The concentrations and sources of PM2.5 in metropolitan New York City

ARTICLE IN PRESS

Atmospheric Environment 40 (2006) S312–S332 www.elsevier.com/locate/atmosenv

The concentrations and sources of PM2.5 in metropolitan New York City Youjun Qin, Eugene Kim, Philip K. Hopke Center for Air Resources Engineering and Science, Clarkson University, Box 5708, Potsdam, NY 13699-5708, USA Received 30 April 2005; received in revised form 12 February 2006; accepted 21 February 2006

Abstract The concentration time series of chemical species measured in PM2.5 samples from four speciation trend network (STN) sites in the New York City metropolitan area and a upwind background site were explored. PM2.5 concentrations and chemical compositions measured in metropolitan area of New York City are uniform. About 69–82% of PM2.5 mass 4
derives from transport. The most important constituents of the PM2.5 were SO2
4 , NH and NO3 and account 54–67% of
PM2.5 mass. More than 93% of SO2
and about 54–65% of NO are likely to have been transported into the NYC area 4 3 based on the concentrations observed at the background site. Backward air parcel trajectories indicate that coal-fired power plants in the border area among West Virginia, Ohio and Pennsylvania are related to typical high PM2.5 events having peak secondary pollutant concentrations in New York City. Positive matrix factorization (PMF) was applied to identify the PM2.5 sources and estimate the source contributions. Sources common to all five sites included secondary sulfate, secondary nitrate, soil and aged sea salt. Oil combustion was identified at four of the sites. At the Elizabeth site, the oil combustion source appears to show an influence from ship emissions. Motor vehicles were apportioned into two sources (gasoline and diesel) at three site and three sources at the Elizabeth site, probably because of its proximity to a major interstate highway. At the Queens College site, only a combined motor vehicle factor could be resolved. The source profiles, source contributions and seasonal or weekday variations derived by PMF are compared to source inventories for the area. It appears that there were more vehicle exhausts and less dust and wood smoke than are indicated by the source inventories. r 2006 Elsevier Ltd. All rights reserved. Keywords: PM2.5; Composition; Positive matrix factorization; PMF; Source apportionment; Emission inventory

1. Introduction In July 1997, the US EPA promulgated new National Ambient Air Quality Standards (NAAQS) for particulate matter with aerodynamic diameter less than 2.5 mm (PM2.5). Since 1988, a national Corresponding author. Tel.: +1 315 268 3861; fax: +1 315 268 4410. E-mail address: [email protected] (P.K. Hopke).

1352-2310/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2006.02.025

chemical speciation monitoring network—IMPROVE (The Interagency Monitoring of Protected Visual Environments) has been set up to monitor visibility and aerosol conditions in national parks or wilderness areas. In 2000, US EPA established the speciation trends network (STN) to determine the composition of PM2.5 in selected urban areas (RTI, 2003). One of the major objectives of this network is to provide chemically speciated data to develop PM2.5 control strategies based on emission

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

inventory development, air quality model evaluation, and source apportionment. To provide a long term perspective on the NYC area PM2.5 composition, data obtained at the four STN sites in the NYC metropolitan area were used to examine the PM2.5 chemical compositions, their spatial and temporal variations, and sources. The distances among the four urban sites (Bronx (2 locations), Brooklyn, and Elizabeth, NJ) are less than 50 km from one another. Another STN site is located in Chester, New Jersey about 100 km in upwind direction from NYC. The Chester site serves as a background site that is principally influenced by particulate matter transported into the region. The data measured at these five sites permit the evaluation of local primary emissions and secondary pollutants in this large metropolitan area. The characteristics of the PM2.5 species are presented in terms of summary statistics, time series

S313

plots, and related discussion. In addition, speciation data measured at four STN sites located in the New York metropolitan area were used to explore the spatial and temporal variations of the PM2.5 chemical compositions. The data from each of these five sites were analyzed using positive matrix factorization (PMF) to obtain the PM2.5 source profiles and contributions. 2. Sampling and chemical analysis The locations of five STN sites in the metropolitan area of New York City are shown in Fig. 1. The New York Botanical Garden site (NYBG) (401510 5800 N, 731520 5000 W) and the Intermediate School 52 site (IS52) (401480 5700 N, 731540 0700 W) are located in Bronx County, New York. The Queens College site (QCII) (401440 1100 N, 731490 2300 W) is located in Queens County, New York. The Elizabeth

Fig. 1. Locations of STN sites in metropolitan New York City and northern New Jersey.

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

S314

Table 1 Summary statistics for data from the NYBG, IS52, QCII, ELIZ and CHES STN sites (mg m
3) NYBG

PM2.5 NO
3 SO2
4 Fe Ni S EC NHþ 4 Si Zn Ca K OC Naþ V Ti Cu Br Pb Na Ba Mn Ta K+ Al Cl Sn Cr Co P As Se Sr Mg Ce La Ag Ga Cs Sc W Au Sb Eu Zr In Hg Ir Cd Rb Mo Nb Hf Sm Tb Y

IS52

QCII

ELIZ

CHES

AVG

STD DEV

BDL (%)

AVG

STD DEV

BDL (%)

AVG

STD DEV

BDL (%)

AVG

STD DEV

13.64 2.00 3.91 0.104 0.024 1.292 1.28 1.756 0.088 0.031 0.05 0.041 2.93 0.229 0.006 0.006 0.004 0.003 0.006 0.088 0.032 0.002 0.011 0.022 0.015 0.024 0.012 0.001 0.001 0.004 0.001 0.001 0.001 0.007 0.014 0.010 0.003 0.001 0.005 0.000 0.003 0.001 0.006 0.001 0.001 0.003 0.001 0.002 0.002 0.00 0.001 0.001 0.003 0.001 0.001 0.001

8.84 2.01 3.16 0.057 0.022 1.029 0.65 1.511 0.083 0.022 0.027 0.037 2.07 0.279 0.004 0.004 0.004 0.003 0.005 0.107 0.027 0.002 0.012 0.041 0.033 0.093 0.010 0.002 0.001 0.008 0.001 0.001 0.002 0.014 0.023 0.016 0.004 0.001 0.009 0.001 0.004 0.002 0.009 0.003 0.004 0.005 0.002 0.002 0.003 0.001 0.002 0.001 0.007 0.001 0.002 0.001

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.2 0.5 0.5 0.7 1.7 2.0 5.9 17.8 25.6 28.0 35.1 52.7 55.4 58.8 59.0 64.6 67.6 68.8 68.8 69.5 75.6 78.5 79.5 80.7 83.9 85.4 88.8 91.5 91.7 92.2 93.7 93.9 93.9 94.9 95.1 95.4 95.4 95.4 95.6 95.6 95.9 96.3 96.3 96.8 96.8 97.1 97.6 97.6 97.8

13.89 2.38 4.18 0.114 0.019 1.394 1.06 2.054 0.095 0.032 0.050 0.048 2.40 0.186 0.006 0.005 0.004 0.004 0.006 0.063 0.016 0.002 0.003 0.019 0.012 0.068 0.004 0.001 0.001 0.003 0.001 0.001 0.001 0.006 0.006 0.004 0.002 0.00 0.003 0.00 0.001 0.001 0.004 0.002 0.001 0.001 0.001 0.001 0.001 0.00 0.001 0.000 0.001 0.000 0.001 0.00

