Photochemical Modeling for the Gulf of Mexico Air Quality Study

Oil and gas exploration and production activities in the Gulf of Mexico produce emissions of ozone and fine particulate matter (PM2.5) precursors which can contribute to on-shore pollutant concentrations in the southeastern U.S. Of particular concern are impacts to urban areas where air quality currently exceeds or nearly exceeds National Ambient Air Quality Standards, and pollutant impacts to protected environments including designated Class I areas such as the Bretton Wilderness and sensitive Class II areas such as the Gulf Islands National Seashore. While meteorological (WRF) and photochemical (CMAQ and CAMx) modeling have been used to evaluate ozone and PM2.5 air quality in urban areas such as Houston and Baton Rouge in support of local air quality management activities, regional scale modeling specifically focusing on impacts from offshore oil and gas sources in federal waters have not been performed in recent years with current models. In this study, we applied the CMAQ and CAMx photochemical models over a pair of two-way nested modeling grids: a 4 km horizontal resolution grid centered over the western Gulf of Mexico and surrounding coastal areas and a 12 km resolution grid covering the entire southwestern U.S. and northeastern Mexico. Boundary conditions for the outer (12 km) grid were obtained from a 36 km resolution grid simulation covering the contiguous states (CONUS), northern Mexico and southern Canada. Boundary conditions for the 36 km simulation were obtained from a global photochemical model (GEOS-Chem) simulation. The CMAQ and CAMx model simulations were driven with 4, 12, and 36 km resolution meteorological fields from a 12 month (calendar year 2012) WRF simulation. While the ultimate objective of the Gulf of Mexico Air Quality Study is to evaluate potential air quality impacts of proposed new off-shore oil and gas lease sales using simulations with future year emission inventories, we focus here on the supporting base-case simulation of conditions during 2012. Initial modeling results for 2012 revealed several areas where improvements in model performance in reproducing observed pollutant concentration levels over the extensive on-shore air monitoring network were needed. While model performance was generally consistent with that achieved in other similar regional simulations used for regulatory analyses, two key performance issues were targeted for further analysis. The primary challenges included a wide-spread tendency towards over prediction of low and moderate ozone concentrations throughout the southeastern states during Q3 (July – September) which was found to be especially severe in coastal areas along the central and eastern Gulf, and large over prediction of particulate nitrate in CAMx which occurred primarily along the coast but also extended well inland. The ozone over prediction bias is consistent with results from other studies. It has been suggested that this bias is at least partially associated with a bias in the GEOS-Chem boundary conditions that may in turn be at least partially due to the absence of ozone loss via sea surface halogen (bromine and iodine) emissions in the GEOS-Chem run used in our initial analysis. Model runs with reduced ozone and ozone precursor boundary conditions ameliorated the bias to some extent but a positive bias persisted in most areas. Some bias reduction has now also been achieved using updated GOES-Chem boundary conditions based on the current GEOS-Chem model with halogen chemistry included. For particulate nitrate, sensitivity tests confirmed that the positive nitrate bias in the initial CAMx results were associated with an overabundance of sea salt aerosol (SSA) which resulted in an overabundance of Na+ for neutralizing nitrate ions via chloride ion substitution. The excess SSA was found to be a result of both high SSA emissions input to CAMx and the assumption that a fraction (set at 20% by mass) of dry “coarse mode” (i.e., 0.5– 4 µm) SSA predicted by GEOS-Chem at the CAMx domain boundaries should be assigned to the CAMx SSA species and treated as fine mode aerosol. Improvements to the sea salt emissions preprocessor used for the CAMx simulations are described which, together with the elimination of GEOS-Chem coarse mode SSA in the GEOS-Chem to CAMx species mapping, significantly reduced the Na and associated particulate nitrate bias. Accurate prediction of particulate nitrate is especially important for assessing visibility and nitrate deposition impacts at Class I and sensitive Class II areas.