Receptor Model Comparisons and Wind Direction Analyses of Volatile

Mar 12, 2004 - The relationship between continuous measurements of volatile organic compounds sources and particle number was evaluated at a Photochem...
29 downloads 8 Views 153KB Size
Environ. Sci. Technol. 2004, 38, 2317-2327

Receptor Model Comparisons and Wind Direction Analyses of Volatile Organic Compounds and Submicrometer Particles in an Arid, Binational, Urban Air Shed S H A I B A L M U K E R J E E , * ,† GARY A. NORRIS,† LUTHER A. SMITH,‡ CHRISTOPHER A. NOBLE,§ LUCAS M. NEAS,| A. HALU ˆ K O ¨ ZKAYNAK,† AND M E L I S S A G O N Z A L E S |,⊥ National Exposure Research Laboratory, U.S. Environmental Protection Agency, MD E205-02, Research Triangle Park, North Carolina 27711, ManTech Environmental Technology, Inc., Durham, North Carolina 27713, Center for Aerosol Technology, RTI International, Research Triangle Park, North Carolina 27709, and National Health and Environmental Effects Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711

The relationship between continuous measurements of volatile organic compounds sources and particle number was evaluated at a Photochemical Assessment Monitoring Station Network (PAMS) site located near the U.S.Mexico Border in central El Paso, TX. Sources of volatile organic compounds (VOCs) were investigated using the multivariate receptor model UNMIX and the effective variance least squares receptor model known as Chemical Mass Balance (CMB, Version 8.0). As expected from PAMS measurements, overall findings from data screening as well as both receptor models confirmed that mobile sources were the major source of VOCs. Comparison of hourly source contribution estimates (SCEs) from the two receptor models revealed significant differences in motor vehicle exhaust and evaporative gasoline contributions. However, the motor vehicle exhaust contributions were highly correlated with each other. Motor vehicle exhaust was also correlated with the ultrafine and accumulation mode particle count, which suggests that motor vehicle exhaust is a source of these particles at the measurement site. Wind sector analyses were performed using the SCE and pollutant data to assess source location of VOCs, particle count, and criteria pollutants. Results from this study have application to source apportionment studies and mobile

* Corresponding author e-mail: [email protected]; phone: (919)541-1865; fax: (919)541-4787. † National Exposure Research Laboratory, U.S. EPA. ‡ ManTech Environmental Technology, Inc. § RTI International. | National Health and Environmental Effects Research Laboratory, U.S. EPA. ⊥ Present address: Department of Internal Medicine, University of New Mexico School of Medicine, Albuquerque, NM 87131. 10.1021/es0304547 CCC: $27.50 Published on Web 03/12/2004

 2004 American Chemical Society

source emission control strategies that are ongoing in this air shed.

Introduction Recent improvements to receptor models have enhanced their user-friendliness and functionality and, as a result, increased their application in atmospheric characterization and modeling studies. Multivariate receptor models such as Positive Matrix Factorization (1) and UNMIX (2-4) have been applied to particulate data to assess source particle contributions from long-range transport (5). These receptor models can also be applied to monitoring network databases such as continuous volatile organic compound (VOC) measurements conducted under the national Photochemical Assessment Monitoring Stations (PAMS) Network and other speciated fine-particle measurements (2, 4) such as those at the U.S. Environmental Protection Agency’s (EPA) newly implemented Speciation Trends Network. In this study a PAMS site in central El Paso, TX, and in close proximity to the U.S.-Mexico border and a major international bridge crossing was used to monitor air pollutants including PAMS VOCs and airborne particles in early winter to assess source influences of these pollutants and their impact in El Paso. A previous source apportionment study using data collected in the summer of 1996 at this PAMS site included application of the source profile-based chemical mass balance (CMB) receptor model (6) to assess the impact of VOC sources from El Paso and its Mexican sister city of Ciudad (Cd.) Jua´rez, Mexico (7). This receptor modeling effort was part of a larger investigation directed by the EPA under the U.S.-Mexico Border XXI Program and Annex V of the La Paz Agreement (8) known as the Paso del Norte Ozone (O3) Study or PDNOS (9). The PDNOS measured and modeled atmospheric transformations of ozone-forming hydrocarbons and meteorological conditions conducive to O3 exceedances (9-13). Levels of O3 along with carbon monoxide (CO) and PM less than 10 µm aerodynamic diameter (PM10) in the Paso del Norte air shed have exceeded U.S. and Mexican air standards (8). In the present study, two receptor models were used to determine and quantify the sources of VOCs: CMB (6), a least squares model that uses collected source profiles and ambient data to calculate SCEs, and UNMIX (2-4), a multivariate model that generates source profiles and calculates SCEs based on the ambient data. Source profiles previously collected in the Paso del Norte air shed (12) (19961997) were used with PAMS measurements acquired during this investigation for the CMB analysis. The profiles used by CMB were compared with profiles generated by the EPA UNMIX 2.3 (2-4) receptor model; this model has been recently developed by the EPA as a stand-alone version. Comparison of CMB and UNMIX estimates provided an opportunity to not only assess Paso del Norte source profiles but also to assess potential changes in mobile source and other emissions since completion of the PDNOS. This would be of particular interest for this air shed and other U.S.Mexico border air sheds since significant increases in border crossing traffic (car and truck) have occurred for many border cities during the 1990s, including Paso del Norte, when commerce related to the North American Free Trade Agreement commenced (14). Mobile sources have been determined to be primary contributors to submicron particles in urban air sheds, and correlations have been found between particle number VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2317

