Characterization of fine particulate matter in Ohio: Indoor, outdoor, and personal exposures

ARTICLE IN PRESS

Environmental Research 106 (2008) 62–71 www.elsevier.com/locate/envres

Characterization of fine particulate matter in Ohio: Indoor, outdoor, and personal exposures Kevin C. Crista,, Bian Liuc, Myoungwoo Kima, Seemantini R. Deshpandea, Kuruvilla Johnb a

Air Quality Center, Department of Chemical Engineering, Ohio University, 177 Stocker Center, Athens, OH 45701, USA b Department of Environmental Engineering, Texas A&M University, Kingsville, TX, USA c School of Public Health, University of Michigan, Ann Arbor, MI, USA Received 9 March 2007; received in revised form 14 June 2007; accepted 27 June 2007 Available online 31 August 2007

Abstract Ambient, indoor, and personal PM2.5 concentrations were assessed based on an exhaustive study of PM2.5 performed in Ohio from 1999 to 2000. Locations in Columbus, one in an urban corridor and the other in a suburban area were involved. A third rural location in Athens, Ohio, was also established. At all three locations, elementary schools were utilized to determine outdoor, indoor, and personal PM2.5 concentrations for fourth and fifth grade students using filter-based measurements. Three groups of 30 students each were used for personal sampling at each school. Continuous ambient PM2.5 mass concentrations were also measured with tapered element oscillating microbalances (TEOMs). At all three sites, personal and indoor PM2.5 concentrations exceeded outdoor levels. This trend is consistent on all week days and most evident in the spring as compared to fall and winter. The ambient PM2.5 concentrations were similar among the three sites, suggesting the existence of a common regional source influence. At all the three sites, larger variations were found in personal and indoor PM2.5 than ambient levels. The strongest correlations were found between indoor and personal concentrations, indicating that personal PM2.5 exposures were significantly affected by indoor PM2.5 than by ambient PM2.5. This was further confirmed by the indoor to outdoor (I/O) ratios of PM2.5 concentrations, which were greater when school was in session than non-school days when the students were absent. r 2007 Elsevier Inc. All rights reserved. Keywords: PM2.5; Personal exposure; Rural; Urban; CPF

1. Introduction Fine particulate matter, or PM2.5, refers to a mixture of solid and liquid atmospheric particles with an aerodynamic diameter (dae) less than or equal to 2.5 mm. It arises mainly from anthropogenic sources such as fossil fuel combustion by electric utilities and motor vehicles, wood burning, and the smelting or other processing of metals. PM2.5 consists of sulfate, nitrate, ammonium, trace elements, carbon compounds, and water (Chow et al., 1994; Chow and Watson, 1998). The majority of PM2.5 components are secondary materials, derived from the chemical reactions of gaseous precursors such as SO2, NOx, volatile organic

Corresponding author. Fax: +1 740 593 4751.

E-mail address: [email protected] (K.C. Crist). 0013-9351/$ - see front matter r 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.envres.2007.06.008

compounds (VOCs), organic and elemental carbon, and a range of trace metals. Recent environmental epidemiological studies suggest that ambient PM2.5, measured at a fixed outdoor site, is more strongly correlated with adverse health effects than particles in other size ranges (Dockery et al., 1993; Schwartz et al., 1996; Pope et al., 1995; Liao et al., 1999; Spengler et al., 1996; Klemm et al., 2000; Pope and Dockery, 2006; Schlesinger et al., 2006; WHO, 2005a, b, 2006). The health effects range from slight respiratory symptoms to increased mortality rates. Certain population groups such as seniors, respiratory and cardiovascular patients, and children are most susceptible to particle pollution. To protect the general public from PM2.5 pollution, the United States Environmental Protection Agency (USEPA) established a standard for the ambient PM2.5 in 1997 (Federal Register, 1997).

ARTICLE IN PRESS K.C. Crist et al. / Environmental Research 106 (2008) 62–71

The associations between ambient PM2.5 concentrations and a variety of adverse health outcomes suggest that ambient concentration may be an indicator for personal PM2.5 exposure, and ambient PM2.5 should correlate well with indoor and personal PM2.5 concentrations (Wilson and Suh, 1997). However, studies have shown inconsistent correlations between outdoor, indoor, and personal PM2.5 levels, with correlation coefficients (R) ranging from below zero to close to one (Wilson and Burton, 1995; Wallace, 1996, 2000; Watson et al., 1997; Wilson et al., 2000; Goswami et al., 2002; Allen et al., 2003). The large range of R-values, on one hand, reflects that personal PM2.5 exposure is impacted by individual lifestyles (e.g. sedentary indoor vs. active outdoor type) and the characteristics of the microenvironment (e.g. poor vs. good ventilation), where the subjects spend time (Wallace, 1996, 2000). On the other hand, it suggests that the interpretation of ambient PM2.5 concentrations as a proxy of personal PM2.5 exposure is perhaps questionable. Since the total personal PM2.5 exposure is a result of PM2.5 concentrations in various microenvironments, a more accurate personal PM2.5 exposure estimation is measured by a personal exposure monitor worn by the subject or obtained by averaging the time-weighted concentration of different microenvironments (Wilson et al., 2000). Limited information is available regarding correlations between personal, indoor, and outdoor PM2.5 concentrations. Most studies focus on senior subjects and respiratory patients, while only a few studies investigate children’s personal PM2.5 exposure and its relationships with indoor and outdoor levels (USEPA, 1996, 2001; Patterson and Eatough, 2000). Studies that focus on children are often conducted in homes, an environment quite different from a classroom (e.g. Janssen et al., 1999). This study evaluates the correlations between personal PM2.5 exposures, indoor, and outdoor PM2.5 concentrations, using data from a 2year health-based study conducted in three elementary schools in southeastern Ohio. These data are part of the Air Pollution and Pediatric Health Impact Assessment (APPHIA) project. In this paper, temporal trends of PM2.5 and the transport of ambient particulate pollutants were also studied to research the possibility of an inherent pattern in the outdoor, indoor, and personal concentrations of PM2.5. This research provides valuable information in examining the relationship between personal, indoor, and outdoor PM2.5 levels. 2. Methods 2.1. Study sites This study was conducted in central and southeastern Ohio (Fig. 1) from January 1999 through August 2000. Two schools in Columbus (Koebel Elementary School and New Albany Elementary School) and one school in Athens (East Elementary School) comprised the monitoring sites. Approximately 30 students of fourth and fifth grades at each site were involved. The three elementary schools are in residential neighborhoods. Koebel is located to the south of Columbus in the industrial center

