Linear-scale models for discharge estimation: asymmetric compound open channel flows

In this study, three recent methods of discharge estimation are evaluated, namely, the interacting divided channel method, the modified divided channel method and the momentum transfer divided channel method, which include momentum exchange on the interface for prediction of discharge. These methods are compared intensively using 22 experimental flume data sets with six river data sets taken from the literature comprising 204 test runs in total. These data sets are categorised as asymmetric compound channels with a wide range of width ratio (B/b), aspect ratio (b/h) and bed slope (⁠$So$⁠), where $B$ is the total width of channel at the bankfull condition, $b$ is the main channel bed width and $h$ is the bankfull main channel height. All these three methods are capable of predicting both overall and zonal discharge for small- to large-scale channels, including natural rivers. However, the interacting divided channel method is the least erroneous, with an average percentage error of 5.83%. Further investigations should include in-depth discussion on zonal discharge prediction, particularly for floodplain discharge, which has been an area of concern and interest in the areas of the water environment and sustainable design of rivers.