8.98 2.12 3.29 0.074 0.015 1.014 0.70 1.658 0.074 0.021 0.024 0.043 1.69 0.267 0.005 0.004 0.003 0.003 0.006 0.08 0.012 0.002 0.004 0.039 0.03 0.192 0.004 0.001 0.001 0.006 0.001 0.002 0.001 0.016 0.01 0.007 0.002 0.00 0.004 0.00 0.002 0.001 0.005 0.004 0.006 0.002 0.001 0.001 0.002 0.00 0.001 0.001 0.002 0.001 0.002 0.00

0.3 0.0 0.0 0.0 0.3 0.0 6.9 0.0 0.0 0.3 0.3 1.3 1.6 4.1 4.4 8.4 8.8 3.4 12.5 45.9 52.5 26.6 77.8 63.8 57.8 47.5 76.6 61.6 60.3 76.6 54.1 54.1 65.0 76.6 91.6 92.8 90.6 99.1 92.2 95.9 95.9 94.7 90.0 82.2 87.8 95.9 93.1 97.5 95.3 97.2 94.4 97.8 98.1 93.8 92.2 98.1

13.16 2.04 4.29 0.1 0.014 1.428 0.74 1.925 0.089 0.026 0.044 0.048 2.32 0.223 0.006 0.005 0.004 0.004 0.005 0.069 0.015 0.002 0.003 0.021 0.012 0.056 0.004 0.001 0.001 0.002 0.001 0.001 0.001 0.006 0.005 0.004 0.001 0.00 0.002 0.000 0.001 0.001 0.003 0.001 0.001 0.001 0.001 0.001 0.001 0.00 0.001 0.000 0.001 0.000 0.001 0.00

9.05 1.81 3.62 0.06 0.012 1.08 0.48 1.603 0.095 0.019 0.027 0.049 1.72 0.626 0.004 0.004 0.003 0.002 0.004 0.092 0.012 0.002 0.005 0.057 0.029 0.197 0.004 0.001 0.001 0.005 0.001 0.001 0.001 0.012 0.009 0.007 0.001 0.00 0.003 0.001 0.002 0.001 0.005 0.003 0.002 0.002 0.001 0.001 0.002 0.00 0.001 0.001 0.002 0.001 0.002 0.00

0.0 0.0 0.0 0.0 0.4 0.0 7.0 0.7 0.0 0.0 0.0 0.0 2.5 3.5 1.4 7.4 7.4 2.8 22.9 46.5 57.0 26.4 79.2 65.1 52.1 49.6 78.9 64.8 65.8 76.4 56.7 51.8 61.3 76.4 91.5 92.3 94.0 99.6 96.1 96.5 95.8 93.0 90.5 85.6 89.8 96.5 93.0 97.9 95.4 96.1 96.8 97.9 98.2 93.3 91.9 98.2

15.48 10.68 2.2 2.07 4.67 3.93 0.13 0.07 0.005 0.005 1.493 1.213 1.82 0.99 2.143 1.743 0.096 0.092 0.024 0.105 0.043 0.026 0.05 0.07 3.51 3.50 0.249 0.298 0.007 0.008 0.007 0.005 0.006 0.005 0.004 0.003 0.005 0.004 0.076 0.116 0.031 0.028 0.003 0.003 0.008 0.012 0.028 0.072 0.021 0.054 0.043 0.123 0.01 0.01 0.003 0.005 0.00 0.001 0.003 0.009 0.001 0.001 0.002 0.001 0.001 0.002 0.008 0.022 0.012 0.023 0.009 0.015 0.003 0.003 0.000 0.001 0.005 0.007 0.000 0.001 0.003 0.004 0.002 0.002 0.006 0.010 0.002 0.005 0.002 0.007 0.003 0.005 0.001 0.001 0.002 0.002 0.003 0.005 0.000 0.001 0.001 0.002 0.001 0.001 0.003 0.008 0.001 0.002 0.001 0.002 0.001 0.001

BDL (%)

AVG

STD DEV

BDL (%)

0.0 0.0 0.0 0.0 19.3 0.0 0.0 0.0 0.9 4.3 1.3 0.4 3.6 8.3 31.8 23.6 14.2 27.0 55.4 64.4 61.8 43.3 76.8 69.1 67.0 60.5 76.4 54.5 96.1 85.8 81.1 76.8 79.4 89.3 92.3 93.6 95.7 99.6 96.6 94.4 95.3 92.3 93.1 88.0 88.8 94.8 95.7 97.4 92.3 96.1 98.7 99.1 95.7 94.4 98.7 98.3

10.73 1.29 4.32 0.037 0.002 1.361 0.393 1.51 0.053 0.007 0.02 0.037 2.33 0.199 0.002 0.003 0.005 0.003 0.003 0.058 0.019 0.002 0.007 0.027 0.006 0.006 0.008 0.002 0.00 0.002 0.001 0.001 0.001 0.003 0.013 0.006 0.003 0.000 0.005 0.000 0.003 0.001 0.005 0.001 0.001 0.002 0.001 0.001 0.003 0.000 0.001 0.001 0.002 0.001 0.000 0.000

8.73 1.33 4.12 0.024 0.002 1.246 0.199 1.36 0.042 0.007 0.015 0.039 3.32 0.247 0.002 0.003 0.007 0.002 0.003 0.087 0.023 0.002 0.011 0.069 0.012 0.021 0.009 0.003 0.00 0.003 0.001 0.001 0.001 0.008 0.025 0.012 0.004 0.001 0.008 0.001 0.004 0.002 0.010 0.002 0.002 0.004 0.002 0.002 0.004 0.001 0.002 0.001 0.006 0.002 0.001 0.001

1.2 0.0 0.0 0.6 49.1 0.0 20.0 0.0 7.5 24.3 8.7 6.9 8.2 13.6 68.2 58.4 51.4 39.9 73.4 69.9 75.7 63.6 78.6 71.5 87.3 82.7 80.9 68.8 99.4 91.3 84.4 86.1 93.1 94.8 89.0 96.0 94.2 100.0 94.8 93.1 94.2 94.2 93.6 95.4 96.0 98.3 93.6 97.1 94.8 97.1 97.7 98.8 97.1 97.7 99.4 99.4

