Models-3/CMAQ is an important air quality modelling system, especially for particulate matter (PM) studies in the lower atmosphere. Despite its widespread applications, Models-3/CMAQ PM results are not directly comparable with size-resolved ambient measurement data, such as PM2.5, since PM is modelled by CMAQ on the basis of log-normal particle number distributions (modes) instead of size bins, and CMAQ does not output quantitative parameters of the particle distributions.

In this paper, we discuss the details of this issue and present the solution that we developed. A complete set of mathematical formulas are derived to convert the current CMAQ modal PM results into the necessary parameters of the log-normal distributions and the speciated PM concentrations in flexible size bins. The newly generated parameters for the particle distributions include the geometric mean diameters and geometric standard deviations of the three particle modes, except the preset geometric standard deviation of the coarse mode. The new PM concentrations include the number concentrations, speciated mass concentrations, and surface area concentrations in the chosen size range of the particle distributions. Special considerations for using the newly generated PM quantities in model evaluations are presented to facilitate understanding of the issue and to ensure correct use of the new PM quantities. Attention is paid to the model results that are needed for PM2.5 evaluations. Results of a Models-3/CMAQ run for the Lower Fraser Valley, which includes part of British Columbia, Canada, and Washington State, U.S., are shown as an example to help visualise the quantitative differences among various PM2.5-related statistical quantities.

In summary, the work presented in this paper converts the Models-3/CMAQ modal PM results into much more useful size-resolved quantities that will help the correct evaluation of this increasingly important modelling system. The methodology and mathematics in the paper may also be applicable to other PM models that are based on statistical particle size distributions.