Metro Commuter Exposures to Particulate Air ... - ACS Publications

Article pubs.acs.org/est

Metro Commuter Exposures to Particulate Air Pollution and PM2.5Associated Elements in Three Canadian Cities: The Urban Transportation Exposure Study Keith Van Ryswyk,*,† Angelos T. Anastasopolos,† Greg Evans,‡ Liu Sun,† Kelly Sabaliauskas,‡ Ryan Kulka,† Lance Wallace,§ and Scott Weichenthal†,∥ †

Air Health Science Division, Health Canada, Ottawa, K1A 0K9, Canada Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, M5S 3E5, Canada § Consultant, Santa Rosa, California 95409, United States ∥ Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, H3A 1A2, Canada ‡

S Supporting Information *

ABSTRACT: System-representative commuter air pollution exposure data were collected for the metro systems of Toronto, Montreal, and Vancouver, Canada. Pollutants measured included PM2.5 (PM = particulate matter), PM10, ultrafine particles, black carbon, and the elemental composition of PM2.5. Sampling over three weeks was conducted in summer and winter for each city and covered each system on a daily basis. Mixed-effect linear regression models were used to identify system features related to particulate exposures. Ambient levels of PM2.5 and its elemental components were compared to those of the metro in each city. A microenvironmental exposure model was used to estimate the contribution of a 70 min metro commute to daily mean exposure to PM2.5 elemental and mass concentrations. Time spent in the metro was estimated to contribute the majority of daily exposure to several metallic elements of PM2.5 and 21.2%, 11.3% and 11.5% of daily PM2.5 exposure in Toronto, Montreal, and Vancouver, respectively. Findings suggest that particle air pollutant levels in Canadian metros are substantially impacted by the systems themselves, are highly enriched in steel-based elements, and can contribute a large portion of PM2.5 and its elemental components to a metro commuter’s daily exposure.

■

1.09M, and Vancouver’s 0.39M daily ridership figures representing the third, fourth, and tenth highest in North America, respectively.13 The objectives of this study were to characterize exposures to particulate air pollutants in Canadian metro systems, evaluate system features as determinants of these exposures, compare metro PM2.5 mass and elemental concentrations to that of ambient air, and estimate the proportion of daily PM2.5 mass and elemental exposure that can originate from time spent in the metro environment. This study is the first to investigate PM and PM-associated elemental exposures in Canadian metro systems and, to our knowledge, the first to assess metro exposures system-wide.

INTRODUCTION

Few studies have examined air pollution exposure in Canadian transportation environments.1,2 However, existing evidence suggests that metro environments may be associated with elevated levels of air pollutants including fine (PM2.5, where PM = particulate matter) and coarse (PM2.5−10) particulate air pollution and PM-associated elements such as heavy metals.3−8 In consideration of this, the activity of commuting may represent a significant proportion of daily exposure to these pollutants for metro commuters.9−11 The Urban Transportation Exposure Study (UTES) was conducted from 2010 to 2013 to characterize air pollution exposures in major transit modes (private vehicles, metro systems, and buses) in Canadian metropolitan centers. The results of the private vehicle monitoring campaign were reported previously.12 This paper presents a comprehensive analysis of metro commuting exposures in Toronto, Montreal, and Vancouver which are all rail-based, predominantly gradeseparated (underground or elevated) systems designed to provide rapid transit across large geographic areas. Metro commuting in Canada is significant by North American standards, with Montreal’s 1.24M (M = millions), Toronto’s © XXXX American Chemical Society

■

MATERIALS AND METHODS Metro Systems and Sampling Routes. Particulate air pollution exposures were assessed in the Toronto, Ontario

Received: November 15, 2016 Revised: March 27, 2017 Accepted: April 17, 2017

A

DOI: 10.1021/acs.est.6b05775 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