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

site (ELIZ) (401380 2800 N, 741120 2800 ) is located in Union County, New Jersey. All four sites were placed in urban commercial areas. The Chester site (CHES) (401470 1400 N, 741400 3100 W) is located in a suburban area in Morris County, New Jersey, about 100 km west of New York City. The distance between IS52 and the other four sites is about 6, 11, 45 and 100 km for NYBG, QCII, ELIZ and CHES, respectively. PM2.5 samples have been taken at NYBG since 15 February 2000. The sample frequency was once every six days before 3 April 2000 and once every third day after that. The first dates that sepciation data were reported for IS52, QCII, ELIZ and CHES, were 22 January 2001, 4 April 2001, 13 May 2001 and 6 June 2001, respectively. The sample frequency is once every third day at these four sites. To the end of 2003, the numbers of available samples with speciation concentrations at NYBG, IS52, QCII, ELIZ and CHES, are 422, 328, 293, 248 and 236, respectively. STN PM2.5 samples were collected on Teflon, Nylon, and quartz filters. Detailed STN sample analyses are shown in Kim et al. (2005). OC concentrations reported in STN data were not blank corrected. A reasonable approach to estimate OC blank values was to use the intercept in PM2.5 mass regression against OC concentrations (Tolocka et al., 2001; Kim et al., 2005). The data marked with error flags and extreme values were removed from the data set before further exploration. The extremely high PM2.5 and OC concentrations measured at all five sites on 7 July 2002 were excluded because they were affected by a large Canadian forest fire. The blank values of OC derived by the regression analysis are 1.54, 1.88, 1.54, 2.47 and 1.57 mg m
3 at NYBG, IS52, QCII, ELIZ and CHES, respectively. These values were subtracted from the OC concentrations before further analysis. Data quality for the STN data is denoted by flags that were used to describe possible errors. In period of September 2001–January 2002, the nylon filters used for STN were contaminated and the Naþ concentrations measured in this period were relatively high. They are flagged as 4, ‘‘possible laboratory contamination’’ and these data required special treatment in the subsequent analyses. The arithmetic means (AVG, mg m
3), standard deviations (STDEV, mg m
3), and the percentage of data values below the MDL (BDL) for the samples collected at the five STN sites are summarized in Table 1. The elements are sorted according to

S315

ascending order of BDL values at NYBG. Correlations between the XRF S values and the ion were very high (r2 ¼ 0:94, chromatography SO2
4 0.98, 0.94, 0.98 and 0.99 at NYBG, IS52, QCII, ELIZ and CHES, respectively). SO2
4 was used in the chemical composition and time series analysis. Both ion and element concentrations for Na and K are reported in STN data. The percentage of concentrations higher than the MDL for ion Na is much higher than that for element Na. Most of ion K concentrations are below the MDL. Thus, ion Na and element K were selected for further analysis. 3. Chemical compositions and time series The arithmetic average concentrations of PM2.5 measured at NYBG, IS52, and QCII were 13.64, 13.89, and 13.16 mg m
3, respectively (Table 1). The differences among these three sites are small. The average concentration of PM2.5 measured at ELIZ was 15.48 mg m
3 while at CHES, the value was 10.73 mg m
3. The major chemical constituents contributing to the PM2.5 at the five STN sites are given in Table 2. The measured components, EC, 2
OC, NH, Naþ , NO
3 , SO4 , and 46 elements (except S and Na) represent more than 92% of PM2.5 mass. This result does not include the conversion of the organic carbon to organic matter (typically a factor of 1.4–1.7) or the conversion of the crustal elements to their oxides. Thus, it appears that there is generally a loss of PM2.5 mass from the Teflon filters that are used for the gravimetric mass measurement.
þ SO2
4 , NH4 and NO3 represent 54%, 61%, 63%, 59% and 67% of PM2.5 mass at NYBG, IS52, QCII, ELIZ and CHES, respectively. Carbonaceous materials (EC and OC) are also important components. Table 2 Fractional contribution to the PM2.5 mass from the major constituents measured in the samples from the NYC area STN sites

EC OC NHþ 4 NO
3 SO2
4 Elements Unmeasured

NYBG (%)

IS52 (%)

QCII (%)

ELIZ (%)

CHES (%)

10.4 28.8 10.7 12.8 25.9 6.3 5.1

9.4 16.3 14.0 16.7 30.4 6.0 7.2

7.3 18.2 13.9 15.7 32.9 6.6 5.4

14.7 20.9 13.7 13.8 31.3 5.7 0.0

5.1 15.2 13.8 13.6 39.6 5.4 7.4

ARTICLE IN PRESS S316

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332


2
Fig. 2. Time series of monthly average concentrations of PM2.5, EC, OC, NHþ 4 , NO3 and SO4 at five STN 2sites.

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

They represent about 39%, 26%, 25%, 36% and 20% of PM2.5 mass at NYBG, IS52, QCII, ELIZ and CHES, respectively. Other elements make up less than 8% of the PM2.5 mass. The marine elements (Naþ and Cl) and crustal elements (Fe, Si, Ca, Al, K, Mn and Ti) are important among the elements. The chemical compositions of PM2.5 at IS52 and QCII are very similar (Table 2). At NYBG and ELIZ, the carbonaceous PM2.5 mass concentration increases to 33% and 36% while the mass of secondary components decreases. The higher percentages of carbonaceous materials at NYBG and ELIS suggest the effects of local emissions. The low percentage of carbonaceous material at CHES implies lower local carbon source emissions. The time series of monthly average concentra
tions for PM2.5 mass and EC, OC, NHþ 4 , NO3 and 2
SO4 measured at the five sites are shown in Fig. 2. There are weak seasonal variations observed in the PM2.5 mass. The PM2.5 concentrations measured at NYBG, IS52, QCII, and ELIZ are very similar to one another. Regression analysis (see supplemental material Table S1) produces r2 values of 0.84, 0.87, 0.83 for NYBG, QCII and ELIS vs. IS52, respectively. With this degree of uniformity, PM2.5 must have common sources. PM2.5 measured at CHES, about 100 km upwind, is also reasonably well correlated with those measured in New York City (r2 is 0.77 for CHES vs. IS52). If PM2.5 measured at CHES is regarded at background concentration, the common source for PM2.5 in metropolitan area of New York City is long range transport. About

S317

69–82% of PM2.5 measured in metropolitan area might come from transport. Local source emissions then only contribute about 18–31% of the observed PM2.5 mass concentrations. Using US EPA’s National Emission Inventories (NEI; EPA, 2003), total PM2.5 primary emissions in Bronx, Queens, Union and Morris counties were estimated (Table 3). Fugitive dust is listed as the major PM2.5 primary emission source in these four counties. It accounts 57%, 38%, 39% and 41% of primary PM2.5 emissions in the Bronx, Queens, Union, and Morris counties, respectively. In the metropolitan area, part of the fugitive dust comes from paved road dust. The average emission strengths are similar in the Bronx and Queens. The PM2.5 concentrations measured in these two counties are similar. The average emission strength in Union County is about half of those in the Bronx and Queens. However, the PM2.5 concentrations measured in Union County are higher than those in the Bronx and Queens. This result implies that the actual emission strength in the Elizabeth urban area is higher than those of the Union County. The average emission strength in Morris Count is estimated to be a factor of 8 lower than those in the Bronx or Queens. High PM2.5 concentrations were measured on 7 July 2002. They are 82, 80, 80, and 83 mg m
3 at NYBG, IS52, QCII and ELIZ, respectively, and higher than the value of 65 mg m
3 in NAAQS. The PM2.5 mass concentration at CHES was unavailable on this date. However, it would also be very high

Table 3 Primary emission sources of PM2.5 in Bronx, Queens, Union and Morris counties (tons mile
2 year
1) as given in the EPA Emissions Inventory Bronx Point Area Coal burning Gas burning Oil burning Wood burning Waste disposal Fugitive dust Other On-Road Gasoline Diesel Off-Road Gasoline Diesel

