Comparison of Calibrated Empirical and Semi-Empirical Methods for Bedload Transport Rate Prediction in Gravel Bed Streams

• Published in: Journal of hydraulic engineering (New York, N.Y.) N.Y.), 2018-07, Vol.144 (7)
• Main Authors: Hinton, Darren, Hotchkiss, Rollin H, Cope, Michael
• Format: Article
• Language: English
• Subjects: Bed load; Calibration; Formulae; Formulas (mathematics); Gravel; Historical account; Hydrologic data; Logarithms; Mathematical models; Performance evaluation; Predictions; Sediment; Sediment transport; Sediments; Streams; Technical Papers; Transport
• ISSN: 0733-9429; 1943-7900
• DOI: 10.1061/(ASCE)HY.1943-7900.0001474

AbstractThe performance of seven sediment transport equations for bedload transport is compared using almost 2,600 of more than 8,000 measurements from a recent compilation. Named equations tested include the Meyer-Peter Muller, Barry, Pagosa good condition, Wilcock, Parker (both calibrated and uncalibrated), Recking, and that of Elhakeem and Imran. The purpose of the tests was to evaluate the performance of several empirical and semiempirical formulae using a single calibration point relative to three uncalibrated equations. The seven equations were included because they either have a calibration procedure already developed, are used frequently in practice, are historically foundational in the field, or have recently been proposed. Results are expressed in the root mean square error of the logarithms (RMSEL) and the relative mean error (RME) and show that the Pagosa good and Barry equations best predict bedload sediment transport (RMSEL of 0.02 and 0.02, respectively). The Pagosa good equation requires a data point for bankfull discharge and the corresponding bedload transport. The uncalibrated Recking (2013) equation resulted in lower errors than two of the calibrated formulae (Wilcock 2001 and Parker 1990) and was not far behind the calibrated Elhakeem and Imran (2016) formula. The Meyer-Peter Muller and uncalibrated Parker (1990) equations performed the worst (RMSEL of up to 0.85 and 0.86, respectively). The results herein demonstrate: (1) empirical formulae were more successful at predicting bedload transport than semiempirical alternatives, (2) a single calibration point significantly improves the predictive accuracy of any formula, and (3) calibration cannot compensate for all the shortcomings of a model.