63

Fig. 1. Locations of the three sites involved in the study.

of the city. The industrial activities include foundries, plastic facilities, and gravel/quarrying operations. This site is also located within 0.5 km of a major transportation artery. New Albany is approximately 8 km northeast of downtown Columbus in the Franklin County and is approximately 32 km northeast of Koebel. New Albany is a bedroom community of Columbus with few commercial facilities and no significant industrial operations within the municipal boundary. Since the prevailing winds are from the southwest, transport of PM2.5 precursors from the Columbus area may influence the particle pollution at this site. The third site, Athens, is approximately 120 km southeast of Columbus and is a rural location. Athens is a university town with a population of 20,000. The site is in a residential area and the only significant local stationary pollution source is Ohio University’s coal-fired power plant. Athens is about 32 km west of the Ohio River Valley, which has numerous coal-fired power generation facilities, chemical manufacturing facilities, and industrial operations. Athens is an upwind remote site for the Department of Energy’s Ohio River Valley PM2.5 monitoring projects (Ambient Monitoring, The Upper Ohio River Valley Project (UORVP)). The Koebel School is a one-story building while the New Albany and Athens schools are both two-storey buildings. Classrooms at each elementary school used for indoor monitoring were selected as far as possible from the kitchen facilities to reduce the impact of cookinggenerated PM2.5. The classrooms at Athens and New Albany are air conditioned. Koebel elementary school has a central heating system but no central air conditioning system. All three schools use natural ventilation during the warm months, so classroom windows are typically open during the months of April, May, June, September, and part of October. With central air conditioning, Athens and New Albany may close their windows during very warm days. However, windows are open a majority of the school days during the spring and fall.

2.2. Measurement methods The monitoring scheme is outlined in Table 1. Personal, indoor, and ambient PM2.5 concentrations were measured concurrently at all sites using Whatman 37, 37, and 47 mm Teflon filters with 2-mm pores size, respectively. In addition, continuous ambient PM2.5 measurements were also conducted.

ARTICLE IN PRESS 64

K.C. Crist et al. / Environmental Research 106 (2008) 62–71

Table 1 Summary of the sampling time and samplers used Sampling type

Sample schedule Sampling period

Sampling time

Monitor s

Vendor

Continuous Ambient PM2.5

January 1999–August 2000

Daily-24 h

TEOM

Filter-based ambient PM2.5

January–June 1999

School day—8 a.m. to 3 p.m. Daily-24 h School day—8 a.m. to 3 p.m.

ACCUTM System

Filter-based indoor PM2.5

September 1999–August 2000 January 1999–June 2000 June 2000–August 2000

Filter-based personal PM2.5

January 1999–August 2000

Non-school day daily24 h School day—8 a.m. to 3 p.m.

Series 1400a

Sampling pump (URG2000-30EH); Cyclone (URG-3000-02Q)

University Research Glass (URG) Corporation

Impactor (URG-200025F)

University Research Glass (URG) Corporation BIOS International Co.

Model AirPros 6000D

2.2.1. Personal PM2.5 measurements Personal PM2.5 measurements were conducted using impactors with a PM2.5 size cut point. The inlets were placed within 3–5 in of the breathing zone directly below the front of the shoulders of the subject minimizing the impact of expired air as a potential PM2.5 sources. Air sampling pumps were operated at a flow rate of 5 L/min and placed in an acoustic shell to reduce pump noise levels. During non-sampling time, personal pumps were charged using direct plug-in converters. The personal PM2.5 sampler devices were placed in personal backpacks (Camelback, personal backpack, H.A.W.G.100oz) worn by selected students. Students were instructed that the sampler should follow them as closely as possible, but they were allowed to place the sampler nearby while involved in indoor sedentary activities (i.e., reading, writing) during which the inlet should be no more than 0.7 m away from the breathing zone or activities during which wearing the sampler would be too inconvenient or impossible (i.e., swimming). One student in each class wore a personal sampler and was representative of the personal exposure level of the entire class. To that end, the teacher ensured that the student moved in a group and wore the sampler at all times. Three groups of 30 students each were sampled in each school per day. Of those 30 students, one student wore the personal sampler. Therefore, approximately 3.3% of the student population was selected to represent the entire community. Selection of students to wear the sampler was based on their capability and willingness to participate. 2.2.2. Indoor PM2.5 measurements Indoor monitors were operated at 10 L/min using flow-controlled indoor sampling pumps. The inlets (a cyclone with 2.5 mm size cut) were placed approximately 1.2 m above the floor. Indoor monitors were conducted typically from 8:00 a.m. to 3:00 p.m. from Monday to Friday throughout the school year, matching the personal PM2.5 sampling period. During the summer months of 2000 (June–August), indoor monitoring was extended to 24 h. 2.2.3. Ambient PM2.5 measurements Continuous ambient PM2.5 concentrations were measured using a Tapered Element Oscillating Microbalances (TEOMs) series 1400a, which has been widely used in PM monitoring around the world (Meyer et al., 2000). Airflow of 16.67 L/min drawn through a cyclone (the same model used for the indoor measurement) is isokinetically split into a 3 L/min main flow for the continuous ambient PM2.5 sample, and a 13.67 L/min bypass flow. The bypass flow is fed into an Automatic Cartridge

Rupprecht & Patashnick (R&P) Co., Inc.