periods of riding and waiting and to record station names. Sampling inlets for the air pollution monitors were connected to shoulder straps to sample the technicians’ breathing zones. All continuous instruments were factory calibrated before each seasonal sampling campaign. Ultrafine particle concentrations over 100 000 pts/cm3 were corrected for bias.14 Further detail on continuous data quality assurance, control, and management are provided in the Supporting Information. Integrated Air Pollution Sampling. Technicians collected integrated samples of PM2.5 and PM10 mass concentrations using the PM2.5 and PM10 personal environmental monitors (PEMs) of the Chempass system (Chempass System R&P/Thermo).15 A personal sampling pump (SKC Leland) was connected to two PEMs (one PM2.5 and the other PM10) providing each with 4.0 L per minute of air flow. Integrated samples reflected 1 week periods (ten 3 h sessions, 30 h total) to ensure detectable levels of PM-associated elements. This resulted in the collection of 18 samples per city for each size fraction. Filters were analyzed gravimetrically following EPA guidelines. Inductively coupled plasma−mass spectrometry (ICP-MS) was used to determine elemental concentrations of PM. Further detail on integrated sampling is provided in the Supporting Information. Relating Metro Exposures to System Features. Metro levels of PM2.5 and UFP were analyzed as a function of system features while controlling for meteorology. Factors of depth, elevation, percent coverage, distance to outside, and section age were selected for evaluation as determinants of exposure on account of their potential relationship with metro ventilation. “Depth” represented the below-grade distance from the ground surface to the rail segment or platform. Independent of “depth”, “elevation” represented the ground level height above sea level. “Percent coverage” was defined as the proportion of a platform or rail segment covered or below grade, while “distance to outside” represented the distance to the closest above-grade section of the system. The section age was calculated for each station and rail segment and is based on the year of construction and the time of sampling. Further detail on the nature and derivation of system feature information is provided in the Supporting Information. Analyses relating system features to measured exposures were conducted while controlling for the effect of meteorology and ambient PM2.5 levels. Meteorological data including temperature, relative humidity, wind speed, wind direction, and atmospheric pressure were obtained from monitoring stations operated by Environment and Climate Change Canada (ECCC). Hourly central site ambient PM2.5 data were downloaded from ECCC’s National Air Pollution Surveillance (NAPS) network Web site. For each city, continuous data concurrent with metro sampling were downloaded from NAPS sites employing tapered element oscillating microbalance monitors (TEOMs) or synchronized hybrid ambient real-time particulate monitors (SHARPs). Data from multiple stations (4 in Toronto, 14 in Montreal, 17 in Vancouver) were downloaded and averaged on an hourly basis to one regional value. Multivariate mixed-effects linear regression models were used to examine the effect of system features on the levels of PM2.5 and UFPs, while controlling for meteorology. Regional PM2.5 levels were also controlled for in PM2.5 models. Separate models were examined for data representing below-grade sections and above-grade sections (Toronto and Vancouver only, as the Montreal metro is completely below grade). Estimates of the relationship between subway features and

subway, the Montreal, Quebec Metro, and the SkyTrain of Vancouver, British Columbia. These systems represent the three largest rail-based commuter systems in Canada. They are similar in length at approximately 68 km but vary in other design features (summarized in Table 1). The main system Table 1. Canadian Metro System Studies in the Urban Transportation Exposure Study system name

subway

metro

SkyTrain

city province opened no. lines no. stations % underground/ indoors system length (km) daily ridership13 track gauge propulsion

Toronto Ontario 1954 4 69 80

Montreal Quebec 1966 4 68 100

Vancouver British Columbia 1985 4 47 20

68.3

69.2

68.6

1.09M 1495 mm third rail 600 V DC steel

1.24M 1435 mm third rail 750 V DC rubber

0.39M 1435 mm third rail 600 and 750 V DC steel

wheels

design differences between the metro systems of these cities deal with their wheel design and the proportion of each system which is below-grade. The Toronto and Vancouver metros use steel wheels rolling on steel track, while the Montreal metro features rubber wheels on concrete rollways. The Vancouver metro is largely at-grade and elevated outdoors (10 of 47 stations are below-grade), while the Toronto metro is predominantly below-grade (55 of 69 stations are belowgrade), and the Montreal metro is entirely below-grade. Sampling was conducted in each metro system for three consecutive weeks in the summer and winter seasons (Toronto and Montreal: summer 2010 and winter 2011; Vancouver: winter and summer 2013). In each city, three technicians carried personal sampling backpacks (Figure S1) and were assigned to one of three separate segments of the system. Each segment was approximately equal in their number of stations and length. Together, they constituted all stations and rail segments of the system with the exception of the Montreal metro where the “yellow line” (consisting of an additional two stations) was excluded for logistical reasons. Technicians sampled on weekday mornings (7−10 AM) and evenings (3−6 PM) and followed a pattern of boarding and disembarking at each station in their segment. Upon encountering the end of a line, they switched directions. This could be done inside a station and required no interruption of data collection. Each 3 h sampling session began at the station at which sampling ended in the preceding 3 h sampling session. Personal exposures measured while the technicians were waiting on the platform are referred to as “waiting” exposures, while those measured on the train are referred to as “riding” exposures. Waiting exposures were sampled at the middle of each platform, and riding exposures were sampled at the middle of each train. Continuous Air Pollution Monitoring. Continuous monitoring was carried out for PM2.5 (TSI DustTrak 8520), ultrafine particles (UFP) (CPC, TSI model 3007), and black carbon (BC) (AethLabs MicroAeth-51). Geographic location data were collected using a global position recorder (GlobalSat DG-100). Digital voice recordings (DVR) were used to identify B


