Abstract:
The accurate prediction of flood levels and velocities is a prerequisite to any appropriate management of
river valleys, where the mitigation of environmental, economic or human losses caused by flood events
is of paramount importance. During these events, rivers frequently acquire a compound channel
configuration.
Due to the 3D nature of compound channel flows, the stage–discharge curves are not as easily
predicted as in single channels. Despite the availability of 2D and 3D flow models that may solve this
question, 1D methods are often preferred due to the reduced data required and to the much shorter
processing time. In the last five decades, important research efforts have been devoted to the
improvement of 1D predictors of stage–discharge curves in compound channels. In this study, the
accuracy of seven of those methods is assessed by comparing their predictions with a large experimental
dataset, comprising symmetrical and asymmetrical compound channels with vertical and inclined main
channel sidewalls, and smooth and rough floodplains. To the authors’ best knowledge, this is the most
comprehensive assessment of stage–discharge predictors for straight compound channels since it
involves the highest number of predictors applied to the widest data set.
It was concluded that the methods that account for the momentum transfer between the main
channel and the floodplains display considerably better results than the traditional methods. For relative
depth (ratio between floodplain and main channel flow depths) higher than 0.25, predicted discharges
for the methods that account for the turbulent momentum exchange are within 5% of observed values.
Depending on whether the flow depth or the flow discharge is the pertinent variable, two different
methods seem to be the most appropriate to produce precise and safe predictions.