Collection Unit (ACCU) System, which is connected to the TEOMs. The ACCU system redirects the bypass flow through one of eight flow channels allowing measurement of filter-based 24-h outdoor PM2.5 concentrations. The 3 L/min air stream passes through a Teflon-coated borosilicate glass fiber filter, which is on the narrow end of a hollow tapered tube. The frequency of the tapered element changes according to filter mass changes under the control of an electronic circuit. A precision electronic counter senses this frequency change in a 2-s period and computes the mass of the particulate collected on the filter. The TEOM was operated at 50 1C to reduce humidity, which might have led to the loss of semi-volatile materials (Meyer et al., 2000). The TEOMs at Athens and Koebel were located on the roofs of the buildings used for the indoor PM2.5 monitoring. The inlets for the monitors were within a 10-m distance above the ground. At New Albany, the TEOM was on top of a high school building approximately 200 m from the site of the indoor monitoring. The monitor’s inlet was within 13 m of the ground. All the three TEOMs were in secure and limited-access locations. 2.2.4. Gravimetric measurements Gravimetric measurements were conducted at Ohio University’s Air Quality Laboratory. Filters were weighed in a temperature and humidity controlled microenvironment (environmentally controlled glove-box, PLAS-LABS). As an alternative to the traditional weighing-room, the glove-box provides an inexpensive, reliable, and convenient method for gravimetric measurements (Allen et al., 2001). The specified temperature range was 20–23 1C with a variability of no more than 72 1C over 24 h, and 30–35% relative humidity with a variability of no more than 75% over 24 h. Each filter was weighed in duplicate—both before and after sampling using a Sartorius analytical microbalance (MC5 UL), with a readability of 1 mg. When these two weights differed by more than 3 mg, the filter was reweighed until sequential weights agreed within the specified range. The average of the two closest weights was the final weight.

2.3. Data validation and analysis In this study, data analyses complied with the published North American Research Strategy for Tropospheric Ozone (NARSTO) Data Management Handbook (Christensen et al., 1999). The PM2.5 concentration was determined by dividing the mass with the sample volume. The continuous ambient 24-h PM2.5 concentrations were based on the 30-min average readings of the TEOMs that were compiled by the TEOM

ARTICLE IN PRESS K.C. Crist et al. / Environmental Research 106 (2008) 62–71 software. Filter-based PM2.5 concentrations were computed using the mass difference between the filter’s initial and final weight obtained from gravimetric measurements. Units for PM2.5 mass and PM2.5 concentrations are reported in microgram (mg) and microgram per cubic meter (mg/m3), respectively. Field blanks, consisting of approximately 10% of the total samples, were used to evaluate the potential contamination during the process of filter assembly, disassembly, and transport. Sample weights from 1999 and 2000 were adjusted by subtracting the mean field blank of 1999 and of 2000, respectively. Limits of detections (LODs) were defined as the ratio of three times the standard deviation of the field blanks to the sampling volume. The LODs for the personal, indoor, and outdoor measurements were 9.01, 6.31, and 1.41 mg/m3, respectively. Data points less than the LODs were included in the data analysis. Negative data points and data points due to operation and equipment failures (i.e. pump malfunction, tube disconnection, and short sampling time) were not included in the analysis. Data capture or completeness, calculated as the number of collected daily samples divided by the target number of samples, were 82%, 88%, and 86% for the personal, indoor, and outdoor measurements, respectively. The data set was analyzed for the descriptive statistics. Temporal analysis (i.e. days of the week, seasons of the year) of the variations of PM2.5 concentrations was also conducted. Linear regression was performed on the filter-based PM2.5 measurements to determine the relationships between personal, indoor, and outdoor concentrations. For pair wise correlation, the 24-h average ambient filter-based measurements were adjusted to match the 7-h indoor and personal samples by using a weighting factor calculated from the TEOM measurements. To estimate the potential location of sources that contributed to the ambient PM2.5 concentrations, conditional probability function (CPF) (Ashbaugh et al., 1985; Kim et al., 2003) was applied to the TEOM measurements. Mathematically, CPF can be expressed as CPFDy ¼ mDy =nDy , where mDy is the number of times air parcels from the wind sector Dy exceeded a threshold criterion and nDy is the total number of data points from the wind sector Dy. Sources are more likely to be located in the directions which have a high CPF value (Kim and Hopke, 2004). Here, we computed the probability that PM2.5 concentrations coming from a given wind direction (Dy ¼ 201) exceeded its average value for each study site.

