Use CMAQ, EPA's Community Air Quality Model, with state-of-the-art emissions, meteorological, and chemical inputs to model air quality in Central California for the summer of 2000.
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA
Although the researchers do not know what is representative in terms of meteorology and emissions, the extensive CCOS data can be used in their modeling efforts to produce results that will serve to suggest effective strategies for improving air quality and to illustrate the limitations of the current practice. It is very difficult to separate true trends from weather-induced fluctuations and the fluctuations in emissions due to variability in human behavior. It is also reasonable to assume that emission reduction controls that are meaningful for complying with the 1-hour standard may not be appropriate or even in the right direction for the 8-hour standard. The vast amount of observational analysis performed by other California researchers indicates that regimes of NOx and VOC-limitation and transitional behavior change in both time and space. This has been demonstrated for the SJV and other areas of the country, and this is not being captured in episodic modeling. Additionally, pollution is often greater on weekends than weekdays, the so-called weekend effect. By modeling over a season, researchers will be able to analyze pollution on about 18 weekends and compare it with the weekdays in a more meaningful manner.
Air quality in the San Joaquin Valley (SJV) was classified as extreme for the 1-hour standard and classified as serious for the new 8-hour standard—and thus will leave little opportunity for offsets. Today, SJV exploration and production is affected by emissions controls for improving air quality. Kern County, in the SJV, is the most oil productive county in California, generating two-thirds of all California petroleum. In 1999, oil wells in Kern County provided over 500,000 barrels of oil per day, ranking behind only Texas (1.4 million barrels per day), Louisiana (1.3 million barrels per day), and Alaska (on the order of 1 million barrels per day). In Kern County, the estimate of emissions for 1999 indicates that petroleum is responsible for 80 percent and 60 percent of all stationary source VOC (volatile organic compounds) and NOx emissions, respectively, which is about one-third of all emissions of each pollutant in the county. The threat of more-stringent air quality controls threatens to decrease present and future production in Kern and Fresno counties and thus threatens oil supply in the Western United States. This, in turn, will be of consequence for national energy security and the local, regional, and national economies.
Current practice for SIP planning involves modeling one or two high ozone episodes each of three or four days. There are over 100 days in exceedance of the new 8-hour standard and over 30 days in exceedance of the 1-hour standard in the SJV. Because of the diverse meteorology in the SJV and other regions, these exceedances will not be understood from modeling one or more episodes. Modeling a single or a small number of episodes as the basis for planning does not capture the diverse number of episodes and may not lead to an integrated strategy that will provide the most effective guidance to emissions reductions needed to meet that standard. A major impediment to modeling more episodes or even modeling a pollutant season is a lack of adequate data for model input and for evaluation of such models to ensure that they are predicting historic concentrations for the right reasons. Although a long-term, robust database is necessary to fully support seasonal modeling, the situation in Central California, with the SJV being designated as “extreme,” requires that researchers demonstrate the importance of seasonal modeling now and derive strategic information from it using the current, perhaps incomplete, database associated with summer 2000. Fortunately, many continuous monitoring stations throughout the region were put in place during that summer and have collected aerometric data continuously.
Current demands for longer-term modeling impact ozone in other parts of the country, as well as other aspects of air quality, since both visibility rules and particulate matter (PM) standards will require annual modeling. Furthermore, if we are to go to a multiple pollutant strategy, which is very much currently on the horizon of the Environmental Protection Agency, we must be able to achieve seasonal and even annual modeling for ozone. It is fortunate to have a robust seasonal data set to support such modeling, which will enable the project performer to evaluate not only performance and suggest potential strategies for control but will also enable the determination of what will actually be needed for seasonal and longer-term modeling. Because of the stochastic nature of atmospheric behavior, the project performer recognizes the need for an ensemble of events that can be analyzed to determine trends in air quality. Seasonal modeling results may begin to indicate that it is patterns that occur in space and time that need to be examined rather than what happens at a single point in time when CMAQ and other models are used to guide control decisions. Much of what researchers are learning is relevant to the application of other state-of-the-art air quality models. The sophisticated modeling tools that the researchers develop and refine are used to assess interactions among meteorology, emissions, and chemistry and to assess uncertainty. These have application to other air quality problems confronting oil and gas E&P operations as well.
