1. Austin – Round Rock Toxics Study (ARTS) Albert Hendler URS Corporation June 13, 2007

2. 2 Contents Part 1 – ARTS Overview Part 2 – Comparison with 1999 NATA Part 3 – Source Apportionment Part 4 – Acrolein Measurements

3. 3 Part 1 – ARTS Overview Exploratory study of air toxics levels 83 common VOCs, carbonyls, and metals 5 sampling sites Every 12 th day sampling

4. 4 Austin-Round Rock Quick Facts 2000 population estimate – 1.25 million  38 th largest U.S. MSA Two largest counties, Travis and Williamson, ranked 85 th and 376 th in 1999 total NEI HAPS Emissions (among 1207 U.S. urban counties) Major source HAP emissions = 1% of total

5. 5 Data Uses Health risk assessment Source apportionment Evaluation of NATA results Baseline for trend analysis Benchmark for other cities

6. 6 Part 2 – Comparison with 1999 NATA Compound Travis Co. Average (ug/m3) Ratio ARTSNATA Acetaldehyde1.421.481.04 Formaldehyde2.831.670.59 1,3-butadiene0.150.171.12 Benzene1.181.611.36 Carbon Tetrachloride0.660.270.41 Chloromethane1.341.210.91 Dichloromethane0.590.520.89 Tetrachloroethylene0.190.251.31 Toluene3.013.431.14 Total Xylene1.772.171.23 Better than a factor of 2 agreement at county level for most frequently detected VOCs and carbonyls Carbon tetrachloride and formaldehyde NATA estimates appear low Poor modeled-monitored agreement for acrolein

7. 7 ARTS data estimated assuming ND = ½ DL NATA estimates driven mostly by background Strong agreement not expected Monitored-Modeled (1999 NATA) Comparison – Travis County Averages for infrequently detected VOCs and Carbonyls

8. 8 Monitored-Modeled (1999 NATA) Comparison – Travis County Averages for frequently detected Metals Better than a factor of 2 agreement for lead, nickel and Cr6+ Worse than a factor of 10 agreement for arsenic, cadmium, and cobalt Arsenic one of top 10 risk drivers based on monitoring data – comparatively unimportant based on NATA

9. 9 Top 10 Cancer Risk Drivers for Travis County from 1999 NATA and 2005-2006 ARTS 1999 NATA 2005-2006 ARTS ND = 1/2 DLND=0 Benzene1,2-DibromoethaneCarbon tetrachloride 1,2-DibromoethaneHexachloro-1,3-butadieneBenzene 1,3-ButadieneCarbon tetrachloride1,3-Butadiene Carbon tetrachlorideBenzeneAcetaldehyde 1,1,2,2-Tetrachloroethane Arsenic AcetaldehydeAcrylonitrilep-Dichlorobenzene Bis_2_ethylhexyl_phthalate1,3-ButadieneChloroform NaphthaleneAcetaldehydeTetrachloroethylene Arsenic1,2-Dibromoethane 1,2-Dichloroethanep-DichlorobenzeneHexachloro-1,3-butadiene Bold type highlights 5 chemicals on all three lists

10. 10 Monitored-Modeled (1999 NATA) Comparison – Site Averages for Total Risk Estimate Monitored-modeled cancer risk in near-perfect agreement based on county average; agreement is within about +/- 40% at site/census tract level

11. 11 Summary of Comparison with 1999 NATA Strong agreement between ARTS 2005-2006 monitoring results and 1999 NATA in terms of:  Estimated Concentrations for most frequently detected VOCs and carbonyls  Identification of key species  Estimated risks Formaldehyde and carbon tetrachloride NATA estimates appear low Wide range of variability in ARTS-NATA agreement for metals

12. 12 Part 3 – Source Apportionment One site – WETX – had a different VOC fingerprint than the other 4 ARTS sites Higher levels of VOC that are constituents of motor vehicle emissions were observed Source apportionment objective was to see if a MV source profile could be confirmed and its contribution to the measured benzene and other measured species quantified

13. 13 U.S. 2006 Average Benzene Levels from http://www.epa.gov/air/data/index.html Webberville Rd. benzene average was greater than 80% of 335 averages for 2006 reported in AQS

14. 14 EPA Positive Matrix Factorization (PMF) Menu-driven source apportionment tool available for download at http://www.epa.gov/heasd/products/pmf/pmf.htm Uses temporal variability in measured concentrations and multi-pollutant relationships to infer source types and their contributions to the observed pollutant levels  Inputs are measured levels of multiple VOC, carbonyl, and/or PM chemical species over a long time series of sampling periods  Outputs are chemical source profiles and their estimated contributions to the observed levels  Model user or data analyst identifies source profiles based on key species

15. 15 Example of Source Profile Produced by PMF and Identified as Motor Vehicle Emissions

