Recent technological advances in remote sensing have enabled investigation of the morphodynamics and hydrodynamics of large rivers. However, measuring topography and flow in these very large rivers is time consuming, and so often constrains the spatial resolution and reach-length scales that can be monitored. Similar constraints exist for Computational Fluid Dynamics (CFD) requiring maximization of mesh- or grid-cell dimensions and implying a reduction in the representation of bedform-roughness elements that are the order of a cell’s dimensions or less, even if they are represented in available topographic data. These ‘subgrid’ elements require parameterizing, and here we aim to apply and consider the impact of roughness-length treatments that include the effect of bed roughness due to ‘unmeasured’ topography. CFD predictions were found to be sensitive to roughness-length specification. Upscaling roughness length was not found to be a particularly effective means of optimizing the agreement between model predictions and field data, because the effects of large bedforms were not well parameterized in roughness-length treatments. Changes in roughness length were shown to have a major impact upon flow routing at the channel-scale, and this was linked to the impacts of roughness length 37 upon channel-scale secondary-flow structures. These results show a poor equivalence between the roughness lengths needed in 2D models and those required for 3D models, especially in deeper areas of the river. Improved shear stress estimation was obtained with a two-layer model. However, this introduces additional parameters that are difficult to estimate a priori.