4. Conclusion
We developed an empirical method for determining OA SSA over the tropical biomass burning regions. A simplified process flow chart for deriving OA SSA is described in Fig. 2. The basic concept of the method is that atmospheric fine particles are composed of sulfate þ nitrate, BC, OA, fdust, and fss. By using fAOD and AAOD data from the monthly ground-based AERONET observation, we formulated an equation for sulfate þ nitrate AOD as a function of OA SSA with the optically separated BC and OA AAODs from total AAOD and the measured BC SSA. We derived two sets of observationally constrained AAOD separation methods from C2012 and B2012. With regard to BC SSA, we used two measured values of 0.25 from Bond and Bergstrom (2006) and 0.19 from Magi (2009, 2011). For the fdust and fss AODs, a semi-observational approach was proposed that obtains their AOD by multiplying the fine-mode ratio of dust þ sea-salt AOD from the multi-model mean by the observed cAOD. Although the multi-model mean fine-mode ratio of dust þ sea-salt AOD was used, their contribution to sulfate þ nitrate AOD was relatively small. Therefore, the final equation, i.e., Eq. (5) or its applied version in Eq. (6), is predominantly an observationally constrained approach in which the equation to determine the sulfate þ nitrate AOD has a sole parameter (i.e., OA SSA). Our empirical method for determining OA SSA contains three independent uncertainty factors in (1) separating BC and OA AAODs from the total AAOD, (2) using model simulations for fdust and fss AODs, and (3) adopting the measured BC SSA. A quantitative comparison of these factors in determining sulfate þ nitrate AOD and OA SSA was presented in detail.