Article

Environmental Science & Technology exposure levels were calculated using the “proc mixed” procedure in SAS EG 5.1 (SAS Institute Inc., Cary, NC, USA). PM2.5 and UFP exposures were modeled with a generalized linear mixed model with an autoregressive covariance structure (AR1) to account for the lack of independence between repeated weekly measurements as well as clustering by each rail segment/station and season (eq 1). Yij = α + βXij + bi + εij

Contribution of Metro Commute to Daily PM2.5 Exposures. Microenvironmental models were developed to estimate the contribution of metro commuting to overall daily exposures of PM2.5 mass and elemental concentrations. The elements included in this analysis were all those which (i) were considerably higher in metro versus ambient, (ii) are known to be related to metro steel and brake composition, and (iii) were measured to be above detection for >50% of samples in both the metro and ambient environments. A key factor in these estimates is the amount of time spent in the metro. Measures of daily time spent in rail-based transportation, including train, subway, or rapid transit, were quantified in the Canadian Human Activity Pattern Survey 2 (CHAPS2).18 Based on this survey, collected from a representative sample of people living in Edmonton, Montreal, Vancouver, Toronto, and Halifax, an average of 70 min per day was determined for those that use the metro for transportation (Dr. Carlyn J. Matz, personal communication). The formula for calculating the contribution of metro commuting to overall daily PM2.5-elemental components is presented in eq 3. GmetroPM2.5jk represents the estimated contribution of a typical 70 min metro commute to daily PM2.5 exposure for element j in city k, XmetroPM2.5jk represents the mean PM2.5 concentration of element j sampled in metro of city k, XAmbientPM2.5jk is the mean PM2.5 concentration of element j sampled in city k, and APEM:Dichot represents the correction factor for the 11% positive bias of the PEM, relative to the dichotomous sampler.17

(1)

Here, Yij is the natural logarithm for PM2.5 or UFP for station/ rail segment i, on week j, α is the regression intercept, β is the fixed effect of X on Y, Xij is the predictor variable for station/rail segment i on week j, bi is the random effect for station/rail segment, and εij is the random error. Effects of influential points were assessed by calculating Cook’s distances. Residual plots were assessed for normality using the Shapiro−Wilk statistic. Predictor parameter estimates and 95% confidence intervals were exponentiated to provide estimates of percent change in exposure per unit increase. Metro versus Ambient PM 2.5 . In each city, the concentration and elemental composition of PM2.5 between the metro and ambient environments were compared. As each city’s metro environment was represented by the 18 PM2.5 samples collected in the study, ambient levels were characterized by the concurrent gravimetric PM2.5 data collected by the ECCC NAPS network. The ECCC conducts the routine collection of 24 h gravimetric PM2.5 for several stations across Canada every 3 days. Ambient gravimetric PM2.5 mass and elemental concentration data was available for one NAPS station in each city. NAPS data collected during and within 2 weeks of each three-week metro sampling period was downloaded to represent ambient levels of PM2.5 and its elemental composition. The NAPS network employs the dichotomous air sampler (R&P/Thermo Scientific) for the collection of gravimetric PM2.5 data. The dichotomous air sampler is listed as an “equivalent method” by the US EPA.16 The PM2.5 PEM of the Chempass system has been shown to have excellent agreement with the PM2.5 measures of dichotomous sampler (R2 = 97%) with a median positive bias of 11%.17 For the purpose of comparing metro PM2.5 to that of ambient, all metro PM2.5 mass and elemental concentrations were adjusted for this 11% bias. Both metro and NAPS gravimetric PM2.5 samples were analyzed for elemental content using ICPMS. Abundance ratios for metro and ambient PM2.5 were calculated (eq 2). AR ij =

