Continuous efforts are being made to improve the procedure for estimation of a stage discharge curve in a compound channel over the last thirty years. The main reason for making these efforts is the fact that the classical method – the divided channel method, which is widely used by the hydraulic engineering community, overestimates both the total discharge and subsection discharges by more than 60% for high floodplain submergence ratios (> 0.3), as proven by a number of laboratory experiments.
One of the methods proposed in the late 1990ies by Bousmar and Zech is the exchange discharge method (EDM). The method was successfully tested against the experimental data for uniform and non-uniform flows in two-stage laboratory canals with smooth and rough floodplains. However, apart from its validation in River Sambre in Belgium, there is a little evidence on its application in rivers, whose non-prismatic compound channels have cross-sections of arbitrary shape and variable floodplain widths. Thus, this paper aims at assessing performance of the EDM in estimating stage-discharge curve at a river gauging station based either on the flood stage records or flood marks. Two gauging stations in Serbia with different floodplain layouts are chosen to assess the contribution of different terms that describe momentum transfer between the main channel and flood plains to the total energy loss. These are the turbulent momentum flux and the momentum flux due to mass exchange. Attention will be paid to the assessment of the contribution and the significance of the second one, since it could be expected that it might have a significant role in the total head loss in a compound channel in case of simultaneously converging or diverging floodplains, or alternately converging and diverging opposite floodplains. Results will be discussed in terms of discharge distributions, and contributions of different sources of energy dissipation to the total head loss, i.e. the turbulent momentum flux and the momentum flux due to mass exchange.