Fine-scale photochemical modeling of ozone and ammonium nitrate over California during CalNex 2010

Multiple areas of California are designated as nonattainment of the ozone and PM2.5 National Ambient Air Quality Standards making air quality (AQ) modeling for California an important aspect of national-scale modeling for EPA rulemaking. Fine-scale AQ modeling for key population centers in California is also useful to inform health and exposure studies. However, AQ modeling is challenging in California due to complex emissions, terrain, meteorology, and chemistry, and understanding model performance is difficult using routine network observations alone. In May-June 2010, the CalNex field study was conducted in California to answer important scientific questions related to air quality and climate processes. The field study provides a rich observational dataset for probing the performance of fine-scale AQ simulations and identifying causes of model performance limitations. In this study, we conduct fine-scale (4-km horizontal resolution) photochemical model simulations for California during May-June 2010 using the Community Multiscale Air Quality model with the Carbon Bond 05 and SAPRC07 gas-phase chemical mechanisms. Model predictions are then evaluated using observations from the CalNex ground, aircraft, and ship platforms in addition routine network observations. Our model evaluations focus on understanding predictions of ammonium nitrate and ozone concentrations given the importance of these pollutants to California's air quality. Model predictions of ammonium and nitrate have a slight low bias but are strongly correlated with network observations in Riverside and Bakersfield. The model tends to under-predict NH3 concentrations observed on the NOAA P3 aircraft near large NH3 sources in the San Joaquin Valley (SJV) and South Coast Air Basin (SoCAB). At the SJV ground site, NH3 predictions are of similar magnitude as observations, while NH3 tends to be over-predicted at the Pasadena ground site. Nitric acid is over-predicted during the day at the SJV ground site but is generally predicted well in Pasadena, except for under-predictions during a high concentration episode. Constrained ISORROPIA box-modeling suggests that model errors are related to total nitrate (HNO3+ NO3) and total ammonium (NH3+NH4) predictions at some hours and related to the remaining aspects of the system (i.e., T, RH, and other components) at other hours. The general observed pattern of low ozone concentration in the LA urban core and high ozone concentrations downwind is captured by the model. However, the model under-predicts peak ozone concentrations observed downwind of the urban core. These under-predictions can be understood by considering model performance for NOx and VOC concentration as well as modeled ozone production efficiencies, chemical indicators, and precursor emission rates.