Computational fluid dynamics (CFD) applications are increasingly utilized for modelling complex flow patterns in natural streams and rivers. Although CFD has been successfully implemented to model many complex flow situations in natural stream settings, adequately characterizing the effects of gravel and cobble beds on flow hydraulics in CFD is a difficult challenge due to the scale of roughness lengths and the inadequacy of traditional roughness representations to characterize flow profiles in situations with large roughness elements. An alternative method of representing gravel and cobble beds is presented. Appropriate drag forces associated with different grain sizes are computed and included in the momentum equations to account for the influence of a hydraulically rough bed. Comparisons with field measurements reveal reasonable agreement between measured and modelled profiles of spatially averaged velocity and turbulent kinetic energy, and model fidelity to the nonlogarithmic behaviour of the velocity profiles. The novel method of representing coarse beds expands the utility of CFD for investigating physical processes in natural channels with large bed roughness.