concentration and traffic patterns (15, 16). Correlations have also been found between ultrafine (20 ppbC) from weekdays to weekends. In contrast, gasoline evaporation showed slight fluctuations ( 0.8) were generally found between mobile source-related VOC sources such as

motor vehicle exhaust and gasoline vapor evaporation with CO and NOx. Strong correlation was also noted for TNMOC with these sources and species. A notable exception was the low correlation of the gasoline evaporation source from UNMIX with CO, NOx, and TNMOC. Of the three particle modes, ultrafine PM (UF) showed the highest correlation with motor vehicle exhaust suggesting that this source was an ultrafine PM contributor. Thus, VOC source apportionment results and correlations with submicron pollutants and TNMOC indicated mobile sources were the dominant contributor to VOCs and ultrafine PM. The correlation of UF with mobile sources was also revealed in the diurnal plots for this particle mode shown as fractions to total particle counts in Figure 4; as with the SCE diurnal analysis, this normalization was performed on the particle counts to remove meteorologically driven correlation in the data. Diurnal variations for UF were similar to that revealed in the motor vehicle exhaust plots (Figure 3a) with higher levels in the early morning and late afternoon suggesting that these sources were the principal factors. Similar diurnal patterns have been found in other air sheds (24) and confirmed at another nearby El Paso site for PM2.5 measurements that were conducted in nearly the same time frame as this investigation (44). A reason for the strong correlation between UF particle counts and mobile source may have been the fact that VOCs are known to form particles during atmospheric transformation (45), in addition to forming ozone. Ultrafine PM was also well-correlated with accumulation PM VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2325

