Evaluation of the Community Multiscale Air Quality (CMAQ) model v5.0 against size-resolved measurements of inorganic particle composition across sites in North America
- 1Atmospheric Modeling and Analysis Division, National Exposure Research Laboratory, Office of Research and Development, US Environmental Protection Agency, Research Triangle Park, NC, USA
- 2Air Quality Assessment Division, Office of Air Quality Planning and Standards, US Environmental Protection Agency, Research Triangle Park, NC, USA
- 3International Centre for Integrated Mountain Development, Kathmandu, Khumaltar, Lalitpur, Nepal
- 4Department of Atmospheric Science, Colorado State University, Fort Collins, CO, USA
- 5Air Quality Research Division, Science and Technology Branch, Environment Canada, Toronto, Ontario, Canada
Abstract. This work evaluates particle size–composition distributions simulated by the Community Multiscale Air Quality (CMAQ) model using micro-orifice uniform deposit impactor (MOUDI) measurements at 18 sites across North America. Size-resolved measurements of particulate SO42−, NO3−, NH4+, Na+, Cl−, Mg2+, Ca2+, and K+ are compared to CMAQ model output for discrete sampling periods between 2002 and 2005. The observation sites were predominantly in remote areas (e.g., National Parks) in the USA and Canada, and measurements were typically made for a period of roughly 1 month. For SO42− and NH4+, model performance was consistent across the USA and Canadian sites, with the model slightly overestimating the peak particle diameter and underestimating the peak particle concentration compared to the observations. Na+ and Mg2+ size distributions were generally well represented at coastal sites, indicating reasonable simulation of emissions from sea spray. CMAQ is able to simulate the displacement of Cl− in aged sea spray aerosol, though the extent of Cl− depletion relative to Na+ is often underpredicted. The model performance for NO3− exhibited much more site-to-site variability than that of SO42− and NH4+, with the model ranging from an underestimation to overestimation of both the peak diameter and peak particle concentration across the sites. Computing PM2.5 from the modeled size distribution parameters rather than by summing the masses in the Aitken and accumulation modes resulted in differences in daily averages of up to 1 μg m−3 (10 %), while the difference in seasonal and annual model performance compared to observations from the Interagency Monitoring of Protected Visual Environments (IMPROVE), Chemical Speciation Network (CSN), and Air Quality System (AQS) networks was very small. Two updates to the CMAQ aerosol model – changes to the assumed size and mode width of emitted particles and the implementation of gravitational settling – resulted in small improvements in modeled size distributions.