3. Results and discussion 3.1. Outdoor, indoor, and personal PM2.5 Table 2 summarizes the results of filter-based measurements. For all three sites, the mean indoor and personal PM2.5 concentrations were consistently higher than the outdoor concentrations. When the data were grouped by days of the week (Fig. 2), the trend of higher indoor and personal PM2.5 concentrations compared to the outdoor levels were still apparent, particularly for Athens. As shown in Fig. 3, outdoor, indoor, and personal PM2.5 concentrations tended to have higher values in spring and summer than in fall and winter for all three sites. During fall and winter, both personal and indoor PM2.5 concentrations were similar to the outdoor levels; while in spring, the average personal and indoor PM2.5 concentrations were approximately three to four times that of the outdoor levels. This may be largely attributed to the increased activities of students during the warm weather, and partly due to the fact that the windows were left open more often during warm season when the outdoor PM2.5 levels were high. The mean indoor PM2.5 concentrations in summer fell to the levels that were similar to the outdoor PM2.5 in

65

Table 2 Summary of personal, indoor, and outdoor PM2.5 measurements (mg/m3) Athens

Koebel

New Albany

Outdoor PM2.5 Mean SD Max Min Median N

13.66 8.91 61.12 0.50 11.49 332

13.89 9.29 77.01 0.24 11.65 315

12.72 8.86 69.30 0.05 10.87 327

Indoor PM2.5 Mean SD Max Min Median N

17.20 13.56 71.57 0.45 12.28 298

14.98 12.30 68.37 1.05 10.55 251

16.52 13.53 69.51 0.24 11.56 270

Personal PM2.5 Three samples per day per school were obtained Mean 17.61 14.59 SD 17.81 13.05 Max 88.38 66.96 Min 0.17 0.42 Median 9.53 10.18 N 207 194

13.93 12.25 56.90 0.95 9.45 205

Note: N is the number of samples obtained at a site for the entire study period.

Athens and New Albany, where no school was in session. Whereas Koebel, which had the summer school, still experienced higher indoor than outdoor PM2.5 concentrations. Koebel School relies entirely on ventilation for cooling purposes in summer (since it has no air conditioning) and summer had the highest outdoor PM2.5 concentration. This may be another reason for the elevated concentrations of indoor concentration values observed at Koebel in summer as compared to the other two schools. The dissimilarity of temporal patterns between outdoor, indoor, and personal concentrations indicates potential differences in their sources. Outdoor PM2.5 differs from non-ambient PM2.5 because it is mainly influenced by regional-scale pollution emissions and meteorological conditions (Wilson and Burton, 1995). As shown in Fig. 4 (CPF plots), the predominant wind directions associated with higher than average PM2.5 level in Athens is likely from the northeast, southeast, and southwest. This region is consistent with the location of coal-fired power plants along the Ohio River Valley. Because Athens is devoid of major anthropogenic sources, the high ambient PM2.5 can be mainly attributed to the regional transported particulate pollution from the Ohio River Valley. Similar relationships between wind direction and PM2.5 concentrations were seen at Koebel, with the most likely source region found to be located to the southeast of the site. The overall higher ambient PM2.5 concentration seen at Koebel as compared to the other schools is likely due to the fact that Koebel Elementary School is located in the midst of an urban area. It is therefore influenced by urban sources of pollution from

ARTICLE IN PRESS 66

K.C. Crist et al. / Environmental Research 106 (2008) 62–71

Fig. 2. Week day trends of PM2.5 concentration at Athens, Koebel, and New Albany. The data distribution has been shown with 95th, 75th, 25th, and 5th percentile and the mean values. Number of measurements (N) is indicated at the bottom of each boxplot.

all directions. New Albany, the suburban site, is impacted by particulate pollution mainly from the west (including southwest and northwest) which engulfs the urban regions of the Franklin County. PM2.5 concentration in New Albany was also affected by the southeast wind direction, indicating impacts from long-range transport of pollutants from the Ohio River Valley region. The consistent similarity found in the ambient PM2.5 concentration levels among all the three sites throughout the week days and across seasons indicates the existence of a common regional source influence. In addition, the Koebel and New Albany sites were also impacted by particulate pollution from the urban and industrialized city of Columbus. A source apportionment study conducted at these three schools featured PSCF results that were in agreement with these CPF results (John et al., 2007).

Fig. 3. Seasonal trends of PM2.5 concentration at Athens, Koebel, and New Albany. The data distribution has been shown with 95th, 75th, 25th, and 5th percentile and the mean values. Number of measurements (N) is indicated at the bottom of each boxplot.

In comparison, larger variations in indoor and personal PM2.5 (non-ambient PM2.5) concentrations were observed among the sites and on different temporal scales. This indicates that the non-ambient PM2.5 is highly subjective to the individual’s activity patterns and the ventilation conditions of where the individual spends time. Several reports have documented human activity as being responsible for high indoor and personal PM concentrations when no apparent indoor source exists (Long et al., 2000;

ARTICLE IN PRESS K.C. Crist et al. / Environmental Research 106 (2008) 62–71

67

Fig. 4. Conditional probability function plots for PM2.5 concentration at (a) Athens, (b) Koebel, and (c) New Albany.

