Abstract
Quantitative evaluation of the performance of one of the most advanced mechanistic secondary organic aerosol (SOA) modules/models, the Model of Aerosol Dynamics, Reaction, Ionization, and Dissolution 2 (MADRID2) in the three-dimensional Models-3/Community Multiscale Air Quality (CMAQ), in urban air is made. Model calculations are compared for the Tokyo, Japan, metropolitan area with measurements made using an Aerodyne quadrupole aerosol mass spectrometer (Q-AMS) at an urban site for 9 days in July and August 2003. In general, model calculations reproduced absolute values and temporal variations of meteorological parameters, C-2-C-8 volatile organic compounds (VOCs), NOx (NO + NO2), inorganic aerosols, and O-3 concentrations reasonably well at this site. However, model-calculated SOA concentrations are a factor of 5 smaller than observed oxygenated organic aerosol (OOA) concentrations, and calculated total organic aerosol (OA = SOA + primary organic aerosol) concentrations are smaller by a factor of 2, indicating missing processes or sources in the current organic aerosol model calculations. On the other hand, observed features of diurnal and day-to-day variations of OOA were captured by our model calculations. Because of the large quantity of unidentified total nonmethane VOCs (NMVOCs) in urban air, a possible contribution of SOA formation from high-molecular-weight VOCs is examined through simple sensitivity studies, in which emissions are increased to account for unidentified NMVOCs. It is found that they are potentially one of the missing SOA sources, demonstrating the importance of reliable measurements of high-molecular-weight VOCs and total NMVOCs. Relationships between SOA and O3, including regional (similar to 150 x 150 km(2)) enhancements around the Tokyo metropolitan area, are also discussed.