COMPARISON BETWEEN CHEMICAL MASS BALANCE RECEPTOR AND CMAQ PM 2.5 SOURCE APPORTIONMENT

Transcription

1 COMPARISON BETWEEN CHEMICAL MASS BALANCE RECEPTOR AND CMAQ PM 2. SOURCE APPORTIONMENT Sun-Kyoung Park 1, Amit Marmur 2, Lin Ke 3, Bo Yan 3, Mei Zheng 3, Armistead G. Russell 4 1 School of Business Administration, Hanyang Cyber University, Hangdang-dong, Sungdong-gu, Seoul, , South Korea, Helena@hycu.ac.kr 2 ENVIRON, 773 San Marin Drive, Suite 211, Novato, CA School of Earth and Atmospheric Science, Georgia Institute of Technology, Atlanta Georgia 3332 USA 4 School of Civil and Environmental Engineering, Georgia Institute of Technology, Atlanta Georgia 3332 USA ABSTRACT Source apportionment of particulate matter has been performed using many different methods. Here, a comparison is done between PM2. apportioned from the Chemical Mass Balance (CMB) receptor model using organic tracers as molecular markers and the source-based Community Multiscale Air Quality (CMAQ) model. Source apportionment is conducted at a station in Atlanta, Georgia for July 21 and January 22. PM 2. source apportionment results had moderate discrepancies, which originate from different spatial scales, fundamental limitations, and uncertainties of the two models. Both models have strengths and limitations, and each model s strengths can be utilized to help overcome the other model s limitations. KEY WORDS: Source Apportionment, PM2., Chemical Mass Balance, Molecular Marker, Community Multiscale Air Quality 1. INTRODUCTION In 1999, the National Ambient Air Quality Standard (NAAQS) for PM 2. (particulate matter with aerodynamic diameter less than 2. micrometers) was promulgated in a response to scientific studies linking elevated fine particle concentrations with adverse health effects (Dockery and Pope, 1994; Metzger et al., 24; Peel et al., 22). Effective control of PM 2. levels requires identifying and quantifying contributions by major sources. Historically, source apportionment of PM 2. largely has been performed via several receptor-modeling techniques. In particular, one of the most widely used receptor modeling techniques is the chemical mass balance (CMB) approach (Core et al., 1982; Friedlander, 1973; Watson et al., 22). In CMB, ambient chemical concentrations are expressed as the sum of products of source profiles and source contributions. This linear system of equations is solved for source contributions by weighted least square fitting. Independent to CMB approach, a bottom-up approach to source apportionment is the use of emission-based, (or source) three dimensional air quality models (AQMs). Such models solve the species conservation equation: where, c i is the concentration of species i, U is wind velocity vector, D i is molecular diffusivity of species i, R i is rate of concentration change of species i by chemical reaction, S i (x,t) is source/sink of i at location x and time t, ρ is air density, and n is the number of predicted species. The conservation equation describes the formation, transport, and fate of air pollutants, including components for processing emissions, meteorology, topography, air quality observations, and chemistry (Russell and Dennis, 2). Three-dimensional models, including California Institute of Technology (CIT) photochemical model (Held et al., 24; Kleeman and Cass, 21; Mysliwiec and Kleeman, 22; Ying et al., 24), the Urban and Regional Multiscale (URM) airshed model (Boylan et al., 22, 2), and the Community Multiscale Air Quality (CMAQ) models (Marmur et al., 2; Napelenok et al., 2; Park et al., 2), have been used for PM 2. source apportionment. This paper compared the source apportionment of PM2. using CMB and CMAQ for two complete months (July 21 and January 22) at a station in Atlanta, Georgia. The goal of this work is to analyze the reasons of discrepancy, and to utilize each model strength to overcome other model s limitations. ci ci + ( U ci ) = ρ Di + Ri ( c1, c2,..., cn, T, t) + Si( x, t) t ρ i=1, 2, 3,, n (1) 2. METHODS Source apportionment of PM 2. is performed using two different approaches: receptor models and source oriented models. The