due to meteorological influences and the fact that both particle size distributions are formed from combustion sources (22), such as motor vehicle exhaust. Diurnal patterns for accumulation mode (AM) particle fractions are also shown in Figure 4. Coarse mode (CM) diurnal patterns are shown in Figure S-6 in the Supporting Information. Unlike the UF particle fraction, AM and CM fractions do not follow the diurnal patterns for motor vehicle exhaust. The fact that even the AM and CM particles showed good correlation (r > 0.69) with vehicle exhaust sources may be a reflection of meteorological influences since the diurnal patterns of these particle fractions did not suggest rushhour traffic patterns (see Figures 4 and S-6). Finally, the two LPG sources estimated from CMB and UNMIX showed strong correlation (r ) 0.98) since the profiles from the two models were based almost entirely on propane. Wind Sector Analysis. Table 6 shows overall medians of wind speed, SCEs, and pollutants as well as medians based on the direction from which the wind was coming (northeast, south, and northwest sectors). These three wind sectors of northeast (10-120° from the north), south (121-240°), and northwest (271-340°) were used in the subsequent directional analyses. Wind sectors were developed from examinations of local topography (Franklin and Sierra Jua´rez Mountains and Rio Grande), how they channeled surface winds (13, 46), and retaining enough observations in each sector for pairwise comparisons. For example, wind directions from 341° to 10° were omitted since these winds may have been influenced by complex terrain effects from the Franklin Mountains. The south sector was created to follow the general path of the Rio Grande and primarily represents winds from Mexico. Figure S-1 in the Supporting Information delineates the wind sectors on the map. To test whether the medians for each SCE/pollutant in Table 6 were equivalent between wind sectors, the Kruskal-Wallis test was initially used to examine whether differences seemed to be present on an overall basis, and the Wilcoxon rank sum test was used to make the subsequent pairwise comparisons (47). A nonparametric approach was used since examination of the distributions of the data by wind direction indicated asymmetric distributions. Tests were conducted using the NPAR1WAY procedure (48) in SAS on the 3-h averaged data; the results are shown in Table 7. The p values in bold indicate statistical significance at the 5% level. For the sake of completeness, all pairwise wind sector comparisons are presented in Table 7. Note, however, that the Kruskal-Wallis test indicated no significant difference (at the 5% level) among the wind sectors for CO, NOx, PM10, and the CM mode. Accordingly, the apparently significant pairwise results for S/NW for CO (p ) 0.03) and PM10 (p ) 0.04) may well be statistical artifacts and should be viewed cautiously. Examination of Table 7 indicates that the northwest sector was significantly lower in most pollutant/SCE levels than either the northeast or south. The AM particle mode, gasoline evaporation, and CNG were all higher (p < 0.05) in the northeast than the northwest. UF and AM particle modes, CO, motor vehicle exhaust, gas evaporation, and LPG were all significantly higher (p < 0.05) in the southern sector than the northwest. Additionally, the UF particle mode and SCE for LPG were higher in the southern section than in the northeast. This may have been due to mobile source and natural gas influences in Cd. Jua´rez or international bridge traffic emissions which were south of CAMS 41. Funk and others (11) revealed that major VOC and CO emissions from mobile and other sources in the Paso del Norte region were from downtown Cd. Jua´rez; this was also determined for PM spatial influences in this air shed (44). The northwest sector was found to have significantly higher wind speeds than either the northeast or south (see 2326

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 8, 2004

Tables 6 and 7); this may partially explain the lower contributions from this sector. Elevation was highest in the northwest sector of El Paso and decreased further east and south (9) and this may have also affected wind speeds from that direction. Although not shown, NMOC SCEs and submicron particle counts showed statistically significant negative correlations with wind speed (r e -0.4) confirming the effect of dilution of pollutants by wind speed. As an adjustment for dilution, pairwise comparisons of wind sectors were reexamined in terms of fluxes by calculating the product of pollutant/SCEs and wind speed (49, 50). Comparison of pollutant concentrations and SCEs as fluxes by wind direction revealed an overall mixing of source emissions at CAMS 41 and/or very local source impacts. VOC/ultrafine PM impacts from the south were not higher than other sectors, negating any implied transboundary effect.

Acknowledgments The first three authors contributed equally to this paper. We acknowledge the technical support of Charles E. Rodes, Philip A. Lawless, Sanjay Natarajan, and Eric A. Myers from the Research Triangle Institute. We also thank Victor Valenzuela and Archie Clause of TCEQ and Allyson Siwik formerly with the U.S.-Mexico Border Program Office of U.S. EPA Region VI in El Paso for assistance in experimental planning and study implementation and Eric Fujita of the Desert Research Institute for providing CMB source profiles. Finally, we thank Shelly Eberly and Charles Lewis of the U.S. EPA for providing review comments on this paper. The U.S. Environmental Protection Agency through its Office of Research and Development funded and managed the research described here under Contract 68-D-99-012 to the Research Triangle Institute and Contract 68-D0-0206 to ManTech Environmental Technology, Inc. The paper has been subjected to Agency review and approved for publication. Mention of trade names or commercial products does not constitute an endorsement or recommendation for use.

Supporting Information Available Discussion of data screening VOCs and factor analysis, averaging to account for autocorrelation, and figures presenting differences in model source contribution estimates by hour and diurnal patterns of particle modes. This material is available free of charge via the Internet at http:// pubs.acs.org.

