The retrieval of aerosol optical depth (τ _a) over land by satellite remote sensing is still a challenge when a high spatial resolution is required. This study presents a tool that uses satellite measurements to dynamically identify the aerosol optical model that best represents the optical properties of the aerosol present in the atmosphere. We use aerosol critical reflectance to identify the single scattering albedo of the aerosol layer. Two case studies show that the Sao Paulo region can have different aerosol properties and demonstrates how the dynamic methodology works to identify those differences to obtain a better τ _a retrieval. The methodology assigned the high single scattering albedo aerosol model ( _o(λ = 0.55) = 0.90) to the case where the aerosol source was dominated by biomass burning and the lower _o model ( _o (λ = 0.55) = 0.85) to the case where the local urban aerosol had the dominant influence on the region, as expected. The dynamic methodology was applied using cloud-free data from 2002 to 2005 in order to retrieve τ _a with Moderate Resolution Imaging Spectroradiometer (MODIS). These results were compared with collocated data measured by AERONET in Sao Paulo. The comparison shows better results when the dynamic methodology using two aerosol optical models is applied (slope 1.06 ± 0.08 offset 0.01 ± 0.02 r² 0.6) than when a single and fixed aerosol model is used (slope 1.48 ± 0.11 and offset −0.03 ± 0.03 r² 0.6). In conclusion the dynamical methodology is shown to work well with two aerosol models. Further studies are necessary to evaluate the methodology in other regions and under different conditions.