Evaluating the Effect of Domain Size of the Community Multiscale Air Quality (CMAQ) Model on Regional PM_2.5 Simulations

A growing number of urban health impact studies use Community Multiscale Air Quality (CMAQ) models for air pollution exposure estimation, although the performance of CMAQ models is likely to be affected by multiple parameters, including the configuration setting of the study domain. We presented an approach for CMAQ model uncertainty assessment with respect to domain size and reported spatial and temporal variations of CMAQ model performance over two study domains, a relatively small domain (D_S) and a large domain (D_L). Specifically, we simulated daily PM_2.5 concentrations over two domains during 2011 and quantified the difference between the model predictions. The model performance was assessed by comparing modeled PM_2.5 against measured PM_2.5 values at monitoring sites located in the region of overlap for each domain. The results suggest that the CMAQ simulations over two domains were in good agreement across the study area except in southwestern areas. We also found that the overall model performance was better for CMAQ simulations with a large domain relative to the smaller domain. Based on our findings, we recommend applying a large domain for PM_2.5 simulations, particularly for urban health risk assessments conducted over summer months, which generally contain more emissions.