GmetroPM2.5jk =

In the calculation of the proportion of daily PM 2.5 attributable to time spent in the metro, metro PM2.5 was calculated as a weighted average between time spent on platforms (10 minbased on train frequencies during rush hour) and time spent in trains (60 min). PM2.5 exposures specific to these two microenvironments were taken from averaged continuous data collected by the DustTrak which was found to be in excellent agreement with the gravimetric method of this study (Figure S2). In eq 4, the formula for calculating the contribution of metro commuting to overall daily PM2.5 is presented where GmetroPM2.5j represents the estimated contribution of a typical 70 min metro commute to daily PM2.5 exposure for city j, XplatformPM2.5j represents the mean PM2.5 platform exposure for city j, XridingPM2.5j represents the mean PM2.5 riding exposure for city j, XAmbient PM2.5j is the mean ambient PM2.5 concentration for city j, and APEM:Dichot represents the correction factor for the 11% positive bias of the PEM, relative to the dichotomous sampler.17

(2)

ARij = abundance ratio of element j in PM2.5 sample i, Cij = concentration of element j in PM2.5 sample i, and CiPM2.5 = concentration of PM2.5 sample i. GmetroPM2.5j =

[X̅ metroPM2.5jk APEM:Dichot × 70 min] + [XAmbientPM2.5 ̅ jk × 1370 min]

(3)

Cij CiPM2.5

X̅ metroPM2.5jk APEM:Dichot × 70 min

[X̅ platformPM 2.5j APEM:Dichot × 10 min] + [X̅ ridingPM2.5j APEM:Dichot × 60 min] [X̅ platformPM2.5j APEM:Dichot × 10 min] + [X̅ ridingPM2.5j APEM:Dichot × 60 min] + [XAmbientPM2.5 ̅ j × 1370 min]

Preliminary data analyses included descriptive statistics, distribution plots, and interpollutant correlations (Pearson’s R, Spearman’s ρ). Statistical tests of difference were applied at the 5% significance level via the nonparametric Kruskal−Wallis test. Statistical analyses and graphics were carried out using SAS EG 5.1 (SAS Institute Inc., Cary, NC, USA). Pollutant maps

(4)

were created using ArcGIS 10 (ESRI 2011, ArcGIS Desktop: Release 10, Redlands, CA: Environmental Systems Research Institute). C


Article

Environmental Science & Technology Table 2. Gravimetric PM2.5 and PM10 Results by City (μg/m3) city Toronto Montreal Vancouver

parameter

n

mean (SD)

median (p25−p75)

range

PM2.5 PM10 PM2.5 PM10 PM2.5 PM10

18 18 15 18 18 18

100 (36) 304 (105) 36 (17) 97 (45) 17 (8) 56 (25)

95 (66−118) 272 (241−396) 35 (21−46) 104 (54−119) 19 (8−22) (35−79)

50−171 155−501 15−81 32−190 4−32 23−99

Table 3. PM2.5, UFP, and Black Carbon Levels on Studied Systems overall descriptive statistics pollutant PM2.5 (μg/m3)

ultrafine particles (103 pts/cm3)

black carbon (μg/m3)

city

n

mean (SD)

median (p25−p75)

p5−p95

Toronto Montreal Vancouver Toronto Montreal Vancouver Toronto Montreal Vancouver

718 795 564 668 775 564 591 795 564

110.8 (58.9) 37.8 (14.1) 23.7 (18) 10.5 (6.1) 19.5 (10.5) 12.6 (12) 7.8 (7.2) 4.6 (2.3) 2.8 (2.4)

97.2 (66.6−150.6) 35.6 (25.8−47.4) 19.3 (11.3−28.4) 9.1 (6.2−12.4) 17.1 (11.6−24.4) 7 (3.8−18.3) 6.3 (2.8−10.6) 4.1 (3−5.4) 2.1 (1.2−3.4)

31.6−219.4 19.5−64.3 8.1−65.6 4.2−23.7 7.5−39.3 2.8−36.3 1.1−19.8 1.8−9.1 0.8−9.5

Table 4. Estimated Changes in Exposure Associated with System Design Features for Below-Grade Portions of Metro Toronto dependent and independent variables

Montreal

% change in exposure

95% CI (LCL, UCL)


95% CI (LCL, UCL)

