Distributed hydrological models are based on the discretization of spatial information into homogeneous computational units (eg, sub-catchment, grid). Watershed discretization attempts to capture the important spatial variability which controls the hydrological response of a basin, while simultaneously minimizing model complexity. This process may result in a level of information loss relative to the raw data (eg, DEM information) and thus the structure uncertainty of a hydrological model. To justify a discretization scheme, it is important to understand the impacts of the associated information loss.