The bulk velocity Ub in streams is conventionally estimated from Manning's equation, but difficulties remain in parameterizing the roughness coefficient n when the streambed is covered with vegetation. A two-layer velocity model is proposed to determine n and Ub for the submerged vegetation case. The modeled n is derived as a function of flow and vegetation properties that can be inferred from remote sensing platforms, such as canopy height, leaf area density, and flow depth. The main novelty in the proposed formulation is that the shear stress is related to the mean velocity profile by considering both ejective and sweeping motions by dominant eddies. The proposed model is tested against a large data set from the literature and is shown to perform well, particularly for rigid vegetation. Poorer model performance for flexible vegetation can be partially attributed to the shape of the assumed mean velocity profile. The roughness coefficient n is found to be robust to variations in the average spacing between canopy elements, allowing the model to be applied to heterogeneous canopies.