View article

The budget of atmospheric secondary organic aerosol (SOA) is very uncertain, with recent estimates suggesting a global source of between 12 and 1820 Tg (SOA) a⁻¹. We used a dataset of aerosol mass spectrometer (AMS) observations from 34 different surface locations to evaluate the GLOMAP global chemical transport model. The standard model simulation (which included SOA from monoterpenes only) underpredicted organic aerosol (OA) observed by the AMS and had little skill reproducing the variability in the dataset. We simulated SOA formation from biogenic (monoterpenes and isoprene), lumped anthropogenic and lumped biomass burning volatile organic compounds (VOCs) and varied the SOA yield from each precursor source to produce the best overall match between model and observations. We assumed that SOA is essentially non-volatile and condenses irreversibly onto existing aerosol. Our best estimate of the SOA source is 140 Tg (SOA) a⁻¹ but with a large uncertainty range which we estimate to be 50–380 Tg (SOA) a⁻¹. We found the minimum in normalised mean error (NME) between model and the AMS dataset when we assumed a large SOA source (100 Tg (SOA) a⁻¹) from sources that spatially matched anthropogenic pollution (which we term antropogenically controlled SOA). We used organic carbon observations compiled by Bahadur et al. (2009) to evaluate our estimated SOA sources. We found that the model with a large anthropogenic SOA source was the most consistent with these observations, however improvement over the model with a large biogenic SOA source (250 Tg (SOA) a⁻¹) was small. We used a …