Literature Cited (1) Paatero, P.; Tapper, U. Environmetrics 1994, 5, 111-126. (2) Henry, R. C. UNMIX Version 2.3 Manual; 2001; available with UNMIX software ([email protected]). (3) Henry, R. C.; Lewis, C. W.; Collins, J. F. Environ. Sci. Technol. 1994, 28, 823-832. (4) Lewis, C. W.; Norris, G. A.; Conner, T.; Henry, R. C. J. Air Waste Manage. Assoc. 2003, 53, 325-338. (5) Poirot, R. L.; Wishinski, P. R.; Hopke, P. K.; Polissar, A. V. Environ. Sci. Technol. 2001, 35, 4622-4636. (6) U.S. Environmental Protection Agency. CMB8 User’s Manual; EPA-454/R-01-XX (Draft); http://www.epa.gov/scram001/ tt23.htm, 2001. (7) Fujita, E. M. Sci. Total Environ. 2001, 276, 171-184. (8) U.S. Environmental Protection Agency. U.S.-Mexico Border XXI Program: Progress Report 1996-2000; EPA 160/R/00/001; U.S. Government Printing Office: Washington, DC, 2001. (9) Roberts, P.; MacDonald, C.; Main, H.; Dye, T.; Coe, D.; Haste, T. Analysis of Meteorological and Air Quality Data for the 1996 Paso del Norte Ozone Study; STI-997330-1754-FR; Final report prepared for the U.S. Environmental Protection Agency, Region 6, by Sonoma Technology, Inc., Santa Rosa, CA, 1997. (10) MacDonald, C. P.; Roberts, P. T.; Main, H. H.; Dye, T. S.; Coe, D. L.; Yarbrough, J. Sci. Total Environ. 2001, 276, 93-109. (11) Funk, T. H.; Chinkin, L. R.; Roberts, P. T.; Saeger, M.; Mulligan, S.; Pa´ramo Figueroa, V. H.; Yarbrough, J. Sci. Total Environ. 2001, 276, 135-151.

(12) Seila, R. L.; Main, H. H.; Arriaga, J. L.; Martı´nez V., G.; Ramadan, A. B. Sci. Total Environ. 2001, 276, 153-169. (13) Brown, M. J.; Muller, C.; Wang, G.; Costigan, K. Sci. Total Environ. 2001, 276, 111-133. (14) Mukerjee, S. Sci. Total Environ. 2001, 276, 1-18. (15) Harrison, R. M.; Shi, J. P.; Jones, M. R. Atmos. Environ. 1999, 33, 1037-1047. (16) Morawska, L.; Thomas, S.; Bofinger, N.; Wainwright, D.; Neale, D. Atmos. Environ. 1998, 32, 2467-2478. (17) Hitchins, J.; Morawska, L.; Wolff, R.; Gilbert, D. Atmos. Environ. 2000, 34, 51-59. (18) Morawska, L.; Thomas, S.; Gilbert, D.; Greenway, C.; Rijinders, E. Atmos. Environ. 1999, 33, 1261-1274. (19) U.S. Environmental Protection Agency. PAMSGRAM Vol. 18. Supplemental Information on the Operation of the Ozone Precursor System; U.S. EPA: Research Triangle Park, NC, June 1, 2000; available at www.epa.gov/ttn/amtic/pamsgram.html (accessed 2000). (20) Noble, C. A.; Mukerjee, S.; Gonzales, M.; Rodes, C. E.; Lawless, P. A.; Natarajan, S.; Myers, E. A.; Norris, G. A.; Smith, L.; O ¨ zkaynak, H.; Neas, L. M. Atmos. Environ. 2003, 37, 827-840. (21) Morawska, L.; Thomas, S.; Jamriska, M.; Johnson, G. Atmos. Environ. 1999, 33, 4401-4411. (22) Finlayson-Pitts, B. J.; Pitts, J. N., Jr. Atmospheric Chemistry: Fundamentals and Experimental Techniques; Wiley: New York, 1986; pp 730-742. (23) Hinds, W. C. Aerosol Technology: Properties, Behavior, and Measurement of Airborne Particles, 2nd ed.; Wiley: New York, 1999; pp 8-11. (24) Chow, J. C. J. Air Waste Manage. Assoc. 1995, 45, 320-382. (25) Lewis, C. W.; Henry, R. C.; Shreffler, J. H. J. Air Waste Manage. Assoc. 1998, 48, 71-76. (26) Henry, R. C.; Lewis, C. W.; Hopke, P. K.; Williamson, H. J. Atmos. Environ. 1984, 18, 1507-1515. (27) Watson, J. G.; Robinson, N. F.; Chow, J. C.; Henry, R. C.; Kim, B. M.; Pace, T. G.; Meyer, E. L.; Nguyen, Q. Environ. Software 1990, 5, 38-49. (28) Watson, J. G.; Chow, J. C.; Fujita, E. M. Atmos. Environ. 2001, 35, 1567-1584. (29) Apel, E. C.; Calvert, J. G.; Riemer, D.; Pos, W.; Zika, R.; Kleindienst, T. E.; Lonneman, W. A.; Fung, K.; Fujita, E.; Shepson, P. B.; Starn, T. K.; Roberts, P. T. J. Geophys. Res. 1998, 103-D17, 2229522316. (30) Conner, T. L.; Lonneman, W. A.; Seila, R. L. J. Air Waste Manage. Assoc. 1995, 45, 383-394. (31) Doskey, P. V.; Porter, J. A.; Scheff, P. A. J. Air Waste Manage. Assoc. 1992, 42, 1437-1445. (32) Graedel, T. E. Chemical Compounds in the Atmosphere; Academic Press: New York, 1978.