2 receptor model chosen in this paper is Chemical Mass Balance (CMB) and the source oriented model used here is Community Multiscale Air Quality (CMAQ) model. CMB was applied to daily PM2. species concentrations measured at the Jefferson St Atlanta, Georgia (JST) (Figure 1) (Zheng et al., 27). CMAQ was run over the United States (Figure 1) for July 21 and for January 22 corresponding to the coordinated intensive monitoring periods by the US-EPA Eastern Supersite Program (ESP 1/2). One of the benefit of choosing the ESP1/2 as a modeling period is that the model results can be fully evaluated using the comprehensive monitored data (Park et al., 26). JST station 14, m (σ=.) 7,2 m (σ=.3) 2,94 m (σ=.6) 1, m (σ=.8) 74 m (σ=.9) 3 m (σ=.96) 14 m (σ=.98) 7 m (σ=.99) 18 m (σ=.997) Multi-scale Air Quality (CMAQ) model v4.3 for air quality modeling (Byun and Ching, 1999). CMAQ was applied for July 21 and January 22 over the continental United States and parts of Mexico and Canada with a 36 km grid and over the Atlanta area with a 12 km grid (Figure 1). The projection used is the unified Regional Planning Organization (RPO) national grid, which is Lambert conformal conic projection with a central meridian of 97 W, a center of latitude of 4 N, and standard parallels of 33 N and 4 N. More information of the air quality modeling system and the model evaluation results are available in elsewhere (Park et al., 26). Source apportionment using CMAQ can be done by direct sensitivity methods such as DDM-3D (Dunker, 1981; Yang et al., 1997) or by Brute Force (BF) (i.e., applying the model once with, then without the target source) (Park et al., 2), the latter of which was applied in this study. The target emission source is removed based on the source category code (SCC) in the emission inventory. Emission sources apportioned were the same as the seven emission categories chosen in CMB for comparison purposes. CMAQ can calculate mass contributions to the secondary PM 2. in addition to those to the primary PM 2.. Only mass contributions to primary PM 2. from CMAQ were compared with those from CMB since CMB only apportions primary sources. Figure 1. The JST monitoring station, and CMAQ modeling domain. Rectangles around the United States and over the Atlanta area are 36 km and 12 km grid domain, respectively. The number of vertical layer is nine with top pressure of 1hPa. 2.1 Receptor-based PM 2. apportionment using a Chemical Mass Balance (CMB) model Receptor-based source apportionment of PM 2. is performed using CMB with organic tracers as Molecular Markers (CMB- MM) (Ke et al., 2; Schauer et al., 1996; Zheng et al., 27). Mass contributions are calculated for seven sources: gasoline exhaust, diesel exhaust, road dust, wood/biomass burning, meat cooking, natural gas, and power plant emissions. CMB has a long record of use, but the co-linearity of profiles relying solely on inorganic species proved problematic in application to the southeastern United States. Thus, CMB-MM, which relies more on speciated organic compounds was developed (Schauer et al., 1996; Zheng et al., 27; Zheng et al., 22). Source profiles are expressed as normalized values to organic carbon. Hence, CMB- MM apportions mass contributions to organic carbon, then the mass contributions to PM 2. is calculated by dividing by the organic carbon to PM 2. ratio of each source. 3. RESULTS & DISCUSSION Daily PM 2. masses at the JST station apportioned using CMB- MM and CMAQ (12 km) were compared (Figure 2). Sulfate, nitrate, and ammonium masses are not shown to give more focus on primary PM 2. mass. Total primary PM 2. masses are different between the two methods as PM 2. mass from CMB-MM is the measured concentration and that from CMAQ is simulated value. Relative PM 2. mass and contributions from each source using CMAQ do not differ significantly between July 21 and January 22, whereas, those from CMB-MM vary markedly from July 21 to January 22. One of the reasons for disagreement between the two models includes uncertainty of the two models. Important sources of uncertainty in CMB results include source profiles. Studies show that mass contributions estimated from CMB are significantly different depending on source profiles chosen (Yan et al., 24). Another source of uncertainty is that CMB apportions primary mass, which is only a fraction of total PM 2. mass (Figures 2). Unknown sources are important not only because they occupy a large parts of PM 2. mass, but also because unknown sources can affect estimating known sources in CMB (Christensen, 24). 2.2 Source-based PM 2. Apportionment using the Community Multiscale Air Quality (CMAQ) model The three-dimensional air quality modeling system used for the source apportionment is EPA Models-3, which includes the Sparse Matrix Operator Kernel Emissions (SMOKE) v1. for emission processing (US-EPA, 24), the NCAR s th generation Mesoscale Model (MM) v3..3 for preparing meteorological fields (PSU/NCAR, 23), and the Community