Queens

Union

Morris

0.89

1.3%

9.03

13.9%

3.18

9.7%

0.14

1.7%

1.32 0.02 0.85 4.51 3.17 38.65 5.33

1.9% 0.0% 1.3% 6.6% 4.7% 56.9% 7.8%

1.09 0.02 0.96 10.87 2.49 24.86 3.59

1.7% 0.0% 1.5% 16.7% 3.8% 38.2% 5.5%

0.05 0.41 0.34 5.66 3.48 12.85 4.63

0.2% 1.3% 1.0% 17.3% 10.6% 39.2% 14.1%

0.01 0.2 0.11 1.64 1.1 3.42 0.61

0.1% 2.4% 1.3% 19.7% 13.2% 41.2% 7.3%

3.06 4.99

4.5% 7.3%

1.91 3.11

2.9% 4.8%

0.78 1.27

2.4% 3.9%

0.16 0.35

1.9% 4.2%

0.52 4.65 67.96

0.8% 6.8% 100%

0.65 6.51 65.09

1.0% 10.0% 100%

0.03 0.07 32.75

0.1% 0.2% 100%

0.21 0.36 8.31

2.5% 4.3% 100%

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

S318

because the peak OC concentration was measured on this date. Fig. 2 shows the elevated monthly average from July 2002 that is driven by the single day high value. This high PM2.5 event was caused by a large forest fire in northern Quebec, Canada (Begum et al., 2005). The smoke plume affected visibility as far south as Virginia. The characteristic of forest fire emissions was a high OC/(EC+OC) ratio of 0.94 (Watson et al., 2001). Similarly high OC concentrations were observed at all five sites on 7 July 2002. The ratios of OC/(EC+OC) were higher than 0.94. Back trajectories analysis clearly showed that the air mass covering the NYC area came from northern Quebec, Canada (Begum et al., 2005). EC concentrations at NYBG, IS52 and QCII show seasonal variations. They are relatively higher in winter and lower in summer. However, there are no obvious seasonal variations for EC concentrations at ELIZ and CHES. The time series of EC for five sites are not correlated with each other with low values of r2. It suggests that EC measured at these sites comes from uncorrelated sources with important impacts of local EC emissions. None of the time series of OC at any of the sites show strong seasonal variations. The time series of OC from NYBG, IS52 and QCII are well correlated with the pairwise values of r2 higher than 0.75. There are lower correlations for the time series of OC at ELIZ and CHES with that from IS52. Except for NO
3 at CHES where the concentra
tions are much lower, the time series of NHþ 4 , NO3 2
and SO4 at the five sites are very similar to one another. The values of r2 between any two sites are all greater than 0.83. Concentrations of NO
3 are higher in winter and lower in summer. In contrast, concentrations of SO2
4 are higher in summer and lower in winter since higher temperatures and

stronger solar radiation enhances the transformation of SO2 during the summer.
2
NHþ are the most important 4 , NO3 and SO4 components of PM2.5 measured in New York. Bari et al. (2003) suggested that on an annual basis, 43% of sulfate, 14% of the sulfur dioxide, 30% of the PM2.5 mass, 27% of HCl, and 24% of HNO3 were attributed to upwind emissions, with the remaining amounts due to emissions in the metropolitan New York. Dutikiewicz et al. (2004) apportioned the contributions of local and transported components of fine particulate sulfate in Queens and concluded that about 44–55% of the SO2
at Queens is 4 transported from distant sources. Lall and Thurston (2005) conclude that nearly half of the total PM2.5 reported in NYC is attributable to transport into the city on annual basis, and more than half (nearly two-thirds) of the PM2.5 in the summertime. For sulfate, they reported that the transported percentage of sulfate is much greater, reaching approximately 90% of the sulfate impacting the site considered in downtown Manhattan. If the concentrations measured at CHES are regard as a measure of the components transported into the region, then based on the ratio of the concentrations of various chemical components at the other sites to the concentrations measured at Chester, about
71–86% of NHþ 4 , 54–65% of NO3 and 493% of 2
SO4 measured in metropolitan area of New York City is transported from distance sources. The peak SO2
and NHþ 4 4 concentrations were measured at NYBG, QCII, ELIZ and CHES on 30 June 2001 with data for this day at IS52 being unavailable. The PM2.5 mass concentrations measured at NYBG, QCII, ELIZ and CHES were 51, 48, 47, and 51 mg m
3, respectively. They are the third highest concentrations for NYBG and QCII, fourth and second highest concentrations for ELIZ

Table 4 Comparison of specific day composition ratio to compositions for Asian and Saharan dusts Element ratio to Al

Asian dusta

Saharan dustb

NYBG 4/22/01

NYBG 7/4/02

IS52 7/4/02

QCII 7/4/02

ELIZ 7/4/02

Ca Fe Si K Ti

1.27 0.48 4.83 0.35 0.052

0.2–0.6 0.52–0.54 2.03–2.32 0.18–0.24 0.052

0.57 1.05 2.92 0.6 0.09

0.34 0.75 2.01 1.1 0.09

0.35 0.82 2.09 0.95 0.1

0.38 0.78 1.99 0.55 0.09

0.29 0.7 1.92 1.56 0.09

a

Nishikawa et al., 2000. Formenti et al., 2003.

b

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

and CHES, respectively. SO2
and NHþ 4 4 represented about 66%, 71% and 74% of the PM2.5 mass at NYBG, QCII and ELIZ, respectively. These values are much higher than the average concentraþ tions. The contribution of SO2
4 and NH4 to PM2.5 at CHES (52%) was at its average value. Examination of the backward trajectory for the air mass arriving in New York City on 30 June 2001 passed through the area of the junction of West Virginia, Ohio and Pennsylvania, an area with many coalfired power plants and other large SO2 emission sources. The air mass then moved eastward in the vicinity of large SO2 emission sources in Pennsylvaþ nia resulting in this high SO2
4 and NH4 event.
þ The peak NO3 and NH4 concentrations were measured at NYBG, IS52, QCII and ELIZ on 9 October 2003. The mass concentrations of PM2.5 with values of 42, 46, 52 and 49 mg m
3 were sixth, third, second, and third highest concentrations measured at NYBG, IS52, QCII and ELIZ, respecþ tively. NO
3 and NH4 comprised about 40%, 51%, 45% and 46% of PM2.5 mass at these four sites, respectively. They are much higher than the average contributions. The 72 h backward trajectory for air mass arriving at New York City on 9 October 2001 was similar to that on 30 June 2001. In both cases, the air mass passed through the area at the intersection of West Virginia, Ohio, and Pennsylvania about 48 h before it arrived to New York City. In this case, the air mass then moved north to Lake Erie and turned to the northeast along Lake Erie and Lake Ontario before moving to NYC. The concentration of the crustal elements Al, Ca, Fe, K, Si and Ti do not show obvious seasonal variations. The time series of K at the five sites are well correlated with the values of r2 higher than 0.82. The correlations for Al and Si among the four sites located in metropolitan area of New York City are also quite high. On 22 April 2001, peak concentrations for Al, Ca, Fe, K, Si and Ti were measured at NYBG. Data for PM2.5 on this day were unavailable for other sites. Back trajectory analysis for this event suggested it was the major dust storm that was observed in Asia in early April and reached the eastern US on this date. The proportions of Ca, Fe, Si, K, and Ti relative to Al (Al assigned as 1) are presented in Table 4. These elemental ratios are in reasonable agreement with the composition of Asian Dust (Nishikawa et al., 2000) although the Si/Al ratio is somewhat lower than is typical for Asian dust indicating some admixture of other crustal materials in transit.

