I nik'd Sliilcs Kmironmcnlnl lYolcclion A»cnc\ Office of Research ;iikI l)c\clo|>menl National Kxposure Research Laboratory Research Abstract (io\ci'iimcni I'cifoi'ni;iiKc Rcsulls \cl(ioal Clean \ir Significant Research Findings: Operational Evaluation of Atmospheric Mercury Simulation Using the Community Multiscale Air Quality (CMAQ) Model Scientific Problem Many areas of the United States currently have public advisories and Policy Issues regarding the consumption of mercury-contaminated fish. The atmosphere is known to be the primary source of mercury entering aquatic ecosystems. The Community Multiscale Air Quality (CMAQ) numerical simulation model is being expanded to help identify the air emission sources responsible for this atmospheric deposition of mercury. Evaluation of simulation models by comparison of their simulation results to actual observations is necessary to establish an appropriate level of confidence in the models and to identify missing or inaccurate process information so that model improvements can be made. The testing performed for this study has shown that CMAQ mercury simulations of mercury wet deposition are more accurate in the cool seasons when precipitation is primarily from large-scale weather features, but less accurate when small-scale convective precipitation is most prevalent. Measurement of mercury dry deposition has only been attempted on an experimental basis, and model testing for this pathway for atmospheric mercury deposition is not yet feasible. Research Approach New CMAQ software has been developed to simulate the emission, transport, chemical and physical transformation, and wet and dry deposition of atmospheric mercury. The CMAQ mercury model (CMAQ-Hg) defines atmospheric mercury in terms of three separate species, elemental mercury (Hg"), reactive gaseous mercury (RGM) and particulate mercury (Phg). Anthropogenic emissions of mercury are based on an emissions inventory developed at the U.S. Environmental Protection Agency's (EPA) Office of Air Quality Planning and Standards and chemical/physical emissions speciation assumptions developed at EPA's National Exposure Research Laboratory (NERL). Emissions of sulfur dioxide (S02), volatile organic compounds (VOCs), halogen gases, and carbon-rich particulate aerosols are believed to have an important effect on chemical and physical transformations of mercury in air and in cloud water. Thus, the CMAQ's "one-atmosphere" approach involving comprehensive multi-pollutant simulation is well National Exposure Research Laboratory — October 2001 ------- suited to the study of atmospheric mercury. In addition to the pre- existing CMAQ model chemistry, the CMAQ-Hg simulates four gas- phase reactions, seven aqueous-phase reactions, six aqueous dissociation equilibria, and two gas-liquid partitioning equilibria involving mercury and mercury compounds based on information published in the peer-reviewed scientific literature. In addition to chemical reactions, the CMAQ-Hg also simulates the physical adsorption of various mercury compounds to carbon aerosol suspended in cloud water. Estimates of total mercury wet deposition from CMAQ- Hg simulations have been compared to weekly observations from the Mercury Deposition Network (MDN) obtained at eleven sites in the central and eastern United States during two four-week periods in 1995. Results and For the spring period from April 4 to May 2, CMAQ-Hg wet deposition Implications results were compared to 28 weekly MDN observations from eight sites, and a Pearson correlation factor of 0.66 was computed. The least- squares linear regression showed an intercept close to zero, but the slope of the regression line was only 0.69, suggesting that the CMAQ- Hg was generally underestimating wet deposition of mercury for this period. However, the mean CMAQ-Hg value was greater than the mean of the MDN observations due to a few model estimates of large Hg wet deposition when observations showed very little. For the summer period from June 20 to July 18, CMAQ-Hg wet deposition results for total mercury were compared to 35 weekly MDN observations from 11 sites, and a Pearson correlation factor of 0.33 was computed. The least-squares linear regression showed a slope similar to that found for the spring period, but the y-axis intercept was quite large. Given the small correlation factor, this least-squares regression was probably not a strong indicator of model skew or bias. When the linear regression y-axis intercept was forced to zero, the slope was 1.20 and overestimation of wet deposition was indicated for the summer period. The general indication for the summer period was a weak relationship between observed and modeled Hg wet deposition. It was found that the difference in model accuracy between the spring and summer periods was largely due to difference in the accuracy of the precipitation estimates from the CMAQ meteorological driver. Comparison of observed and modeled precipitation amounts at the MDN sites showed a Pearson correlation factor of 0.80 for the spring period, but only 0.13 for the summer period. A tendency for underestimation of precipitation during the spring period was also found. During the summer period, when most of the precipitation was from small thunderstorms rather than large-scale weather systems, the National Exposure Research Laboratory — October 2001 ------- precipitation was simply not being defined accurately at the 36-km horizontal resolution used in the CMAQ-Hg simulations. Completion of this CMAQ-Hg model evaluation satisfied Annual Performance Measure #39 (Operational Evaluation of the CMAQ Mercury Model) under the Government Performance and Results Act (GPRA) Goal 1 (Clean Air). The results suggest that the CMAQ-Hg model applied using a 36-km horizontal resolution is capable of simulating atmospheric mercury with moderate accuracy when the majority of precipitation is from large-scale weather features. However, when a significant fraction of precipitation is from small- scale features like thunderstorms and isolated showers, a more refined horizontal resolution is required. This project was aided by model application and programming support obtained through inter-agency agreement #DW47938686 with the GSA, and by NERL's participation in a mercury model intercomparison project sponsored by the European Monitoring and Evaluation Programme (EMEP). Examples of recent publications from this study include: Bullock, O. R., Jr., and Brehme, K.A. Description and Evaluation of Atmospheric Mercury Simulation Using the CMAQ Model. Submitted to Atmospheric Environment Future Research Further CMAQ-Hg model development and evaluation will be conducted as additional process information for atmospheric mercury becomes available through field study and basic physicochemical research described in EPA's Mercury Research Strategy (http://www.epa.gov/ORD/NRMRL/mercury/) Questions and inquiries can be directed to: O. Russell Bullock, Jr. US EPA, Office of Research and Development National Exposure Research Laboratory Research Triangle Park, NC 27711 Phone: 919/541-1349 E-mail: bullock.russell@epa.gov Research Collaboration and Publications Contacts for Additional Information National Exposure Research Laboratory — October 2001 ------- |