Application and Validation of Regression Analysis in the Prediction of Discharge in Asymmetric Compound Channels

Abstract

A series of laboratory experiments was performed to present the overbank flow in asymmetric rectangular compound channels. For this purpose, two different sets of asymmetric models with rectangular compound cross sections were tested for a wide range of discharges. The first set consisted of nine compound cross-section models formed by a combination of three step heights and three main channel widths. The second set consisted of six compound cross section-models formed using a combination of two step heights and three main channel widths. The mean flow measurements were then related to a dimensionless parameter called the relative depth defined as the ratio of the depth above the floodplain bed to the depth above the main channel bed. The variations and interactions of the three outlined mean flows were investigated with respect to relative depth. A set of single-variable regression models has been developed for estimating the three mean flow types using relative depth as the only independent variable. Another set of multiple-variable regression models was derived using two additional dimensionless parameters, which take into account the width dimensions of the constructed asymmetric compound channel. The application of several key statistics and validation procedures indicated the high significance and reliability of the developed models in predicting the three mean flow types.