S319

Table 5 SN values for the STN data measured at NYBG, IS52, QCII, ELIZ and CHES NYBG NO
3 Fe Ni S EC NHþ 4 Si Zn Ca K OC Naþ V Ti Cu Br Pb Na Ba Mn Ta K+ Al Cl Sn Cr Co P As Se Sr Mg Ce La Ag Ga Cs Sc W Au Sb Eu Zr In Hg Ir Cd Rb Mo Nb Hf Sm Tb Y

42,800 1456 3193 1309 344 586 128 14.0 6.13 7.72 4.38 1.78 1.89 1.21 1.28 0.97 2.28 1.12 3.97 0.57 0.55 0.46 0.57 0.36 0.29 0.26 0.21 0.13 0.11 0.10 0.08 0.08 0.12 0.07 0.07 0.07 0.09 0.13 0.06 0.06 0.05 0.05 0.04 0.04 0.04 0.04 0.03 0.03 0.02

IS52

QCII

10,350

6819

93

64 24,279

12,812 7747 1361 760 228 149 41.0 58.0 168 25 4.27 1.69 8.55 0.44 3.28 2.95 43 0.41 1.17 1.74 1.17 1.67 2.12 1.02 0.73 0.14 0.10 0.14 0.01 0.11 0.06 0.05 0.08 0.19 0.68 0.62 0.06 0.10 0.03 0.08 0.04 0.08 0.03 0.02 0.13 0.19 0.02

489 310 448 46.0 61.0 184 9.99 4.56 1.44 8.72 0.5 3.36 3.27 34 0.37 1.04 1.31 0.93 1.44 2.17 1.16 0.76 0.13 0.11 0.08 0.00 0.05 0.09 0.06 0.10 0.17 0.49 0.27 0.05 0.12 0.03 0.06 0.05 0.04 0.03 0.02 0.11 0.17 0.02

ELIZ

17

CHES

2.39 7.66

904 281 642 1642 410 102 9.36 8.03 23 6.82 1.31 1.42 1.03 2.92 0.55 3.04 1.76 8.36 0.41 2.62 0.06 0.41 0.31 0.44 0.35 0.29 0.11 0.08 0.05 0.00 0.04 0.08 0.07 0.10 0.12 0.25 0.30 0.08 0.05 0.03 0.13 0.05 0.02 0.01 0.06 0.08 0.02 0.02

57 15 44 75 120 50 0.87 1.19 4.3 3.22 0.52 0.95 0.46 0.89 0.47 2.76 0.27 0.74 0.3 1.01 0.01 0.13 0.25 0.22 0.10 0.08 0.16 0.04 0.08 0.00 0.07 0.10 0.08 0.07 0.11 0.05 0.06 0.03 0.09 0.03 0.09 0.04 0.04 0.01 0.04 0.03 0.01 0.00

ARTICLE IN PRESS S320

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

On 4 July 2002, high crustal element concentrations were measured at four sites with data not available at CHES. The proportions of Ca, Fe, S, K and Ti on this date are given relative to Al (Table 4). It is likely that this dust event was intercontinental transport of Saharan dust. Saharan dust events in the eastern US usually occurring in the summer (June–August) have been reported (Prospero, 1999; Gatz and Prospero, 1996). These values are compared to those of Formenti et al. (2003) for Saharan dust. These ratios are similar to the measured concentrations lending support to the assignment of Saharan dust. Naþ and Cl usually represent the marine aerosol. New York is a coastal city. However, the PM2.5 concentrations of Naþ and Cl measured in metropolitan New York are relatively low. More than 47% of Cl concentrations are below the MDL (Table 1). Winter road salt does not appear to contribute to the Naþ and Cl concentrations in PM2.5. At CHES, more than 82% Cl concentrations were lower than the MDL. There is a weaker correlation between two variables that generally come from a common source. The average ratios of Naþ =Cl are 9.8, 2.7, 4.0, 5.7 and 33 at NYBG, IS52, QCII, ELIZ and CHES, respectively. They are

much higher than the value of 0.56 in fresh marine source for PM2.5 and coarse particles (Watson et al., 1994) because of the depletion of Cl through reactions with gaseous acids. 4. Source apportionment The PMF model was used here to identify PM2.5 sources and estimate source contributions. It is a multivariate receptor modeling has been described in detail by Paatero (1997) and is implemented in the PMF2 program. This program has now been widely used to analyze airborne particulate matter composition data (Lee et al., 1999; Ramadan et al., 2000; Kim et al., 2003; Kim and Hopke, 2004a; Buzcu et al., 2003; Liu et al., 2003a,b; Qin and Oduyemi, 2003; Qin et al., 2002; Chueinta et al., 2000). 4.1. Data screening The application of PMF depends on variable selection and uncertainty estimation. Since prior to July 2003, the STN data were not accompanied by uncertainties, the comprehensive set of error structures estimated by Kim et al. (2005) was used in this study. Environmental data typically has higher

Fig. 3. Scatter plots of measured and model predicted PM2.5 at NYBG, IS52, QCII, ELIZ and CHES.

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

noise content in some variables than others. The ‘‘bad variables’’ that provide more noise than signal should be excluded from the analyses. The signal-tonoise (SN) ratios (Paatero and Hopke, 2003) for the measured variables at each of the 5 sites are presented in Table 5. Because of the high correlation between the XRF S values and the ion chromatography SO2
4 values measured at the five sites, S was selected for PMF analyses and only S will be

S321

subsequently discussed. Both ion and element concentrations for Na and K are reported in STN data. The percentage of concentrations higher than the MDL for ion Na is much higher than that for element Na. Most of ion K concentrations are below the MDL. Thus, Na ion and elemental K were selected for further analysis. The values of SN shown in Table 5 indicate that
EC, OC, NHþ 4 , NO3 , Fe, Ni, Si, Zn, Ca, Cu, and Br

Table 6 Average contributions (mg m
3) of identified sources to PM2.5 concentrations

Secondary sulfate Secondary nitrate Soil dust Aged sea salt Oil combustion Spark ignition Highway vehicle Diesel

NYBG

IS52

QCII

ELIZ

CHES

5.7775.26 2.1072.45 1.4871.06 0.6870.69 0.5270.43 2.1471.65

7.2075.88 2.5772.44 1.0770.76 0.4970.61 1.3671.30 1.1170.91

4.8774.47 1.8172.55 0.7570.64 0.4470.61 1.2571.27 2.5572.09

6.4376.71 1.0571.41 0.9970.78 0.8771.74

0.4370.34

0.4570.34

6.6776.89 1.9872.38 0.9971.41 1.1471.37 0.8570.95 2.5771.93 1.3071.26 2.1671.32

Fig. 4. Source profiles for the secondary sulfate factors.