Wallace, 2000; Vette et al., 2001). Long et al. (2000) reported an increase in PM2.5 concentration of approximately 23723 and 1279.1 mg/m3 associated with 11 dusting and 15 vigorous walking events, respectively. In studying the indoor and outdoor concentration ratios (I/O ratios) of PM (size range ¼ 0.01–2.5 mm) at Fresno, CA, researchers found re-suspension from activities during the day responsible for the observed diurnally I/O ratio variations for PM with size range 41 mm (Vette et al., 2001). This study was based in a school to minimize the impact of indoor PM2.5 sources including tobacco smoke and cooking activities which would otherwise be present in any other indoor setting. Consequently, the elevated indoor and personal PM2.5 concentrations appear to be mainly related to human activity patterns. As the subjects of this study were active and spent most of their time in large groups, their activity patterns would influence resuspensions and re-entrainment, resulting in elevated personal and indoor concentrations. To further investigate the influence of human activities on the measured PM2.5 concentrations, I/O ratios were evaluated for school days and non-school days. As shown in Table 3, higher values and ranges of I/O ratios were found during the school days than non-school days. The time series plot in Fig. 5 also shows a decrease of I/O ratios at all three sites when students were on vacation (starting from early June 2000). Overall, the I/O ratios of non-school days dropped 69%, 72%, and 26% compared to school days at Athens, New Albany, and Koebel, respectively. The less dramatic decline of indoor concentrations and I/O ratios associated with the absence of students at Koebel is likely contributed by the activities of the summer school students in the same classroom at Koebel. Despite the fact that the extent of the I/O ratio decrease varied among the sites, the general decline trend of I/O ratios when students were not present in the buildings suggests that the measured indoor and personal PM2.5 concentrations are highly affected by human activity. Similar results were reported by other studies. In nine homes in the Boston area, (Long et al., 2000) reported the mean I/O ratios for

Table 3 PM2.5 mass concentrations (mg/m3) indoor/outdoor ratios (school days vs. non-school days at all sites) Mean

SD

Max

Min

N

Athens School-day Non-school-day

2.61 0.8

5.76 0.7

57.43 3.34

0.17 0.1

235 31

Koebel School-day Non-school-day

1.71 1.27

3.17 1.16

31.69 4.94

0.16 0.04

196 43

New Albany School-day Non-school-day

2.98 0.82

5.47 0.6

49.44 2.96

0.05 0.13

208 54

PM2.5 as 2.4714 for daytime while 0.7470.41 for nighttime. In a fifth-grade classroom in Lindon, Utah, researchers reported prominent indoor PM2.5 concentration peaks occurred from 9:00 a.m. to 4:00 p.m. each school day; while no midday peaks existed during the weekend (Patterson and Eatough, 2000). Studies (e.g. Wallace, 1996) also found that I/O ratios are typically less than or equal to 1 in the absence of indoor sources. In this study, the mean I/O ratios ranged from 1.71 to 2.98 during school days, while they ranged from 0.80 to 1.27 during non-school days. 3.2. Correlations between personal, indoor, and outdoor PM2.5 Results from the regression analysis for personal, indoor, and outdoor PM2.5 concentrations are presented in Table 4 and Fig. 6. While both Athens and Koebel had moderate personal–outdoor PM2.5 correlations, no statistically significant correlations were found at New Albany. The indoor–outdoor PM2.5 correlations were similar among the three sites. Statistically significant (po0.0001) correlations existed between personal and indoor PM2.5 concentrations at all three sites with R values ranging from 0.32 to 0.87.

ARTICLE IN PRESS K.C. Crist et al. / Environmental Research 106 (2008) 62–71

68

Fig. 5. I/O ratios: school-day vs. non-school-day (2000 data). I/O ratios greater than 5 were not included in the graph. Table 4 Summary of correlations of personal, indoor, and outdoor PM2.5 concentrations (mg/m3) Slope

R

N

p

Personal vs. outdoor Athens 10.22 Koebel 8.42 New Albany 13.37

0.58 0.50 0.05

0.27 0.32 0.04

202 186 200

0.0001 o0.0001 0.59

Personal vs. indoor Athens 2.03 Koebel 4.96 New Albany 8.19

0.87 0.70 0.32

0.68 0.56 0.30

196 169 182

o0.0001 o0.0001 o0.0001

Indoor vs. outdoor Athens 14.28 Koebel 10.43 New Albany 10.77

0.19 0.36 0.44

0.13 0.26 0.30

292 240 265

0.03 o0.0001 o0.0001

Intercept

Note: Data for personal–outdoor and personal–indoor correlation analyses are based on school day data. Indoor–outdoor correlation analysis includes both school day and non-school day data. Excluding data for non-school day yields indoor–outdoor correlations at three sites with R of 0.13, 0.32, and 0.36 for Athens, Koebel, and New Albany, respectively.

Generally, the indoor–personal correlations were stronger than correlations between personal and outdoor PM2.5 concentrations and between indoor and outdoor concentrations. Similar results were reported in Pellizzari et al. (1999). In that study, pooled data for all subjects resulted in low personal–outdoor correlations (R ¼ 0.19–0.27, po0.01), moderate indoor–outdoor correlation (R ¼ 0.21–0.33), and strong personal–indoor concentrations (R ¼ 0.79, po0.01). The results of poor personal–outdoor correlations and strong personal–indoor correlations from this study and other related studies (Lai et al., 2004; Mohammadyan and Ashmore, 2005) indicate that personal PM2.5 exposure is more closely affected by indoor PM2.5 levels than the ambient PM2.5 correlations. Other factors that influence the personal–indoor–outdoor correlations of PM2.5 include study designs. In the pre-pilot PTEAM study, the cross-sectional personal–out-