60.8 6.3 11.1 −3.4 8.5

(46.0, 77.2) (3.9, 8.8) (−8.8, 35.4) (−5.5, −1.2) (2.7, 14.6)

−2.1

(−6.7, 2.8)

2.1 0.2 −2.8

(−1.8, 6.2) (−0.8, 1.2) (−5.2, −0.4)

55 −2.1 15.7 −2.5 9.2

(41.7, 69.6) (−4.2, 0.1) (−5.2, 41.2) (−4.5, −0.4) (3.5, 15.2)

−1.8

(−10.4, 7.6)

12.3 1.4 9.8

(4.3, 21.1) (−0.5, 3.3) (4.8, 15.0)

Vancouver % change in exposure

95% CI (LCL, UCL)

60.7 8.6 32.7 −8 −3.5

(40.3, 84.0) (0.6, 17.1) (19.8, 47.0) (−10.6, −5.3) (−12.1, 6.0)

45.6 6.9 9.6 2.2 9.8

(28.2, 65.4) (−0.5, 14.9) (−0.5, 20.8) (−0.4, 4.9) (0.5, 20.0)

PM2.5a waiting vs riding distance to outside (km) depth (tens of meters) elevation (tens of meters) section age (decades) ultrafine particlesb waiting vs riding distance to outside (km) depth (tens of meters) elevation (tens of meters) section age (decades)

a Models adjusted for wind speed, atmospheric pressure, relative humidity, temperature, and ambient PM2.5. bModels adjusted for wind speed, atmospheric pressure, relative humidity, and temperature.

■

RESULTS Air Pollution Exposures. Median PM2.5 exposures were markedly higher in Toronto (95 μg/m3) than in Montreal (35 μg/m3) and Vancouver (19 μg/m3) (Table 2). Median PM10 exposures trended similarly, with Toronto levels (272 μg/m3) nearly three times higher than Montreal (104 μg/m3) and over five times higher than Vancouver (50 μg/m3). Continuously monitored PM2.5 data agreed well with gravimetric samples (Figure S2). Between-city differences in UFP exposures were less dramatic than for PM2.5 with mean UFP exposures highest in Montreal (19.5 × 103 pts/cm3) followed by Vancouver (12.6 × 103pts/cm3) and Toronto (10.5 × 103 pts/cm3) (Table 3). BC showed the same intercity trend as PM2.5 with mean levels highest in Toronto (7.8 μg/m3), followed by Montreal (4.6 μg/m3) and Vancouver (2.8 μg/m3). Exposure Trends by Season, Platform/Train, and Above/Below-Grade. Median metro exposures by season (winter/summer) and commuting activity (waiting/riding) are shown in Figures S3−S5 for PM2.5, UFP, and BC, respectively.

Statistics for these comparisons are presented in Table S3 with the comparison of above-grade/below-grade sections included for Toronto and Vancouver metros. PM2.5 showed seasonality in all three metros: higher winter levels for Montreal (winter, 46.9 μg/m3; summer, 26.3 μg/m3) and Vancouver (winter, 24.4 μg/m3; summer, 11.3 μg/m3); higher summer levels in Toronto (winter, 82.3 μg/m3; summer, 119.6 μg/m3). UFP levels were significantly higher in winter in all three metros with the most pronounced difference in Vancouver (winter, 18.3 × 103 pts/cm3; summer, 3.8 × 103 pts/cm3). The seasonality of BC reflected PM2.5 seasonal trends. In Toronto, waiting (platforms) PM2.5 exposures were significantly higher than riding (trains) exposures (waiting: 140.0 μg/m3, riding: 80.8 μg/m3). Waiting and riding PM2.5 were similar in Montreal and Vancouver (nearly equal medians and interquartile ranges). Platform UFP exposures tended to be higher than train levels in Toronto (waiting, 11.0 × 103 pts/ cm3; riding, 6.7 × 103 pts/cm3) and Vancouver (waiting: 8.3 × 103 pts/cm3; train: 6.0 × 103 pts/cm3); Montreal UFP levels D


Article


Table 5. Estimated Changes in Exposure Associated with System Design Features for Above-Grade Portions of Metro Toronto dependent and independent variables

Vancouver


95% CI (LCL, UCL)


95% CI (LCL, UCL)

12.1 −5.9 −6.8

(−14.9, 47.7) (−11.5, 0.0) (−23.3, 13.2)

