Improving Aerosol Simulations: Assessing and Improving Emissions and Secondary Organic Aerosol Formation in Air Quality Modeling

Both long-term and short-term exposure to fine particulate matter (PM2.5) has been shown to increase the rate of respiratory and cardiovascular illness, premature death, and hospital admissions from respiratory causes. It is important to understand what contributes to ambient PM2.5 level to establish effective regulation, and air quality model can provide guidance based on the best scientific understanding available. However, PM2.5 simulations in air quality models have often found performance less than desirable, particularly for organic carbon levels. Here, some of major shortcomings of current air quality model are addressed and improved by using CMAQ, receptor models, and regression analysis. CMAQ modeling is performed for two months (July 2001 and January 2002) for the continental U.S., and detailed analysis of source apportionment results and scaling factors are conducted in the southeastern U.S.;Secondary organic aerosol (SOA) is another topic that is investigated since organic carbon is one of major components of PM2.5 in the U.S. especially in summer. CMAQ studies on organic aerosol usually underestimate organic carbon with larger than a 50% bias. Formation of aged aerosol from multigenerational semi-volatile organic carbon is added to CMAQ version 4.5, significantly improving performance of organic aerosol simulations. An increase in SOA due to aged aerosol is maximum in the south U.S., with maximum value of 8 mugm-3. In the Southeast, SOA contribution is estimated as 70% of total organic carbon. Aged aerosol also decreased discrepancies between measured and simulated hourly OC concentrations.;Detailed source apportionment of PM2.5 is performed using the CMAQ-tracer method suggests that wood combustion (10% of total PM 2.5 in summer and 25% in winter in the Southeast), fugitive dust (7 to 10% in the Southeast), fuel combustion (10% in winter) and mobile sources (5% in both seasons) are the largest sources of PM2.5, followed by meat cooking and industrial processes. Source impacts of PM2.5 simulated from CMAQ and resolved in four receptor models are compared to each other at the Southeastern Aerosol Research and Characterization (SEARCH) study monitoring sites. CMAQ identified the extended number of sources (28 vs. typically 10 in receptor models) with good performances of PM2.5 simulations with less temporal variations in source contributions than and disagreement with those from receptor models. Discrepancies between results from CMAQ and receptor models come from biases in input data for each model and limitation in model mechanisms. Biases in CMAQ modeling is decreased by investigating emission estimates using tracer species, such as organic molecular markers and trace metals that are used in receptor models. Comparison of simulated and observed tracer species shows some consistent discrepancies, which enables us to quantify biases in emissions and improve CMAQ simulations. For example, PM2.5 emissions from biomass burning is overestimated by 100% in January 2002, those from mobile sources are underestimated by 50% and more in both seasons. Biases in fugitive dust emissions are largest, especially in winter.