door correlations (R2 ¼ 0–0.02) were lower than the longitudinal regressions in which individual correlation was performed (R2 ¼ 0.01–0.58) (Wallace, 1996). However, regardless of the study design (longitudinal vs. crosssectional), there is a tendency that good personal–outdoor PM correlations are likely to occur on subjects with minimum indoor sources. In a longitudinal-designed research, Williams et al. (2000) reported the R-value of daily average personal–outdoor correlation of 0.89 on elderly subjects (72–93 years old), who usually have low incidence of indoor sources and low activity levels. Researches found that pooled PM10 personal–outdoor correlations (cross-sectional study) were significant with R of 0.83 for seven elderly adults, who were not exposed to apparent indoor sources (Tamura et al., 1996). The data analysis also excluded days with high indoor source events (i.e. incense burning, smoking visitors). In general, this study was longitudinal because measurements over multiple days for the subjects were obtained. For daily personal monitoring, one subject was chosen from the class. Although each subject tended to have similar activity patterns with the whole group, different personal exposure to non-ambient PM sources existed due to discrepancies in individual’s activity pattern. Because of the large number of subjects involved and the length of the study, no activity diaries were maintained. This may limit the correlation analysis in differentiating the subjects from limited nonambient PM2.5 to those with high non-ambient PM2.5 exposures, and result in poor to moderate personal–outdoor correlations. Measurement method can also influence correlations between personal, indoor, and outdoor PM2.5 concentrations. In this study, cautions were taken to minimize measurement biases through procedures such as using the same inlet type for outdoor and indoor monitors, and conducting concurrent measurement at each site. However, differences existed between monitoring devices used for personal, indoor, and outdoor measurement. More fibers or visible particles resided on personal sample filters than indoor and outdoor sample filters. It is possible that the

ARTICLE IN PRESS K.C. Crist et al. / Environmental Research 106 (2008) 62–71

69

Fig. 6. Correlations of personal–outdoor, indoor–outdoor, and personal–indoor during the entire sampling period at Athens, Koebel, and New Albany.

impactors, used in personal monitoring, collected some coarse particles due to low flow rate or due to design capability. Another common effect is the existence of ‘‘personal cloud’’ (Wu et al., 2005; Corsi et al., 2007). Personal cloud or personal-activity PM2.5 is an increased personal exposure derived from personal measurements in comparison with a stationary monitors or a time-weighted average of indoor and outdoor concentrations. The cause of personal cloud is still unknown, however, since the chemical composition of personal cloud is similar to the indoor particles, personal cloud is thought to be associated with indoor activity-generated sources. Studies show personal cloud as a major reason for the poor personal–outdoor correlations since the personal cloud for healthy persons could count for up to 50 mg/m3 during their active

period (Howard-Reed et al., 2000). In comparing retirement centers to apartments, Rodes suggested a personal cloud might be a result of ‘‘re-suspended particles from carpeting, collection of body dander and clothing fibers, personal proximity to open doors and windows, and elevated PM levels in non-apartment indoor microenvironments’’ (Rodes et al., 2001). Electrostatic effects which happen within a close vicinity of a subject to the sources may also contribute to the cloud. Although resuspension effects are more pronounced in coarse-model particles, increased PM2.5 concentration in re-suspension events (i.e. physical activity) were also recorded (Thatcher and Layton, 1995; Sarnat et al., 2000; Chang et al., 2000). The re-suspension-related PM2.5 increase in this study could be quite sizable due to large group (30 subjects) involved.

ARTICLE IN PRESS 70

K.C. Crist et al. / Environmental Research 106 (2008) 62–71

4. Conclusions This study was designed to study personal exposure, indoor, and outdoor levels of PM2.5 mass concentrations of fourth and fifth grade elementary school children in Ohio. To that end, personal, indoor, and outdoor PM2.5 mass measurements were conducted at three elementary schools in central and southeastern Ohio from January 1999 through August 2000. At all three sites, personal PM2.5 exposures were significantly affected by indoor PM2.5, presumably the result of re-suspension by human activity. This conclusion was supported by the consistently higher personal and indoor PM2.5 concentrations as compared to outdoor levels seen at each site throughout the school days. Furthermore, the I/O ratios of PM2.5 were greater than unity at all sites when school was in session with large fluctuations; while lower and steady I/O ratios associated with lower indoor sources were found during non-school days when the students were absent. Due to the strong influence of indoor PM2.5 source influence (i.e. student activity patterns, building ventilation condition), only moderate personal–outdoor correlations existed at two sites, Koebel and Athens. No personal–outdoor correlations occurred at New Albany. In comparison, there were strong personal–indoor correlations at all three sites. The findings of this study support the conclusions of previous research, in which the influence of non-ambient PM2.5 contributes to the poor to moderate personal–outdoor relationships. One of the central assumptions of some epidemiological studies is that ambient PM concentration, monitored by a fixed site, is a sufficient surrogate for personal PM exposure. The highly variable correlations between personal and outdoor concentrations in this study indicate that the ambient PM2.5 may not be a strong indicator for indoor concentration or total personal exposures. However, the personal–indoor–outdoor correlations in this study were established in three groups of 30 primary school students who tended to have group activities in the classroom environment. The question arises whether the results can be generalized to children in other microenvironments and to other types of population groups. Further studies that include personal activity diaries and chemical composition of the measured PM2.5 are clearly needed in order to provide detailed characterization of the personal exposure level of PM2.5. Transport characteristics of PM2.5 were studied at all the three sites using CPF analysis and the Athens site experienced an inflow of fine particulate matter mainly from the south and southeast direction which coincides with the location of several coal-fired power plants along the Ohio River Valley. This PM2.5 was most likely of a regional origin as Athens being a rural site is lacking in major anthropogenic sources. The Koebel site being an urban site is influenced by particulate pollution from all directions whereas the suburban site of New Albany is impacted by particulate pollution from the west (i.e. the