16. 16 Other Source Profile Produced by PMF

17. 17 Set of 6 Source Profiles Produced by PMF from ARTS VOC and Carbonyl Measurements Source Profile No. DescriptionKey SpeciesKey Months/Sites 1Acrolein/MEKAcrolein, MEKJun.-Sep./All sites 2Toluene Mar.-Nov./RRTX 3Motor Vehicle Emissions Benzene, toluene, ethyl benzene, xylene, trimethylbenzene, acetylene, propylene, butadiene All months/WETX 4BackgroundCarbon tetrachloride, chloromethane, dichlorofluoromethane, trichlorofluoromethane All months/All sites 5Secondary Aldehydes Formaldehyde, acetaldehyde,Jun.-Sep./All sites 6Undefined (possible contaminant) Ethyl benzene, MEK, styrene, xylene, MIBK Jun.-Sep. 2005/All sites

18. 18 Relative Source Strengths by Site Less than a factor of two variability in average source strengths between sites, except for an apparent toluene source impacting RRTX and motor vehicle emissions impacting WETX

19. 19 Reconstructed vs. Measured Benzene (R-Square = 0.87) Reconstructed Benzene (y-axis) is the sum of all the source contributions to a particular benzene measurement (x-axis) For the ARTS source apportionment, reconstructed benzene agreed reasonably well with the measurements

20. 20 Benzene Source Apportionment for Webberville Road (WETX)

21. 21 ARTS Sampling Sites and Major Roadways WETX was the farthest of all ARTS sites from a major freeway

22. 22 Webberville Road Sampling Site

23. 23 Summary of Source Apportionment PMF produced 6 source profiles from the ARTS data set The greatest motor vehicle impact was detected for Webberville Road (WETX), where benzene levels averaged about 2x the other ARTS sites 79% of the benzene mass measured at WETX was attributed to motor vehicle emissions WETX was the farthest of all ARTS sites from a major freeway

24. 24 Part 4 – Acrolein Measurements Analyte 2005-2006 Average Concentrations (μg/m 3 ) RfC (μg/m 3 ) Average Hazard Quotient MUTXPITXRRTXTRTXWETX Acrolein3.732.514.602.583.770.02172 Formaldehyde2.782.853.262.982.729.80.30 Acetaldehyde1.341.331.381.401.6290.16 1,3-Butadiene0.07 0.140.3320.07 Manganese0.000.01 0.050.12 1,2-Dibromoethane0.10 0.110.100.80.13 Acrylonitrile0.070.06 0.1820.04 Benzene0.940.800.981.111.88300.04 1,2-Dichloropropane0.08 0.090.08 40.02 Arsenic0.00 0.030.02 Top 10 Non-Cancer Hazard Quotients ARTS acrolein measurements exceeded the EPA RfC by more than 100 times

25. 25 U.S. 2005 Average Acrolein Levels from http://www.epa.gov/air/data/index.html ARTS acrolein averages were the highest reported to EPA AQS in 2005

26. 26 Comparison of Monitored and Modeled (1999 NATA) Acrolein Estimates ARTS acrolein averages exceeded 1999 NATA estimates by factors of about 20 to 60

27. 27 Acrolein Monitoring Method and Data Quality Indicators Measurements made using TO-15 (canister – GC/MS) Acceptable stability in canisters and recovery demonstrated by ERG Same method and lab used in 2005 and 2006 for the EPA Urban Air Toxics Monitoring Program Good measurement precision in ARTS for 13 of 16 collocated sample pairs Acrolein Canister Duplicates Average RPD = 49% (N=16) The data quality appears to be OK but…

28. 28 Acrolein Emissions Densities from the 1999 NEI No known emission sources explain why the ARTS acrolein levels were the highest in the U.S.

29. 29 U.S. 2006 Average Acrolein Levels from http://www.epa.gov/air/data/index.html In 2006, seven Indiana sites joined four ARTS sites to round out the top 10% of reported acrolein

30. 30 2006 Acrolein Averages from AQS

31. 31 1999 NATA County-Averaged Acrolein Estimates NATA modeling of known emissions does not explain the Indiana measurements

32. 32 Acrolein/MEK Source Profile

33. 33 Acrolein Time Series

34. 34 Acrolein Follow Up Study – Summer 2007 Collaborative effort between EPA, TCEQ, and CAPCOG Conducted by URS, ERG, UTSPH, and Czartech Analytical 3 measurement techniques  TO-15  Modified TO-11A (DNPH) – Includes analysis of acrolein decay products  Dansylhydrazine (DNSH) passive sampler (Herrington et al, 2006) Field Duplicates Replicate analyses Through the probe field blanks and challenges Blind spiked QC samples Data quality assessments for accuracy and precision

35. 35 Acrolein Summary ARTS acrolein levels were among the highest reported to EPA AQS Very poor agreement with NATA estimates No known sources account for the measured levels or differences between Austin-Round Rock compared and most other urban sites  Similarly high levels measured in Indiana and Tulsa show the anomaly is not limited to central Texas Repeatability of measurements and reliability of measurement method to be tested in summer 2007

36. 36 Acknowledgements ARTS was sponsored by a grant from the U.S. EPA to the Capitol Area Council of Governments (CAPCOG)  Kuenja Chung was the EPA Project Officer  Mike Fishburn was the CAPCOG Project Manager  Samples were analyzed by Eastern Research Group (Julie Swift) The acrolein follow up study was designed by David Brymer, David Carmichael, and David Manis, TCEQ