3.0172.40 0.3570.36

ARTICLE IN PRESS S322

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

Fig. 5. Time series of source contributions for the secondary sulfate factors.

measured at the five STN sites are good variables because their SN ratio are higher than 2. These good variables were selected for analysis. V, Ti, Cl, Pb, Mn, Ba, Al, Ta, Sn, Cr, As and Se were either good (SN42) or weak variables at five sites because their SN are higher than 0.2, but less than 2. They were also selected for analysis. Some trace elements were bad variables at some sites and weak variables at others. They were selected for analysis if they can serve as source markers. For example, Mg was bad variable at CHES. It was a weak variable at other four sites. Mg was selected for inclusion in the PMF model analysis at NYBG, IS52, QCII and ELIZ because it could serve as a marker of marine aerosol.
A total of 27 valuables, EC, OC, NHþ 4 , NO3 , S, þ Na , K, Fe, Ni, Si, Zn, Ca, Cu, Br, V, Ti, Cl, Pb, Mn, Ba, Al, Ta, Sn, Cr, As, Se and Mg, were selected from STN data measured at NYBG, IS52 and QCII for the PMF analyses. Except for As, these 26 valuables and Eu, P, Sc, Sr and Zr were selected from the ELIZ data. Except for As and Mg,

these 25 valuables plus Ce and Se were selected from the STN data at CHES. The ‘‘weak variables’’ provide comparable signal and noise. The weights of weak variables were decreased in the analysis. Pb, As, Se, Al, Cr, As, Ta, Sn, Ba and Mg at NYBG, Cr, As, Ta, Sn, Ba and Mg at IS52 and QCII, Al, Pb, Ba, Ta, Se, Sn, Cr, Mg, Eu, P, Sc, Sr and Zr at ELIZ, and, Ti, Cr, Mn, V, Cl, Pb, Ta, Ba, Sn, Al, Se, Ce and Sc at CHES are weak variables. The weak variables were downweighted by increasing their uncertainties by a factor of 5. The Naþ values obtained during the period in which the Nylon filters were flagged as likely to be contaminated were downweighted by a factor of 30. Concentrations below the MDL were replaced by half of the MDL, and 5/6 of MDL was set as overall uncertainty. Missing concentration was replaced by the geometric mean of the measured concentrations, and its uncertainty was set to 4 times this geometric mean value. The measured PM2.5 mass concentrations were included as an independent variable in PMF model

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

S323

Fig. 6. Source profiles for the secondary nitrate factors.

analysis to obtain the mass apportionment directly without the usual multilinear regression. The PM2.5 mass values were downweighted in the analysis by setting the uncertainties to 4 times the mass concentrations. The initial results were then normalized by the apportioned PM2.5 mass concentrations so that the quantitative source contributions for each source were obtained. PMF2 was run with different factor numbers and the results that produced a good fit to the data as shown by the Q-value and the scaled residuals as well as providing the most interpretable results. The robust mode was used to reduce the influence of extreme values on the solutions. The global optima of the PMF solutions were tested using multiple starting values. To examine the likely directions for local point sources, the conditional probability function (CPF) (Kim and Hopke, 2004a) were calculated for each of the resolved sources at each of the five sites. The CPF provides a polar plot showing the probability that the

sources that provided the source contributions are located in the directions of high probability. 4.2. Source analysis results Previously, Li et al. (2004) applied PMF to a limited set of PM2.5 data measured using six hour time interval samples collected at the NYC Supersite at Queens College, New York City during July 2001. Six PM2.5 sources of secondary sulfate, secondary nitrate, motor vehicle, road dust, sea salt and oil combustion were identified successfully. Ito et al. (2004) examined the STN data from these same sites for the limited period of April 2001– November 2002 using Absolute Principal Components Analysis and PMF. Eight factors were extracted using APCA. However, they aggregated these factors into 4 sources, soil, secondary sulfate, traffic, and residual oil/incineration, to improve the interpretability of the results and agreement with the PMF results.

ARTICLE IN PRESS S324

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

In the current study, six sources for CHES and QCII, seven sources for NYBG and IS52 and eight sources for ELIZ have been identified successfully by PMF2. The source profiles and time series of source contributions derived by PMF model for these five STN sites are shown in figures in the supplemental material for this paper. In the profile plots, the error bars provided by the PMF2 program are displayed. These errors are estimated by a procedure similar to that described by Roscoe and Hopke (1981). The scatter plots to compare the reconstructed PM2.5 mass concentrations using the model results and measured PM2.5 mass concentrations are shown in Fig. 3. The high values of r2 (all greater than 0.89) indicate that the identified sources account for most of the variations in PM2.5 concentrations at these five sites. The average source contributions are summarized in Table 6. A source with high loading for S and NHþ 4 is identified by PMF model at all five STN sites. These profiles are shown in Fig. 4 and the source

contributions shown in Fig. 5 demonstrate obviously season variations with high concentrations in summer and low concentration in winter. This source represents secondary sulfate formed from SO2 emitted upwind of New York City. The derived ratios of S=NHþ 4 for this source are 1.00, 1.02, 1.08, 0.88 and 1.00 at NYBG, IS52, QCII, ELIZ and CHES, respectively. They closely approach the stoichiometric ratio of S=NHþ 4 for (NH4)2 SO4 of 1.13. Similar to previous studies (Kim et al., 2003) in the eastern US, secondary sulfate is the largest contributor to PM2.5. On average, it contributes 38–51% of concentration for PM2.5 mass at these STN sites. These values are similar to those observed in other studies of NYC. Assuming the concentrations observed at Chester represent the regional aerosol transported into the region, these analyses suggest that more than 93% of sulfate measured in the New York metropolitan area was transported into the metropolitan New York City area from sources outside of the region. This value

Fig. 7. Time series of source contributions for the secondary nitrate factors.

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

is in agreement with the estimates of Lall and Thurston (2005), but are substantially higher than those of Bari et al. (2003) and Dutkiewicz et al. (2004) who based their analysis on a trajectory analysis rather than a comparison to upwind sites that are thought to represent areas that are not strongly affected by sources within the NY metropolitan area. þ A source with high loading for NO
3 and NH4 was identified at all sites. The source profiles are shown in Fig. 6 and the time series of contributions are presented in Fig. 7. The ammonium nitrate factor shows strong seasonal variations with high concentrations in winter and low concentrations in summer similar to earlier studies based on shorter 4 data records. The observed ratios of NO
3 =NH for these source profiles are 3.7, 2.7, 3.8, 3.3 and 2.9 at NYBG, IS52, QCII, ELIZ and CHES, respectively. These values approach the stoichiometric ratio of þ NO
3 =NH4 for NH4NO3 (3.44). Secondary nitrate is a major component of PM2.5. It contributes about

S325

8–18% of concentration for PM2.5. Again comparing the contributions at the urban sites to those at Chester suggests about 54–65% of the ammonium nitrate measured in metropolitan New York City is the result of transport into the region. A source with high loading for Si, Fe, Ca and Al is identified at all sites (profiles not shown). This source represents the suspended soil. Si is the major component for soil dust. The source contributions do not show average seasonal variations (Fig. 8). The peak contributions appear to be intercontinental dust events as discussed above. Ito et al. (2004) also point to this event. There are some anomalies in the results. The Fe concentration is higher than that for Si in the source profiles at NYBG and IS52. The soil dust sources derived at NYBG and IS52 must be intermixed with other source materials, but not to a sufficient extent that a separate source can be resolved. The contributions of soil dust to PM2.5 are 6–11% at five STN sites. According to the EPA’s National Emission Inventories, fugitive dust is the

Fig. 8. Time series of source contributions for the soil factors.

ARTICLE IN PRESS S326

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

Fig. 9. Source profiles for oil-fired power plants.

Fig. 10. Time series of source contributions for oil-fired power plants.