Seasonal modeling is being performed to estimate how regional control strategy options for the new 8-hour, 80 ppb ozone standards change with respect to time and spatial location in Central California. The project performer’s approach is to consider a regional modeling domain and model a summer season in which multiple exceedances of the 8-hour ozone standards occur. The research for DOE focuses on the emission inventory for oil and gas production, creation of new modeling tools, reducing the computational burden of chemistry and model diagnostics, and using satellite products for model input and validation. The Central California Ozone Study (CCOS) aerometric database is used to support and benchmark modeling that is performed to understand how interactions of emissions, chemistry, and meteorology affect the sensitivity of pollution to local versus upwind sources (intra- and inter-basin transport).
Researchers have assembled model inputs and conducted a 15-day simulation for a portion of the CCOS domain for the summer of 2000. Additionally, they have conducted diagnostic simulations of various model inputs, formulations, and operational parameters and performed sensitivity calculations. They have written a new version of the Decoupled Direct Sensitivity code for CMAQ v4.5, used satellite products to begin to improve the treatment of photochemistry in the model, and examined emissions in counties that produce oil in California. Should the classification of air quality in the SJV be changed, more careful examination of oil emissions in the “other” and “area” categories will be important in the future. Parts of the studies are directed toward acquiring improvements in CMAQ photochemistry through the use of satellite products. The project performer finds MODIS surface albedo products exhibit substantial temporal and spatial variations relative to the current CMAQ treatment of albedo. Ozone production increases with surface albedo, and the absolute increase is larger at more polluted sites. Total column ozone is another input to the photochemistry module, and spatial, inter-annual, seasonal, and monthly variations are not considered. Researchers are currently examining it using products from TOMS. Total column ozone varies spatially and in a daily, monthly, seasonal, and annual fashion. It decreased over California for the period 1979 to 2005 as it did in the Eastern United States and globally. Annual modeling, as is required for PM and visibility, would be affected by these variations more than seasonal ozone modeling. In general, it would be better to include actual or averaged profile data as model input for specific time periods and regions of the country. After adjusting the default column ozone in CMAQ by TOMS data, the modeled J values indicate ozone concentration changes as large as 6 percent. Researchers have also used sensitivity analysis to determine how the limiting reagents, NOx or VOCs, change in time and space as you move through the modeling domain. They have also investigated inter- and intra-basin transport. Closest (mostly localized) emission sources affect local ozone formation most, and this is especially true for NOx emissions.
The project performer received emissions data from the California Air Resources Board (CARB) for the entire modeling domain and is working with CARB to improve emissions. The National Oceanic and Atmospheric Administration is providing meteorological data for 5-day periods on each side of the 15 days currently under consideration. Researchers will continue to investigate how limiting reagents change spatially and temporally along with the impacts of intra- and inter-basin transport of emissions. A paper is being written to be submitted for review and publication by the end of 2007.
This project was funded under the Gas Environmental Program.
Other Government Organizations Involved: California Air Resources Board (for Cal-Berkeley and LBNL staff), California Energy Commission (for LBNL research), and National Oceanographic and Aeronautic Administration (for met fields determined by NOAA staff).
Jin, L., C. Agnoux, J.W. Bao, N.J. Brown, R. Harley, X.L. Mao, S.A. Michelson, S. Tonse, and J. Wilczak, “Diagnostic and Mechanistic Evaluations of MM5-CMAQ for the Summer 2000 Central California Ozone Study,” abstract and poster presented at the 2006 American Geophysical Union (AGU) Fall Meeting, San Francisco, CA, December 2006 (LBNL Report No. 62435).
Brown, N., L. Jin, S. Tonse, and R. Harley, “Ozone Sensitivity to Emissions and Changes of Limiting Reagents,” abstract and poster presented at the 2006 AGU Fall Meeting, San Francisco, CA, December 2006 (LBNL Report No. 62436).
Mao, X., L. Jin, C. Agnoux, S. Tonse, and N. Brown, “Using MODIS Surface Albedo and TOMS Data for Improving the Description of Air Quality Model,” abstract and poster presented at the 2006 AGU Fall Meeting, San Francisco, CA, December 2006 (LBNL Report No. 62437).