We use the SIBI to classify sea ice scenes in order to quantify the effect of surface brightness on the anisotropy. For clear-sky scenes we use 3 SIBI bins. For overcast scenes we use a monthly map of the SIBI to classify the surface below the clouds. For overcast scenes we also split the scenes into liquid and water clouds, and use a linear regression between the reflectance and optical depth in order to calculate the anisotropic factor used in the radiance-to-flux inversion. We evaluate the performance of the angular dependence models (ADM) by comparing their ability to predict the measured CERES radiances in latitude/longitude boxes. Using this method we find the RMS difference decreases from 8% to 7% in May and from 17% to 12% in July, when using the SIBI ADMs instead of the existing ADMs. Additional tests using multi-angle data from the MISR instrument show a reduction in the standard deviation of instantaneous fluxes retrieved from different viewing zenith angles from 7.3%, using the existing ADMs, to 6.6% when using the SIBI based ADMs. This suggests that using the SIBI to classify sea ice scenes will lead to improvements in the CERES SW fluxes.