11.3 −1.6 3.3

(5.6, 17.4) (−2.9, −0.3) (−0.7, 7.4)

82 −4.4 −18.8

(48.8, 122.6) (−8.3, −0.2) (−29.4, −6.7)

30.8 −2.3 0.3

(15.6, 48.1) (−5.0, 0.5) (−8.1, 9.4)

PM2.5a waiting vs riding elevation (tens of meters) section age (decades) ultrafine particlesb waiting vs riding elevation (tens of meters) section age (decades)

a Models adjusted for wind speed, atmospheric pressure, relative humidity, temperature, and Ambient PM2.5. bModels adjusted for wind speed, atmospheric pressure, relative humidity, and temperature.

the metro (2 orders of magnitude higher in Toronto, one order higher in Vancouver). In comparison, Fe was only 1 order of magnitude more abundant in the Montreal metro (4.3%) than ambient (0.11%), and Mn metro abundance was comparable to outdoors. Other metallic elements with increased abundance ratios (greater than 1 order of magnitude) were Cr, Cu, and Ni, noted in all three systems. Contribution of Metro Commuting to Estimated Daily PM2.5-Associated Element Exposures. Assuming a 70 min metro commute and 1370 min of ambient exposure, a metro commute was estimated to contribute the majority of daily exposure for several elements (Table 6). Of particular note was

were similar on platform and train. BC platform/train patterns reflected those of PM2.5 and were higher on the platform only in Toronto. Toronto and Vancouver metro systems run above or below grade for various portions of their routes; approximately 20% of the Toronto system and 80% of the Vancouver system is abovegrade. Median below-grade PM2.5 exposures were more than twice above-grade levels (Toronto: 106.4 μg/m3 below-grade, 46.7 μg/m3, above-grade; Vancouver: 39.7 μg/m3 below-grade, 15.1 μg/m3 above-grade). Differences were less dramatic for UFPs, but median UFP exposures in Toronto were higher above-grade (12.0 × 103 pts/cm3) than below-grade (8.8 × 103 pts/cm3). Conversely, median UFP levels in Vancouver were higher below-grade (11.0 × 103 pts/cm3) than above-grade (6.8 × 103 pts/cm3). BC patterns were higher below-grade in both Toronto and Vancouver. System Features and Commuter Exposures. Multivariate models evaluating system design features as determinants of commuter exposures are presented for below-grade and above-grade metro routes in Tables 4 and 5. In the belowgrade analysis (Table 4), exposures were higher waiting on platforms versus riding for PM2.5 and UFP for Toronto and Vancouver. Platform versus train was not a determinant of exposures in Montreal. System features found to relate to increased PM2.5 exposures included: distance to outside air (Toronto = 6.3%/km, Vancouver = 8.6%/km), depth (Vancouver = 32.7%/10 m), and elevation (Toronto = −3.4%/10 m, Vancouver = −8.0%/10 m). System features related to increased UFP exposures included depth (Montreal =12.3%/10 m) and elevation (Toronto = 2.5%/10m). Section age was also related to PM2.5 exposure (Toronto = 8.5%/ decade, Montreal = −2.8%/decade) and UFP (approximately 10%/decade in all three cities). Table 5 presents the multivariate analysis for the above grade portions of the Toronto and Vancouver metro. The inverse relationship between elevation and all three pollutants was still present in the Toronto system but only for PM2.5 in Vancouver. Platform levels were higher than riding levels for both PM2.5 and UFPs in the above-grade portion of the Vancouver system. PM-Elemental Composition Analysis. Fourteen elemental components of PM2.5 were compared between metro and ambient environments. Metro PM2.5 was found to be highly enriched in several elements including Fe, Cr, Co, Ni, and Ba (Table S4). The most abundant of the sampled elements in Toronto and Vancouver was Fe. In these cities, Fe was at least 2 orders of magnitude more abundant in the metro (Toronto, 54%; Vancouver, 17%) than outdoors (Toronto, 0.02%; Vancouver, 0.12%). Mn was also notably more abundant in

Table 6. Percent Contribution of Metro Commute to Daily Exposure of PM2.5 Elementsa element

Toronto

Montreal

Vancouver

Al As Ba Cd Cr Cu Fe Mn Mo Ni Pb Sr V Zn

63% 27% 96% 14% 94% 88% 99% 93% 69% 79% 36% 29% 32% 24%

51% 8% 8% 15% 41% 97% 80% 37% 17% 43% 28% 12% 9% 17%

24% 15% 15% 13% 53% 95% 42% 91% 49% 19% 15% 16% 12%

a

Percent contribution of 70 min metro commute (4.9% of day) to overall daily exposure (eq 3).

exposure to barium (96%) and manganese (93%) in Toronto, copper in Montreal (97%), molybdenum in Vancouver (91%), and iron in all three cities (99%, 80%, and 95%, respectively). The estimated 4.9% of time spent in the metro contributed 21.2%, 11.3%, and 11.5% of daily PM2.5 exposure in Toronto, Montreal, and Vancouver, respectively (Table 7).