3 (a) (b) effects or overestimation. To build multiple monitors in the area of interests may diminish the non-representativeness (in space) of CMB results although this approach would encounter additional costs. To decrease the grid size of the air quality model can enhance the spatial resolution of CMAQ results. The mass contributions calculated from CMAQ using different sizes of the grid are compared (Figure 3). The correlation coefficient (r) of.87 between the two results from the air quality model with different grid size is not that low, but not close to 1, indicates that the significant part of differences between CMB and CMAQ is due to the different spatial scale. (point vs. volume) CMAQ(36km) CMAQ(12km) Diesel Gasoline Road dust Wood burning Figure 3. Scatter plot of daily average mass contributions to PM 2. from CMAQ in July 21 and January 22 in JST. Figure 2. Daily average mass contributions to PM 2. in JST (sulfate, nitrate, and ammonium were excluded). CMB-MM, CMAQ (12km) [left to right]. Results of CMB-MM are not available on July 1, 2,, 11, 22, 24, and 28, and on January 1, 11, and 2. (a) July 21. (b) January 22 Major sources of uncertainty of CMAQ results include the emission inventory, speciation profiles, and meteorological inputs (Placet et al., 2). Currently, the emission inventory is known to be one of the more uncertain inputs (Abdel-Aziz and Frey, 24; Gilliland et al., 23; Hogrefe et al., 23; Mannschreck et al., 22; Mendoza-Dominguez and Russel, 21; Placet et al., 2; Taghavi et al., 2; Vautard et al., 23). Two recent 21 emission inventories over the southeastern United States are significantly different. Depending on sources, up to 3% of difference was observed. The different spatial scales of the two models have an important implication for use of the results. CMB is done based on the measurement, so results are specific to the monitoring location. However, mass contributions calculated from CMB may not be representative to the area if a local (spatial) concentration gradient exists. The site representativeness problem is important to epidemiological studies, because source impacts determined at the monitoring station are used to analyze the health effect of the pollutants over the area in which the monitor is located (Wade et al., 24). On the other hand, CMAQ simulates average concentrations of the grid, so results are less subject to local One method to improve the accuracy of emissions would be to calculate scaling factors of emissions via inverse modeling by incorporating measured concentrations. Indeed, both models have strengths and limitations, and each model s strength can be utilized to overcome the other model s limitations. ACKNOWLEDGEMENT This research was supported by the U.S. Environmental Protection Agency under Agreements RD , RD831761, and RD REFERENCES Abdel-Aziz, A., and H. C. Frey, 24, Propagation of uncertainty in hourly utility NOx emissions through a photochemical grid air quality model: A case study for the Charlotte, NC, modeling domain: Environmental Science & Technology, v. 38, p Boylan, J. W., M. T. Odman, J. G. Wilkinson, A. G. Russell, K. G. Doty, W. B. Norris, and R. T. McNider, 22, Development of a comprehensive, multiscale "oneatmosphere" modeling system: application to the Southern Appalachian Mountains: Atmospheric Environment, v. 36, p Boylan, J. W., M. T. Odman, J. G. Wilkinson, A. G. Russell, K. G. Doty, W. B. Norris, and R. T. McNider, 2, Integrated assessment modeling of atmospheric pollutants in the Southern Appalachian Mountains. Part