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

most important primary PM2.5 emission source in this area. It accounts about 38–57% of local emissions of PM2.5 in the Bronx, Queens, Union and Morris counties (Table 3). A source with significant concentrations of Ni and V was identified at NYBG, IS52, QCII and ELIZ (Fig. 9), but was not resolved from the data at Chester. Ni and V are tracers of residual oil combustion. The source contributions at NYBG, IS52, and QCII show strong seasonal variations with higher concentrations in winter and lower concentrations in summer (Fig. 10). At these sites, this source profile is indicative of oil-fired power plants. However, a very different pattern is observed at Elizabeth. The CPF for the oil combustion factor

S327

at Elizabeth points in the direction of the Port of Elizabeth and thus, in this case, the residual oil factor may also include significant contributions from ship engines from the Port of Elizabeth and related activities. The oil combustion profiles varied with site. Loading for OC is higher or much higher than the EC at NYBG, IS52 and ELIZ. However, the loading for OC is much lower than that for EC at QCII. A diesel factor could not be resolved at QCII and thus, this factor may be a combination of oil combustion and diesel emissions. The oil combustion contributes about 4–11% of concentration for PM2.5 at these four sites. These values are much higher than the EPA estimated contribution of oil

Fig. 11. Source profiles for spark-ignition vehicle emissions.

ARTICLE IN PRESS S328

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

burning to the emissions of PM2.5 (Table 3). The oil combustion source derived by PMF model may be mixed with other sources such as waste incineration as suggested by Ito et al. (2004). Motor vehicle emissions were apportioned into two source profiles at two of the three New York City sites and CHES. These two sources are sparkignition vehicle emissions with high loadings for OC and diesel vehicle emissions with high for EC and some metal elements. The source contributions of diesel were lower on weekends as has been observed in a number of other recent studies (Kim and Hopke, 2004a). However, this source resolution needs to be viewed in terms of recent work by Shah et al. (2004). They report that diesels operating at very slow speeds (creep) and in stop and go traffic produce OC/EC ratios that are very similar to

typical spark-ignition emissions with approximately equal amounts of OC and EC. Only under more continuous motion at higher speeds (transient and cruise) is there significantly more EC than OC in the emissions. Thus, the ‘‘diesel’’ profile that is being extracted probably represents only diesel vehicles moving at reasonable speed under fluid traffic conditions. Diesels in stop and go traffic are likely to be apportioned into the ‘‘spark-ignition’’ category. Fig. 11 shows sources with high loadings for OC with less EC at all five sites. The source contributions (Fig. 12) do not show seasonal variations. They also do not show weekday and weekend variations. OC and EC are major components for motor vehicle emissions (Watson et al., 1994). With higher OC than EC, a lack of trace species, and

Fig. 12. Time series of source contributions for spark-ignition vehicles.

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

Fig. 13. Source profiles for diesel vehicle emissions.

Fig. 14. Time series of source contributions for diesel vehicle emissions.

S329

ARTICLE IN PRESS S330

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

approximately equal weekday and weekend contributions, this source is assigned to be spark ignition vehicle emissions. The ratios of OC/EC for this source derived by PMF model are 5.3, 2.8 and 3.7 at NYBG, IS52 and QCII, respectively. They are higher than the typical ratio of 2.1 for fresh gasoline exhaust (Cadle et al., 1999). The spark-ignition factor resolved at Chester has a very high ratio of OC to EC (30) suggesting a poorer separation of diesel from spark-ignition. In Elizabeth, two high OC factors were resolved. One was identified as highway traffic because the CPF plot for this factor aligns with the direction of the New Jersey Turnpike while the ‘‘local’’ traffic factor does not show a clear directionality. The source contributions of spark ignition vehicles are 8–22% at the five sites. They are higher than EPA highway vehicle contribution to primary emission sources of PM2.5 listed in Table 3. Source profiles (Fig. 13) with high loadings for EC and metal elements such as Cu, Zn and Ca were identified at CHES, ELIZ, NYBG and IS52, but not at QCII. The source contributions (Fig. 14) are higher in winter and lower in summer. This source is assigned to be the diesel vehicle emissions. The source contributions were much lower on weekend days when there is a much lower level of heavy duty diesel truck and bus use. Diesel emissions contribute between 3% and 15% of the PM2.5 at the five sites. The emissions inventory indicates that highway vehicle emissions account for about 6–12% local PM2.5 emissions (Table 3). A source with high loading for Naþ , S and NO
3 was identified at all sites. The source contributions do not show seasonal variations. It is suggested that this is aged sea salt with Cl replaced by SO2
4 and NO
3 . It appears that there is sufficient production of gaseous acids to produce significant displacement of chlorine from the sea salt particles. This factor also suggests the possibility of heterogeneous conversion of S in sea-salt particles as has been previously observed over the North Atlantic Ocean (Sievering et al., 1991). Thus, in winter when there is lower production of gas-phase acids, there could still be displacement by in situ conversion of SO2 within the particles. Aged sea salt contributes about 3–7% of concentration for PM2.5 at the five sites. These results are quite consistent with those seen by Kim and Hopke (2005) for sites in New Jersey and Delaware. The emissions inventory suggest wood combustion as a major PM2.5 source. It is normally

observed as a factor containing OC, EC, and K. However, such a source profile could not be isolated at any of the sites including Chester. Kim and Hopke (2004b) also did not identify a wood combustion source at Brigantine, NJ or at any of the NJ or Delaware sites (Kim and Hopke, 2005). 5. Conclusion Secondary sulfate, secondary nitrate, soil dust and aged sea salt were ubiquitous sources and were identified at all five STN sites. Oil combustion was identified at all four urban sites. The sites within NYC appear likely to have been affected by oil-fired power plants whereas the Elizabeth, NJ site shows the apparent influence of ship emissions. Motor vehicle emissions were identified as a source at three sites with their contributions apportioned between compression and spark-ignition vehicles. At the Elizabeth, NJ site, three motor vehicle sources were identified whereas at Queens College, it was not possible to separate the motor vehicle emissions into diesel and gasoline vehicle contributions. Waste disposal emissions were identified at the two sites in the Bronx. Wood burning was listed by EPA as an import primary emission source for PM2.5 in Queens, Union and Morris counties. Wood burning is supposed to account 16.7%, 17.3% and 19.7% of primary PM2.5 emission in these three counties. However, there does not seem to be sufficient measured potassium concentrations that such emissions would be expected to generate (Watson et al., 2001) and thus, wood burning could not be identified at any of the sites. In contrast, the source contributions derived by PMF model for oil combustion and motor vehicle are higher than the contributions to primary emissions. Acknowledgements This research was supported in part by the New York Sate Energy Research and Development Authority under Agreement No. 7919 and by the U.S. Environmental Protection Agency through Science to Achieve To Results (STAR) grant number RD83107801. Appendix A. Supplementary Material Supplementary data associated with this article can be found in the online version at doi:10.1016/ j.atmosenv.2006.02.025.