urban areas in Franklin County). The CPF analysis will serve as a basis for source apportionment studies in the future as well as provide a direction to evaluate the meteorological influences governing outdoor PM2.5 concentrations. Acknowledgments This study was supported by the Ohio Environmental Protection Agency and the Ohio Air Quality Development Authority. The authors would like to acknowledge Michele Morrone, Ohio University, for her technical guidance on this project. References Allen, R., Box, M., Liu, L.S., Larson, T.V., 2001. A cost-effective weighing chamber for particulate matter filters. J. Air Waste Manage. Assoc. 51, 1650–1653. Allen, R., Larson, T., Wallace, L., Sheppard, L., Liu, L.-J.S., 2003. Investigation of indoor and outdoor contributions to total indoor particulate matter exposure. Environ. Sci. Technol. 37, 3484–3492. Ambient Monitoring, the Upper Ohio River Valley Project (UORVP) [online] /www.netl.doe.gov/coalpower/environment/air_q/am_monitor. htmlS, 28 February 2002. Ashbaugh, L.L., Malm, W.C., Sadeh, W.D., 1985. A residence time probability analysis of sulfur concentrations at Grand Canyon National Park. Atmos. Environ. 19, 1263–1270. Chang, L., Koutrakis, P., Catalano, P.J., Suh, H.H., 2000. Hourly personal exposure to fine particles and gaseous pollutants—results from Baltimore, Maryland. J. Air Waste Manage. Assoc. 50, 1223–1235. Chow, J.C., Watson, J.G., 1998. Guideline on Speciated Particulate Monitoring. Desert Research Institute, Prepared for Neil Frank, Jim Homolya, Office of Air Quality Planning and Standards (MD-14), NC. US Environmental Protection Agency, Research Triangle Park. Chow, J.C., Watson, J.G., Fujita, E.M., Lu, Z.Q., Lawson, D.R., Ashbaugh, L.L., 1994. Temporal and spatial variations of PM2.5 and PM10 aerosol in the Southern California air-quality study. Atmos. Environ. 28, 2061–2080. Christensen, S.W., Boden, T.A., Hook, L.A., Cheng, M.D., 1999. NARSTO Quality Systems Science Center Oak Ridge National Laboratory. Environmental Sciences Division Publication No. 4917, Oak Ridge, TN. Revision date: August 16, 1999. Corsi, R.L., Siegel, J., Karamalegos, A., Simon, H., Morrison, G.C., 2007. Personal reactive clouds: introducing the concept of near-head chemistry. Atmos. Environ. 41, 3161–3165. Dockery, D.W., Pope, C.A., Xu, X.P., Spengler, J.D., Ware, J.H., Fay, M.E., Ferris, B.G., Speizer, F.E., 1993. An association between air-pollution and mortality in 6 United-States cities. New Eng. J. Med. 329, 1753–1759. Federal Register, 1997. 63, 38702–38752. Goswami, E., Larson, T., Lumley, T., Liu, L.-J.S., 2002. Spatial characteristics of fine particulate matter: identifying representative monitoring locations in Seattle. J. Air Waste Manage. Assoc. 52, 324–333. Howard-Reed, C., Rea, A.W., Zufall, M.J., Burke, J.M., Williams, R.W., Suggs, J.C., Sheldon, L.S., Walsh, D., Kwok, R., 2000. Use of continuous neophelometer to measure personal exposure to particles during the US environmental protection agency Baltimore and Fresno Panel Studies. J. Air Waste Manage. Assoc. 50, 1125–1132. Janssen, N.A., Hoek, G., Harssema, H., Brunekreef, B., 1999. Personal exposure to fine particles in children correlations closely with ambient fine particles. Arch. Environ. Health 54, 95–101.

ARTICLE IN PRESS K.C. Crist et al. / Environmental Research 106 (2008) 62–71 John, K., Karnae, S., Crist, K., Kim, M., Kulkarni, A., 2007. Analysis of trace elements and ions in ambient fine particulate matter at three elementary schools in Ohio. J. Air Waste Manage. Assoc. 57, 394–406. Kim, E., Hopke, P.K., 2004. Source apportionment of fine particles in Washington, DC, utilizing temperature-resolved carbon fractions. J. Air Waste Manage. Assoc. 54, 773–785. Kim, E., Hopke, P.K., Edgerton, E.S., 2003. Source identification of Atlanta aerosol by positive matrix factorization. J. Air Waste Manage. Assoc. 53, 731–739. Klemm, R.J., Mason Jr., R.M., Heilig, C.M., Neas, L.M., Dockery, D.W., 2000. Is daily mortality associated specifically with fine particles? Data reconstruction and replication of analyses. J. Air Waste Manage. Assoc. 50, 1215–1222. Lai, H.K., Kendall, M., Ferrier, H., Lindup, I., Alm, S., Ha¨nninen, O., Jantunen, M., Mathys, P., Colvile, R., Ashmore, M.R., Cullinan, P., Nieuwenhuiksen, M.J., 2004. Atmos. Environ. 38, 6399–6410. Liao, D.P., Creason, J., Shy, C., Williams, R., Watts, R., Zweidinger, R., 1999. Daily variation of particulate air pollution and poor cardiac autonomic control in the elderly. Environ. Health Persp. 107, 521–525. Long, C.M., Suh, H.H., Koutrakis, P., 2000. Characterization of indoor particle sources using continuous mass and size monitors. J. Air Waste Manage. Assoc. 50, 1236–1250. Meyer, M.B., Patashnick, H., Ambs, J.L., Rupprecht, E., 2000. Development of a sample equilibration system for the TEOM continuous PM monitor. J. Air Waste Manage. Assoc. 50, 1345–1349. Mohammadyan, M., Ashmore, M.R., 2005. Personal exposure and indoor PM2.5 concentrations in an urban population. Indoor Built Environ. 14, 313–320. Patterson, E., Eatough, D.J., 2000. Indoor/outdoor relationships for ambient PM2.5 and associated pollutants: epidemiological implications in Lindon, Utah. J. Air Waste Manage. Assoc. 50, 103–110. Pellizzari, E.D., Clayton, C.A., Rodes, C.E., Mason, R.E., Piper, L.L., Fort, B., Pfeifer, G., Lynam, D., 1999. Particulate matter and manganese exposures in Toronto, Canada. Atmos. Environ. 33, 721–734. Pope, C.A., Dockery, D.W., 2006. Health effects of fine particulate air pollution: lines that connect. J. Air Waste Manage. Assoc. 56 (6), 709–742. Pope, C.A., Bates, D.V., Raizenne, M.E., 1995. Health-effects of particulate air-pollution-time for reassessment. Environ. Health Persp. 103, 472–480. Rodes, C.E., Lawless, P.A., Evans, G.F., Sheldon, L.S., Williams, R.W., Vette, A.F., Creason, J.P., Walsh, D., 2001. The relationships between personal PM exposures for elderly populations and indoor and outdoor concentrations for three retirement center scenarios. J. Exp. Anal. Environ. Epidemiol. 11, 103–115. Sarnat, J.A., Koutrakis, P., Suh, H.H., 2000. Assessing the relationship between personal particulate and gaseous exposures of senior citizens living in Baltimore, MD. J. Air Waste Manage. Assoc. 50, 1184–1198. Schlesinger, R.B., Kunzli, N., Hidy, G.M., Gotschi, T., Jerrett, M., 2006. The health relevance of ambient particulate matter characteristics: coherence of toxicological and epidemiological inferences. Inhal. Toxicol. 18 (2), 95–125. Schwartz, J., Dockery, D.W., Neas, L.M., 1996. Is daily mortality associated specifically with fine particles? J. Air Waste Manage. Assoc. 46, 927–939.