4 I: Hourly and seasonal ozone: Journal Of The Air & Waste Management Association, v., p Byun, D. W., and J. K. Ching, 1999, Science algorithms of the EPA Models-3 Community Multiscale Air Qualtiy (CMAQ) modeling system: Atmospheric modeling division, National Exposure Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC Christensen, W. F., 24, Chemical mass balance analysis of air quality data when unknown pollution sources are present: Atmospheric Environment, v. 38, p Core, J. E., J. A. Cooper, P. L. Hanrahan, and W. M. Cox, 1982, Particulate Dispersion Model Evaluation - A New Approach Using Receptor Models: Journal Of The Air Pollution Control Association, v. 32, p Dockery, D. W., and C. A. Pope, 1994, Acute Respiratory Effects of Particulate Air-Pollution: Annual Review of Public Health, v. 1, p Dunker, A. M., 1981, Efficient Calculation Of Sensitivity Coefficients For Complex Atmospheric Models: Atmospheric Environment, v. 1, p Friedlander, S., 1973, Chemical Element Balances And Identification Of Air-Pollution Sources: Environmental Science & Technology, v. 7, p Gilliland, A. B., R. L. Dennis, S. J. Roselle, and T. E. Pierce, 23, Seasonal NH3 emission estimates for the eastern United States based on ammonium wet concentrations and an inverse modeling method: Journal Of Geophysical Research-Atmospheres, v. 18. Held, T., Q. Ying, A. Kaduwela, and M. Kleeman, 24, Modeling particulate matter in the San Joaquin Valley with a source-oriented externally mixed threedimensional photochemical grid model: Atmospheric Environment, v. 38, p Held, T., Q. Ying, M. J. Kleeman, J. J. Schauer, and M. P. Fraser, 2, A comparison of the UCD/CIT air quality model and the CMB source-receptor model for primary airborne particulate matter: Atmos. Environ. accepted for publication. Hogrefe, C., G. Sistla, E. Zalewsky, W. Hao, and J. Y. Ku, 23, An assessment of the emissions inventory processing systems EMS-21 and SMOKE in grid-based air quality models: Journal Of The Air & Waste Management Association, v. 3, p Ke, L., F. Wang, E. S. Edgerton, J. J. Schauer, A. G. Russell, G. R. Cass, and M. Zheng, 2, Source Apportionment of Organic Carbon in PM2. from the SEARCH study. : Atmospheric Environment, Submitted for publication. Kleeman, M. J., and G. R. Cass, 21, A 3D Eulerian sourceoriented model for an externally mixed aerosol: Environmental Science & Technology, v. 3, p Mannschreck, K., D. Klemp, D. Kley, R. Friedrich, J. Kuhlwein, B. Wickert, P. Matuska, M. Habram, and F. Slemr, 22, Evaluation of an emission inventory by comparisons of modelled and measured emission ratios of individual HCs, CO and NOx: Atmospheric Environment, v. 36, p. S81-S94. Marmur, A., S.-K. Park, J. Mulholland, and A. G. Russell, 2, PM2. Source Apportionment Using Receptor and Source-oriented Models: Conceptual Differences and Implications for Health Studies: Proceedings of an AAAR international specialty conference -- Particulate Matter Supersites Program and Related Studies, Atlanta, Georgia Mendoza-Dominguez, A., and A. G. Russel, 21, Estimation of emission adjustments from the application of fourdimensional data assimilation to photochemical air quality modeling: Atmospheric Environment, v. 3, p Metzger, K. B., P. E. Tolbert, M. Klein, J. L. Peel, W. D. Flanders, K. Todd, J. A. Mulholland, P. B. Ryan, and H. Frumkin, 24, Ambient air pollution and cardiovascular emergency department visits: Epidemiology, v. 1, p Mysliwiec, M. J., and M. J. Kleeman, 22, Source apportionment of secondary airborne particulate matter in a polluted atmospbere: Environmental Science & Technology, v. 36, p Napelenok, S., D. S. Cohan, A. G. Russell, and Y. Hu, 2, Implementation of the direct decoupled method (DDM) into the CMAQ model for aerosol species: Proceedings of an AAAR international specialty conference -- Particulate Matter Supersites Program and Related Studies, Atlanta, Georgia Park, S.-K., C. E. Cobb, K. Wade, J. Mulholland, Y. Hu, and A. G. Russell, 26, Uncertainty in Air Quality Model Evaluation from Spatial Variation: Atmospheric Environvironment, v. 4, p. S63-S73. Park, S.-K., L. Ke, Y. B., R. A. G., and M. Zheng, 2, Source apportionment of PM2. using a three-dimensional air quality model and a receptor model: Proceedings of an AAAR international specialty conference -- Particulate Matter Supersites Program and Related Studies, Atlanta, Georgia Peel, J., P. Tolbert, M. Klein, K. Metzger, W. D. Flanders, K. Todd, J. Mulholland, P. B. Ryan, and H. Frumkin, 22, Ambient air pollution and respiratory emergency department visits in Atlanta, August 1998-August 2 (ARIES/SOPHIA): Epidemiology, v. 13, p. S124-S124. Placet, M., C. O. Mann, R. O. Gilbert, and M. J. Niefer, 2, Emissions of ozone precursors from stationary sources: a critical review: Atmos. Environ., v. 34, p PSU/NCAR, 23, PSU/NCAR Mesoscale Modeling System Tutorial Class Notes and User s Guide: MM Modeling System Version 3: Mesoscale and Microscale Meteorology Division, National Center for Atmospheric Research. Rogge, W. F., L. M. Hildemann, M. A. Mazurek, and G. R. Cass, 1994, Sources Of Fine Organic Aerosol.6. Cigarette- Smoke In The Urban Atmosphere: Environmental Science & Technology, v. 28, p B. R. T. Simoneit, 1993a, Sources Of Fine Organic Aerosol.2. Noncatalyst And Catalyst-Equipped Automobiles And Heavy-Duty Diesel Trucks: Environmental Science & Technology, v. 27, p B. R. T. Simoneit, 1993b, Sources Of Fine Organic Aerosol.3. Road Dust, Tire Debris, And Organometallic Brake Lining Dust - Roads As Sources And Sinks: Environmental Science & Technology, v. 27, p B. R. T. Simoneit, 1993c, Sources Of Fine Organic

