A previous investigation of large-scale grain-size sorted patterns on inner continental shelves used an exploratory numerical model that represented with simple parameterizations what was hypothesized to be a key relationship between grain size and bottom roughness from ripples. We now attempt to reproduce the results of that initial model using more explicit representations of processes at smaller scales. The new model produces sorting and morphology patterns that agree qualitatively with the earlier study, consistent with the original hypothesis. However, the new results differ quantitatively from the earlier model, and results also differ among alternative models with different parameterizations of the sediment transport details.We conclude that the existing parameterizations for ripple geometries and near-bottom flow do not properly treat the conditions under which sorted bedforms occur. Our attempts to use both top-down models and more explicit numerical reductionist models provide an instructive case study highlighting the advantages and pitfalls of different modeling strategies for dealing with interactions spanning a range of scales. © 2006 Taylor & Francis Group.