is performed upon non-physical parameters in assumed A(y) and K(y) relationships. The optimization method selected for this application is the Nelder and Mead Simplex Algorithm. The objective function is expressed in terms of the relative differences between observed and simulated stages and discharges, which are evaluated based on the complete numerical solution of St Venant equations. This approach to formulating the optimization problem was applied to unsteady flow data sets for an experimental reach of the River Main in Northern Ireland. Based on statistical analysis, simulated and observed stages were found to be in good agreement.