ARTICLE IN PRESS Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

References Bari, A., Dutkiewicz, V.A., Judd, C.D., Wilson, L.R., Luttinger, D., Husain, L., 2003. Regional sources of particulate sulfate, SO2, PM2.5, HCl, and HNO3, in New York, NY. Atmospheric Environment 37, 2837–2844. Begum, B.A., Kim, E., Jeong, C.H., Lee, D.W., Hopke, P.K., 2005. Evaluation of the potential source contribution function using the 2002 Quebec forest fire episode. Atmospheric Environment 39, 3719–3724. Buzcu, B., Fraser, M.P., Kulkarni, P., Chellam, S., 2003. Source identification and apportionment of fine particulate matter in Houston, TX, using positive matrix factorization. Environmental Engineering Science 20, 533–545. Cadle, S.H., Mulawa, P.A., Hunsanger, E.C., Nelson, K., Ragazzi, R.A., Barrett, R., Gallagher, G.L., Lawson, D.R., Knapp, K.T., Snow, R., 1999. Composition of light-duty motor vehicle exhaust particulate matter in the Denver, Colorado area. Environmental Science and Technology 33, 2328–2339. Chueinta, W., Hopke, P.K., Paatero, P., 2000. Investigation of sources of atmospheric aerosol at urban and suburban residential areas in Thailand by positive matrix factorization. Atmospheric Environment 34 (20), 3319–3329. Dutkiewicz, V.A., Qureshi, S., Khan, A.R., Ferraro, V., Schwab, J., Demerjian, K.L., Husain, L., 2004. Sources of fine particulate sulfate in New York. Atmospheric Environment 38, 3179–3189. EPA, 2003. 2002 National Emission Inventory Data & Documentation. http://www.epa.gov/ttn/chief/net/2002inventory. html Formenti, P., Elbert, W., Maenhaut, W., Haywood, J., Andreae, M., 2003. Chemical composition of mineral dust aerosol during the Saharan Dust Experiment (SHADE) airborne campaign in the Cape Verde region, September 2000. Journal of Geophysical Research 108 (D18), 8576. Gatz, D.F., Prospero, J.M., 1996. A large silicon-aluminum aerosol plume in central Illinois: North African desert dust? Atmospheric Environment 30, 3789–3799. Ito, K., Xue, N., Thurston, G.D., 2004. Spatial variation of PM2.5 chemical species and source-apportioned mass concentrations in New York City. Atmospheric Environment 38, 5269–5282. Kim, E., Hopke, P.K., 2004a. Source apportionment of fine particles at Washington, DC utilizing temperature resolved carbon fractions. Journal of Air and Waste Management Association 54, 773–785. Kim, E., Hopke, P.K., 2004b. Improving source identification of fine particles in a rural northeastern U.S. area utilizing temperature resolved carbon fractions. Journal of Geophysical Research 109, D09204. Kim, E., Hopke, P.K., 2005. Identification of fine particle sources in mid-Atlantic US area. Water, Air, Soil Pollution 168, 391–421. Kim, E., Hopke, P.K., Edgerton, E.S., 2003. Source identification of Atlanta aerosol by positive matrix factorization. Journal of Air and Waste Management Association 53, 731–739. Kim, E., Qin, Y., Hopke, P.K., 2005. Estimation of organic carbon blank values and error structures of the speciation trends network data for source apportionment. Journal of Air and Waste Management Association 55, 1190–1199.

S331

Lall, R., Thurston, G.D., 2005. Identifying and quantifying transported vs. local sources of New York City PM2.5 fine particulate matter air pollution. Atmospheric Environment, submitted for publication. Lee, E., Chan, C.K., Paatero, P., 1999. Application of positive matrix factorization in source apportionment of particulate pollutants in Hong Kong. Atmospheric Environment 33, 3201–3212. Li, H., Hopke, P.K., Husain, L., Qrueshi, S., Dutkiewicz, W.A., Schwab, J.J., Drewnick, F., Demerjian, K.L., 2004. Sources of fine particle composition in New York City. Atmospheric Environment 38, 6521–6529. Liu, W., Hopke, P.K., Han, Y.J., Yi, S.-M., Holsen, T.M., Cybart, S., Kozlowski, K., Milligan, M., 2003a. Application of receptor modeling to atmospheric constituents at Potsdam and Stockton, NY. Atmospheric Environment 37 (36), 4997–5007. Liu, W., Hopke., P.K., VanCuren, R.A., 2003b. Origins of fine aerosol mass in the western United States using positive matrix factorization. Journal of Geophysical Research 108 (D23), 1 Art. No. 4716 DEC 3 2003. Nishikawa, M., Hao, Q., Morita, M., 2000. Preparation and evaluation of certified reference materials for Asian mineral dust. Global Environmental Research 4 (1), 103–113. Paatero, P., 1997. Least square formulation of robust nonnegative factor analysis. Chemometrrics and Intelligent Laboratory Systems 37, 23–35. Paatero, P., Hopke, P.K., 2003. Discarding or downweighting high-noise variables in factor analytic models. Analytica Chimica Acta 490 (1–2), 277–289. Prospero, J.M., 1999. Long-term measurements of the transport of African mineral dust to the southeastern United States: Implications for regional air quality. Journal of Geophysical Research-Atmospheres 104 (D13), 15917–15927. Qin, Y., Oduyemi, K., 2003. Atmospheric aerosol source identification and estimates of source contributions to air pollution in Dundee, UK. Atmospheric Environment 37, 1799–1809. Qin, Y., Oduyemi, K., Chan, Y.L., 2002. Comparative testing of PMF and CFA models. Chemometrics and Intelligent Laboratory Systems 61, 75–87. Ramadan, Z., Song, X.H., Hopke, P.K., 2000. Identification of sources of Phoenix aerosol by positive matrix factorization. Journal of Air and Waste Management Association 50L, 1308–1320. Roscoe, B.A., Hopke, P.K., 1981. Error estimation of factor loadings and scores obtained with target transformation factor analysis. Analytica Chimica Acta 132, 89–97. RTI, 2003. Extended semi-annual data summary report for the chemical speciation of PM2.5 filter samples projects. RTI/ 07565/06DS. Shah, S.D., Cocker, D.R., Miller, J.W., Norbeck, J.M., 2004. Emission rates of particulate matter and elemental and organic carbon from in-use diesel engines. Environmental Science & Technology 38 (9), 2544–2550. Sievering, H., Boatman, J., Galloway, J., Keene, W., Kim, Y., Luria, M., Ray, J., 1991. Heterogeneous sulfur conversion in sea-salt aerosol particles: the role of aerosol water content and size distribution. Atmospheric Environment 25A, 1479–1487. Tolocka, M.P., Solomon, P.A., Mitchell, W., Norris, G.A., Gemmill, D.B., Wiener, R.W., Vanderpool, R.W., Homolya, J.B., Rice, J., 2001. East versus West in the US: Chemical

ARTICLE IN PRESS S332

Y. Qin et al. / Atmospheric Environment 40 (2006) S312–S332

characteristics of PM2.5 during the winter of 1999. Aerosol Science and Technology 34, 88–96. Watson, J.G., Chow, J.C., Lu, Z., Fujuta, E.M., Lowenthal, D.H., Lawson, D.R., 1994. Chemical mass balance source apportionment of PM10 during the southern California air quality study. Aerosol Science and Technology 21, 1–36.

Watson, J.G., Chow, J.C., Houck, J.E., 2001. PM2.5 chemical source profiles for vehicle exhaust, vegetative burning, geological material, and coal burning in Northwestern Colorado during 1995. Chemosphere 43, 1141–1151.