■

DISCUSSION This exposure assessment and analysis of Canadian metro exposures has revealed substantial differences in exposure between systems related to differences in wheel and brake design. As well, differences in exposure were revealed within systems between platform and train environments and below and above grade sections of a system. Further, these metro E


Article


component of metro system-related PM rather than infiltrated vehicle traffic or in-metro diesel combustion emissions. UFPs were higher in winter in all metros, resembling the seasonality in other studies sampling UFPs,28,29. As well, a 5year study of particle count exposures in Toronto found winter UFP levels to be significantly higher than summer levels for particles ranging from 8 to 50 nm in size.30 In the below-grade sections of the Toronto and Vancouver metro systems, platform levels were substantially higher than those in trains (∼60% higher on platforms in both systems). This could be on account of train ventilation systems providing particle deposition and filtration. Some metro studies have shown PM2.5 platform levels to be notably higher than train levels.9,31 In the case of Toronto and Vancouver, time spent waiting on the station platform may contribute a substantial proportion of total commuter exposure. Conversely, platform exposures were similar to train exposures in the Montreal metro, as seen in some other international metros.25,26,32 UFP levels also showed significant differences in above- and belowgrade sections, but concentrations were generally low (Table S3). For example, on-road UFP concentrations measured in the private vehicle commuting component of this study were 2−4 times higher than in subways.12 The PM2.5 and BC exposures in the below-grade sections of the Toronto and Vancouver metros were notably higher than in the above-grade sections (see Table S3). As well, in the belowgrade sections of these two systems, these two pollutants were found to increase with distance to an above-ground segment of the system (positive association with “distance to outside”). Combined, these two factors provide an indication of the capacity by which access to open air dilutes PM2.5 and BC levels. The sections of the below-grade portions of these two systems located on low-lying land (low elevation) were also found to have higher levels of PM2.5 (inverse association with “elevation”). The mechanism behind this association may be related to the finding that a high proportion of metro PM is sourced from the system itself through frictional forces. These frictional forces such as braking and grinding between wheels and rails may be higher at lower parts of the system where grade changes are sharper and braking forces need to be increased. PM2.5 Composition and Estimated Metro Contributions to Daily Exposures. Iron was the most abundant of the sampled elements in Toronto and Vancouver and the second most abundant in Montreal (second to copper) (Table S4). High enrichments of iron have also been seen in other metro sampling studies.3−5,10,23,24,33 Metro PM was much more enriched in iron and manganese than ambient PM, indicating the presence of an Fe-rich and Mn-rich source within the metro environments. In the cases of Toronto and Vancouver, this is likely the result of the conventional steel wheel−steel rail interface known to generate steel “rail dust” through friction.19 This is not the case in Montreal where rubber wheels and concrete rollways are used. However, the iron and manganese enrichments still suggest a significant steel source in the metro. This may be explained by the existence of redundant steel wheels installed on the bogies in behind the rubber wheels. Their purpose is to provide a surface for the wooden brake pads as well as flanges that run along a steel rail guidance track. The ratio of Mn/Fe further suggests the source to be steel abrasion. In all three metros, the Mn/Fe ratio was approximately 0.01 (see Figure S9−11), consistent with composition of rail steel.5,10,19 Other metallic elements found

Table 7. Microenvironmental Concentrations of PM2.5 and Percent Contribution of Metro Commute to Daily PM2.5 microenvironment

Toronto

Montreal

Vancouver

71.9 124.6 15.1 21.2%

32.3 31.3 12.9 11.3%

16.6 18.3 6.6 11.5%

3

mean PM2.5 (μg/m ) traina platforma ambient % contribution of metro PM2.5 exposure to daily PM2.5b a

Adjusted for 11% positive bias.17 bPercent contribution of 70 min metro commute (4.9% of day) to overall daily exposure (eq 4).