5 Aerosol.4. Particulate Abrasion Products From Leaf Surfaces Of Urban Plants: Environmental Science & Technology, v. 27, p B. R. T. Simoneit, 1993d, Sources Of Fine Organic Aerosol.. Natural-Gas Home Appliances: Environmental Science & Technology, v. 27, p B. R. T. Simoneit, 1997a, Sources of fine organic aerosol.7. Hot asphalt roofing tar pot fumes: Environmental Science & Technology, v. 31, p B. R. T. Simoneit, 1997b, Sources of fine organic aerosol.8. Boilers burning No. 2 distillate fuel oil: Environmental Science & Technology, v. 31, p B. R. T. Simoneit, 1998, Sources of fine organic aerosol. 9. Pine, oak and synthetic log combustion in residential fireplaces: Environmental Science & Technology, v. 32, p B. R. T. Simonelt, 1991, Sources Of Fine Organic Aerosol.1. Charbroilers And Meat Cooking Operations: Environmental Science & Technology, v. 2, p Russell, A., and R. Dennis, 2, NARSTO critical review of photochemical models and modeling: Atmospheric Environment, v. 34, p a, Measurement of emissions from air pollution sources. 1. C-1 through C-29 organic compounds from meat charbroiling: Environmental Science & Technology, v. 33, p b, Measurement of emissions from air pollution sources. 2. C-1 through C-3 organic compounds from medium duty diesel trucks: Environmental Science & Technology, v. 33, p , Measurement of emissions from air pollution sources. 3. C-1-C-29 organic compounds from fireplace combustion of wood: Environmental Science & Technology, v. 3, p a, Measurement of emissions from air pollution sources. 4. C-1-C-27 organic compounds from cooking with seed oils: Environmental Science & Technology, v. 36, p b, Measurement of emissions from air pollution sources.. C-1-C-32 organic compounds from gasoline-powered motor vehicles: Environmental Science & Technology, v. 36, p Schauer, J. J., W. F. Rogge, L. M. Hildemann, M. A. Mazurek, and G. R. Cass, 1996, Source apportionment of airborne particulate matter using organic compounds as tracers: Atmospheric Environment, v. 3, p Taghavi, M., S. Cautenet, and J. Arteta, 2, Impact of a highly detailed emission inventory on modeling accuracy: Atmospheric Research, v. 74, p US-EPA, 24, SMOKE User's manual Vautard, R., D. Martin, M. Beekmann, P. Drobinski, R. Friedrich, A. Jaubertie, D. Kley, M. Lattuati, P. Moral, B. Neininger, and J. Theloke, 23, Paris emission inventory diagnostics from ESQUIF airborne measurements and a chemistry transport model: Journal Of Geophysical Research-Atmospheres, v. 18. Wade, K., A. Marmur, J. Mulholland, A. Russell, J. Peel, and M. Klein, 24, Analyses of air quality data for an emergency department study, : Epidemiology, v. 1, p. S61-S61. Watson, J. G., T. Zhu, J. C. Chow, J. Engelbrecht, E. M. Fujita, and W. E. Wilson, 22, Receptor modeling application framework for particle source apportionment: Chemosphere, v. 49, p Yan, B., M. Zheng, and A. Russell, 24, Uncertainty and Sensitivity Analysis of Chemical Mass Balance Modeling Using Organic Tracers for PM 2. Source Apportionments Proceeding of the AAAR 4 annual conference, Atlanta, GA.. Yang, Y. J., J. G. Wilkinson, and A. G. Russell, 1997, Fast, direct sensitivity analysis of multidimensional photochemical models: Environmental Science & Technology, v. 31, p Ying, Q., M. Mysliwiec, and M. J. Kleeman, 24, Source apportionment of visibility impairment using a threedimensional source-oriented air quality model: Environmental Science & Technology, v. 38, p Zheng, M., G. R. Cass, L. Ke, F. Wang, J. J. Schauer, E. S. Edgerton, and A. G. Russell, 27, Source apportionment of PM2. and their daily variations at Jefferson Street, Atlanta GA during Summer and Winter: Journal of Air and Waste Management Association, v. 7, p Zheng, M., G. R. Cass, J. J. Schauer, and E. S. Edgerton, 22, Source apportionment of PM2. in the southeastern United States using solvent-extractable organic compounds as tracers: Environmental Science & Technology, v. 36, p