We used the regional air quality model CMAQ to simulate organic aerosol (OA) concentrations over the Pearl River Delta region (PRO) and compared model results to measurements. Our goals were (1) to evaluate the potential contribution of the aqueous reactive uptake of dicarbonyls (glyoxal and methylglyoxal) as a source of secondary organic aerosol (SOA) in an urban environment, and (2) to quantify the sources of SOA in the PRO in fall. We improved the representation of dicarbonyl gas phase chemistry in CMAQ as well as added SOA formation via the irreversible uptake of dicarbonyls by aqueous aerosols and cloud droplets, characterized by a reactive uptake coefficient gamma = 2.9 x 10(-3) based on laboratory studies. Our model results were compared to aerosol mass spectrometry (AMS) measurements in Shenzhen during a photochemical smog event in fall 2009. Including the new dicarbonyl SOA source in CMAQ led to an increase in the simulated mean SOA concentration at the sampling site from 4.1 mu g m(-3) to 9.0 mu g m(-3) during the smog event, in better agreement with the mean observed oxygenated OA (OOA) concentration (8.0 mu z m(-3)). The simulated SOA reproduced the variability of observed OOA (r = 0.89). Moreover, simulated dicarbonyl SOA was highly correlated with simulated sulfate (r = 0.72), consistent with the observed high correlation between OOA and sulfate (r = 0.84). Including the dicarbonyl SOA source also increased the mean simulated concentrations of total OA from 8.2 mu g m(-3) to 13.1 mu g m(-3) closer to the mean observed OA concentration (16.5 mu g m(-3)). The remaining difference between the observed and simulated OA was largely due to impacts from episodic biomass burning emissions, but the model did not capture this variability. We concluded that, for the PRD in fall and outside of major biomass burning events, 75% of the total SOA was biogenic. Isoprene was the most important precursor, accounting for 41% of the total SOA. Aromatics accounted for 13% of the total SOA. Our results show that the aqueous chemistry of dicarbonyls can be an important SOA source, potentially accounting for 53% of the total surface SOA in the PRD in fall.