(33) Kenski, D. M.; Wadden, R. A.; Scheff, P. A.; Lonneman, W. A. Receptor Modeling of VOCs in Chicago, Beaumont, and Detroit; No. 91-82.3; 84th Annual A&WMA Meeting, Vancouver, BC, 1991. (34) Kenski, D. M.; Wadden, R. A.; Scheff, P. A.; Lonneman, W. A. Receptor Modeling of VOCs in Atlanta, Georgia; No. 92-104.06; 85th Annual A&WMA Meeting, Kansas City, MO, 1992. (35) Lewis, C. W.; Conner, T. L.; Stevens, R. K.; Collins, J. F.; Henry, R. C. Receptor Modeling of Volatile Hydrocarbons Measured in the 1990 Atlanta Ozone Precursor Study; No. 93-TP-58.04; 86th Annual A&WMA Meeting, Denver, CO, 1993. (36) Scheff, P. A.; Wadden, R. A. Environ. Sci. Technol. 1993, 27, 617-625. (37) Wadden, R. A.; Scheff, P. A.; Franke, J. E.; Conroy, L. M.; Kiel, C. B. J. Air Waste Manage. Assoc. 1995, 45, 547-555. (38) Henry, R. C.; Spiegelman, C. H.; Collins, J. F.; Park, E. S. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 6596-6599. (39) SAS/ETS User’s Guide, Version 8; SAS Institute: Cary, NC, 1999; pp 191-396. (40) Henry, R. C. The Application of Factor Analysis to Urban Aerosol Source Indentification; Proceeding of the Fifth Conference on Probability and Statistics, American Meteorological Society, Boston, 1977; pp 134-138. (41) Einfeld, W.; Church, H. W. Winter Season Air Pollution in El Paso-Ciudad Juarez; EPA-906-R-95-001; U.S. EPA, Region 6: Dallas, TX, 1995. (42) Fujita, E. M.; Campbell, D. E.; Zielinska, B.; Sagebiel, J. C.; Bowen, J. L.; Goliff, W. S.; Stockwell, W. R.; Lawson, D. R. J. Air Waste Mange. Assoc. 2003, 53, 844-863. (43) SAS Procedures Guide, Version 8; SAS Institute: Cary, NC, 1999; pp 1317-1454. (44) Li, W. W.; Orquiz, R.; Garcia, J. H.; Espino, T. T.; Pingitore, N. E.; Gardea-Torresday, J.; Chow, J. C.; Watson, J. G. J. Air Waste Manage. Assoc. 2001, 51, 1551-1560. (45) Grosjean, D.; Seinfeld, J. H. Atmos. Environ. 1989, 23, 17331747. (46) Pearson, R.; Fitzgerald, R. J. Air Waste Manage. Assoc. 2001, 51, 669-680. (47) Hollander, M.; Wolfe, D. A. Nonparametric Statistical Methods; Wiley: New York, 1973; 503 pp. (48) SAS/STAT User’s Guide, Version 8; SAS Institute: Cary, NC, 1999; pp 2507-2552. (49) Harrison, H. J. Air Waste Manage. Assoc. 1991, 41, 1195-1198. (50) Willis, R. D.; Ellenson, W. D.; Conner, T. L. J. Air Waste Mange. Assoc. 2001, 51, 1142-1166.

Received for review May 2, 2003. Revised manuscript received January 13, 2004. Accepted January 20, 2004. ES0304547

VOL. 38, NO. 8, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2327