Abstract
This paper presents an empirical, non-linear, multivariable regression model for predicting discharge in smooth asymmetric compound rectangular channel. The model is developed using experimental discharge data generated from testing nine different channel cross-sections with varied geometric dimensions. The predictive strength of the developed regression model is validated using several major statistics. All deployed statistics have indicated that the developed model is highly significant. In addition, the area method has been used to validate the model’s discharge predictive strength. The area method predicts discharge mainly based on the cross-section geometry and apparent shear stress. As obtained from the literature, a sample of three different regression-based models has been used to estimate the apparent shear stress. Therefore, three different sets of discharge have been predicted using the area method. The four sets of discharge predicted using the developed regression model and area method have been compared to their corresponding experimental values using the sum of squared errors (SSE). The outcome is that seven channel cross-sections out of nine tested ones resulted in minimum SSE values when discharge predicted using the developed regression model.
Acknowledgments
The authors are thankful and acknowledge the financial support by the Research Committee and Academic Affairs at Birzeit University, Palestine.