In this work, we constructed two typical non-spherical dust models with medium volume size of 0.2 µm and 2 µm and assorted shape parameters. Assuming a lognormal size distribution, the optical properties of each subgroup are interpolated from a non-spherical optical database developed at the Texas A & M and integrated to create the lookup tables of a fine mode and coarse mode dust models for the DT-O. Dust retrieval is achieved with two steps: first, identifying dusty pixels using a combination of near-UV (Deep blue), visible, and thermal infrared (TIR) wavelength spectral tests; second, choosing the new dust models in the retrieval process. The retrieved dust AOD is compared to retrievals from a small set of AERONET stations.