Aerosol size distribution modeling with the Community Multiscale Air Quality modeling system in the Pacific Northwest: 1. Model comparison to observations

This paper evaluates aerosol number and mass concentrations from the Community Multiscale Air Quality (CMAQ) modeling system at the 4 km scale against detailed airborne and surface measurements from the coordinated Pacific Northwest 2001 and Pacific 2001 field campaigns. Several investigators have tested CMAQ's prediction of PM2.5 mass, but relatively little attention has been paid to its representation of other subranges of the particulate size distribution. In addition, few have examined CMAQ aerosol performance in the unique Pacific Northwest emission and meteorological conditions. Airborne measurements show an underprediction of aerosol number by a factor of 10-100, especially in the Aitken mode (i.e., ultrafine, < 100 nm diameter). This result cannot simply be explained by errors in gas-phase constituents. Errors in modeled ozone, NO, NOy, and SO2 exist, but the aerosol number underprediction is relatively constant while the gas-phase errors fluctuate in time and location. For the period of aircraft measurements, the surface PM2.5 mass normalized mean bias is 2%, -33%, and -38%. Speciated aerosol measurements reveal a large but intermittent positive bias in aerosol nitrate and organic mass. Errors in aerosol mass and composition are not consistent or large enough to explain the negative bias factor of 5 - 10 in accumulation mode particle number concentration and of 10 - 100 in total particle number concentration. The few other studies of CMAQ size distributions have found similar performance, the root of which has many potential causes and is worthy of future investigation.