Simulation of dust emission flux is very fundamental for evaluating climatic effects of dust aerosols and controlling desertification process and dust storm. The two shear stress partition models were proposed by Marticorena and Alfaro respectively for evaluating effect of roughness elements on dust production. The existing problems of them are discussed on basis of sensitive tests, previous wind tunnel and field experiments. First, the influence of porosity and spatial heterogeneity of roughness elements on shear stress partition was not taken into account because all equations about roughness length can not involve such factors as porosity and spatial heterogeneity of roughness elements. However, it is proved that these factors cannot be neglected by many experiments. Second, the result from sensitive test of Marticorena's Model is in disagreement with that of Alfaro's Model, implying that the models need to be investigated further by wind tunnel experiment. Third, regarding the roughness length to efficiently suppress wind erosion, the values predicted respectively by the models are questioned by some wind erosion experiment data, therefore the relationship between shear stress partition and roughness length need to be reassessed. In terms of these problems of the models, two approaches are proposed: improving roughness length formula and tuning empirical constants of the models by wind tunnel experiment.