Movable bed resistance to the flow in an alluvial river is fundamental to river dynamics and riverbed evolution

playing a critical role in calculations of flood routing and sediment transport. This paper discusses the traditional method and its significant limitations in using the techniques of hydraulic radius separation and slope superposition

based on a systematic review of previous resistance methods in literature and their underlying assumptions. Specifically

under nonrigorous iterative closure calculations

these methods often yield an overestimated grain hydraulic radius—sometimes exceeding the measured total hydraulic radius—resulting in physical anomalies such as a negative bedform hydraulic radius. Model tests on the Yellow mainstream indicate that the friction loss caused solely by bed grains is a non-existent scenario

and that grain resistance and bedform resistance are interdependent and therefore their simple superposition is not reasonable. To address this

we adopt a high-accuracy turbulent eddy model velocity formula that is able to resolve those theoretical defects inherent in the classical logarithmic velocity law

and take the river as an approximate uniform flow (with an energy slope equal to the bed slope). Then

we apply the continuity equation in closure calculations

using the inputs of discharge

channel width

and bed sediment size

and thereby determine the flow depth and velocity that should reflect the real flow resistance. Validation against extensive field data from the wide and narrow reaches of the lower Yellow mainstream shows that this new method yields a satisfactory accuracy for engineering applications

achieving a meaningful switching from empirical or semi-empirical estimations to theoretical calculations of movable river bed resistance. Compared to Einstein’s method

our new velocity formula overcomes its deficiencies in calculating the two layers near water surface and riverbed

resulting in a higher accuracy. In the case of imbalance between riverbed erosion and deposition

the turbulent eddy model yields validation water depth correlation coefficients of 0.85 and 0.96 for the wide and narrow reaches

respectively

better than those of Einstein's formula of 0.84 and 0.95; Its relative errors are 14.4% and 14.4%

lower than 18.6% and 18.5% of Einstein's

respectively. If sidewall resistance is considered using Einstein’s approach

its validation performance improves for the wide reach

and its overall validation accuracy can be further enhanced. However

in the case of narrow reach

the validation accuracies of both formulas decrease

and the validation performance of Einstein's is quite unsatisfactory.