A Study of PM2.5 Formation in Central California during 2013 Discover-AQ Field Experiment

Abstract:

Five high PM2.5 episodes occurred in the Central Valley of California in January and February, 2013. Two of these episodes took place during the 2013 Discover-AQ field experiment. We used observations and CMAQ model simulations to study PM2.5 formation during these episodes. The study domain covered all of central and portions of northern California. Analyses were conducted with special emphasis on the differences on the meteorology and PM2.5 components over three sub-regions: the San Francisco Bay Area (SFBA), the Sacramento area (SAC), and the San Joaquin Valley area (SJV).

The CMAQ had 15 vertical layers and 4 km horizontal grid resolution. The SAPRC99 chemical and AE5 aerosol mechanisms were used in the simulations. Meteorological inputs to CMAQ were generated using the WRF model. An available 2012 emissions inventory was used for 2013 simulations.

In all three sub-regions, the WRF model slightly under predicted wind speed while correctly predicted the wind direction. The predicted boundary layer thickness had good correlation with observed average PM2.5 concentrations, especially in SJV.

The CMAQ model reproduced all five high PM2.5 episodes. The predicted PM2.5 almost matched the observed values in the SFBA. For the two episodes captured by the Discover-AQ field experiment, CMAQ under predicted PM2.5 in the SJV area This under prediction may be attributed to the thickness of the first layer of CMAQ, which is about 32 m. The nighttime PBL height computed by WRF can be as low as 15 m in SJV during this period.

There were considerable differences in the ratio of primary to secondary PM2.5 among in the three sub-regions. Secondary PM2.5 averaged 27% of total PM2.5 in SFBA. The corresponding ratio was 36% in SAC and 45% in SJV. The biggest component of secondary PM2.5 in SJV was ammonium nitrate, which is consistent with large ammonia emissions there from dairy and feedlot operations.

We found large sensitivity of CMAQ simulated PM2.5 to the model layer structure and the PBL schemes. We plan to rerun CMAQ with additional vertical layers and finer vertical resolution near the surface to better delineate the nighttime boundary layer structure. The WRF model will also be tested with several PBL schemes. The simulated PM2.5 at upper layers will be compared with aircraft observations from the Discover-AQ campaign.