Understanding the spatial distribution of fine particle sulfate (SO42-) concentrations is important for optimizing emission control strategies and assessing the population health impact due to exposure to SO42-. Aerosol remote sensors aboard polar orbit satellites can help expand the sparse ground monitoring networks into regions currently not covered. We developed a generalized additive model (GAM) using MISR fractional aerosol optical depths (AODs) scaled by GEOS-Chem aerosol profiles to predict ground-level SO42- concentrations. This advanced spatial statistical model was compared with alternative models to evaluate the effectiveness of including simulated aerosol vertical profiles and adopting an advanced statistical model structure in terms of improving the AOD- SO42- association. The GAM is able to explain 70% of the variability in SO42- concentrations measured at the surface, and the predicted spatial surface of annual average SO42- concentrations are consistent with interpolated contours from ground measurements. Comparisons with alternative models demonstrate significant advantages of using model-scaled lower-air fractional AODs instead of their corresponding column values. The nonlinear association between SO42- concentrations and fractional AODs makes the GAM a more suitable model structure than conventional linear regressions.