71

Spengler, J.D., Koutrakis, P., Dockery, D.W., Raizenne, M., Speizer, F.E., 1996. Health effects of acid aerosols on North American children: air pollution exposures. Environ. Health Persp. 104, 492–499. Tamura, K., Ando, M., Sagai, M., Matsumoto, Y., 1996. Estimation of levels of personal exposure to suspended particulate matter and nitrogen dioxide in Tokyo. Environ. Sci. Technol. (Tokyo) 4, 37–51. Thatcher, T.L., Layton, D.W., 1995. Deposition, re-suspension and penetration of particles within a residence. Atmos. Environ. 29, 1487–1497. US Environmental Protection Agency, 1996. Air Quality Criteria for Particulate Matter Chapter 1, Executive Summary, EPA/600/P-95/ 001aF. Office of Research and Development, Washington, DC. US Environmental Protection Agency, 2001. Air Quality Criteria for Particulate Matter, vol. 1, EPA 600/P-99/002a. Office of Research and Development, Washington, DC. Vette, A.F., Rea, A.W., Lawless, P.A., Rode, C.E., Evans, G., Highsmith, V.R., Sheldon, L., 2001. Characterization of indoor–outdoor aerosol concentration relationships during the Fresno PM exposure studies. Aerosol Sci. Tech. 34, 118–126. Wallace, L., 1996. Indoor particles: a review. J. Air Waste Manage. Assoc. 46, 98–126. Wallace, L., 2000. Correlations of personal exposure to particles with outdoor air measurement: a review of recent studies. Aerosol Sci. Tech. 32, 15–25. Watson, J.G., Wyzga, R.E., Burnett, R.T., Vostal, J.J., Romieu, I., Chow, J.C., Holcombe, J.T., Lipfert, F.W., Marrack, D., Thompson, R.J., Vedal, S., 1997. Ambient particles and health: lines that divide-discussion. J. Air Waste Manage. Assoc. 47, 995–1008. WHO, 2005a. Effects of Air Pollution on Children’s Health and Development. World Health Organization, Regional Office for Europe, Bonn, Germany. WHO, 2005b. WHO Air Quality Guidelines Global Update 2005. World Health Organization, Regional Office for Europe, Bonn, Germany; Copenhagen, Denmark. WHO, 2006. WHO Air Quality Guidelines for Particulate Matter, Ozone, Nitrogen Dioxide and Sulfur Dioxide. Geneva, Switzerland. Williams, R., Suggs, J., Creason, J., Rodes, C., Lawless, P., Kwok, R., Zweidinger, R., Sheldon, L., 2000. The 1998 Baltimore particulate matter epidemiology-exposure study: Part 2. Personal exposure assessment associated with an elderly study population. J. Exp. Anal. Environ. Epidemiol. 10, 533–543. Wilson, W.E., Burton, R.M., 1995. Exposure relationships relevant to particulate matter mortality studies. In: Paper Presented at A&WMA International Conference on Particulate Matter: Health and Regulatory Issues, Pittsburgh PA, April 4–6, 1995. Wilson, W.E., Suh, H.H., 1997. Fine particles and coarse particles: concentration relationships relevant to epidemiologic studies. J. Air Waste Manage. Assoc. 47, 1238–1249. Wilson, W.E., Mage, D.T., Grant, L.D., 2000. Estimating separately personal exposure to ambient and nonambient particulate matter for epidemiology and risk assessment: why and how. J. Air Waste Manage. Assoc. 50, 1167–1183. Wu, C.-F., Delfino, R.J., Floro, J.N., Quintana, P.J.E., Samimi, B.S., Kleinman, M.T., Allen, R.W., Liu, L.-J.S., 2005. Exposure assessment and modeling of particulate matter for asthamatic children using personal nephelometers. Atmos. Environ. 39, 3457–3469.