Simulating reactive nitrogen, carbon monoxide, and ozone in California during ARCTAS-CARB 2008 with high wildfire activity

- We evaluate model simulations for O3, CO, total and speciated NOy in California.
- Simulating wildfire impacts is still challenging even with updates to fire emissions.
- Model biases for speciated NOy are larger than for total NOy.
- Emissions and chemistry near dairy farms and oil fields are not well understood.
- Modeled O3 and HNO3 were significantly biased low off shore of the South Coast.

Predictions of O3, CO, total NOy and individual NOy species (NO, NO2, HNO3, PAN, alkyl nitrates and aerosol nitrate) from a fine resolution regional air quality modeling system for the South Coast Air Basin (SoCAB) and San Joaquin Valley Air Basin (SJVAB) of California are presented and evaluated for the 2008 ARCTAS-CARB campaign. The measurements of the chemical compounds from the fire plumes during the field campaign allow for the evaluation of the model's ability to simulate fire-influenced air masses as well. In general, the model successfully simulated the broad spatial distribution of chemical compounds in both air basins as well as the variation within the basins. Using inventories that reflect 2008 emissions levels, the model performed well in simulating NOx (NO + NO2) in SoCAB. Therefore, the under prediction of O3 over these areas is more likely caused by uncertainties with the VOC emissions, chemistry, or discrepancies in the meteorology. The model did not capture the relatively high levels of O3, and some reactive nitrogen species that were measured off shore of the SoCAB, indicating potential missing sources or the transport from on shore to off shore was not successfully captured. In SJVAB, the model had good performance in simulating different chemical compounds in the Fresno and Arvin areas. However, enhanced concentrations of O3, NOx, HNO3 and PAN near dairy farms were significantly underestimated in the model. Negative biases also exist for O3 and HNO3 near oil fields, suggesting larger uncertainties associated with these emission sources. While the model simulated the total NOy mixing ratios reasonably well, the prediction for partitioning between individual compounds showed larger uncertainties in the model simulation. Although the fire emissions inventory was updated to include the latest emissions estimates and speciation profiles, our model shows limited improvement in simulating the enhancement of O3, CO, and PAN under fire impact as compared to a previous version of the modeling system. Further improvements in simulating fire emissions, especially the timing and the plume injection heights, are desired in order to better simulate the impact of fires.