environments are characterized as having concentrations of fine-fraction, steel-related elemental exposures which surpass ambient levels by one or more orders of magnitude. In all cities, a typical 70 min commute, which constitutes 4.9% of the day, was estimated to contribute over 50% of the estimated daily exposure to several PM2.5 metals (Fe, Mn, Mo, Cu, and Ni) and 21% of daily PM2.5 exposure in Toronto, 11% in Montreal, and 12% in Vancouver. Considering the combined daily ridership of these three metro systems (Table 1), a significant portion of Canada’s population is being exposed to these particulates on a daily basis. Overall, PM2.5 levels were found to be highest in the Toronto metro, a system that is largely below-grade and using a traditional steel wheel-steel rail rolling stock design. Steel wheel-steel rail metro design has been noted to generate PM “rail dust”.5,19−23 This environment is also subject to the regular resuspension of particles due to the piston effect of train movement. The piston effect has been suggested by the examination of temporal variability in metro PM levels which coincide with a sharp rise in wind speed/turbulence and the motion of a train.21,24 Montreal is also a below-grade system, but the much lower sampled PM2.5 levels may be due to the use of rubber wheels and concrete rollways compared to a conventional steel wheel-steel rail design. Vancouver shares Toronto’s steel wheel−steel rail design, but the low PM2.5 levels sampled are consistent with its predominantly above-grade and elevated system, open to ambient environment, and thus permitting greater dispersion of system-generated PM. The higher levels of PM associated with these metro features (steel wheel−steel rail interface, below-grade versus above-grade) have also been reported in other studies.25,26 While BC levels are frequently attributed to fresh combustion emissions, by exploring their relationship with UFPs (primarily sourced by fresh diesel emissions), we found evidence suggesting that BC levels monitored in below-grade metro environments are unlikely due primarily to vehicle exhaust emissions. Spearman’s rank correlations between UFP and BC were very low in the below-grade sections of all three metros (Toronto, 0.2; Montreal, 0.09; Vancouver, 0.11; Table S5), and BC correlated strongly with UFP only in the abovegrade portion of the Vancouver metro (0.59, Table S6). In addition, below-grade BC levels were more correlated with PM2.5 than with UFP. While high BC levels in metro systems have been hypothesized to be partly due to diesel-burning maintenance units,27 these would have operated during nighttime hours outside the sampling period. The graphite connections between the third rail and the train could also be a source of BC in these systems.23 Taken together, our findings suggest that sampled BC levels could be representing a F


Article


guide decision-making processes by transportation planners in Canada and elsewhere to help reduce PM exposures for the commuting public. For existing subway infrastructure experiencing high particulate matter concentrations, interventions to improve metro air quality should be considered, for example improved ventilation, in car filtration or conducting rail dust cleaning. Future proposed exposure sampling work will explore source types contributing to the unique element-rich PM2.5 composition found in the metro microenvironment as well as the evaluation of interventions designed to improve metro air quality.

to be enriched in metro PM were Cr, Cu, Mo and Ni, which are all associated with train-related friction processes (e.g., abrasion of steel rail, brake pads, electrification rail).5,19−23 Ba, which is a useful marker of wear from friction brake pads,23,24 is used to supplement the regenerative braking systems and was found in highest levels in Toronto and Vancouver. Conversely, Ba levels were much lower in Montreal, suggesting a system design difference and which is likely associated with Montreal’s use of wooden brake pads instead of conventional metallic brake pads. Exposures measured in this study may overestimate typical metro exposures experienced outside of rush hour as sampling was conducted exclusively during peak hours. Indeed, train activity has been linked to exposure levels34 and it is therefore likely that levels are lower during off-peak hours. As well, it is important to note that PM2.5-based elemental exposures represented each metro as a whole; exposures for above and below grade sections of the Toronto and Vancouver metros are likely different for fine fraction metals, as observed with the continuously sampled PM2.5 data. As well, sampling represented more time on platforms in comparison to that of an actual commute where most time is spent riding. This may have resulted in higher measured elemental concentrations of PM2.5 elemental levels than may be experienced by a typical commuter for metros where platform PM2.5 levels are higher than train levels; however, a sensitivity analysis assuming a more typical commute of 14% platform exposure time and 86% train exposure time was found to result in a less than two percent difference in estimated metro contributions to elemental PM2.5 daily exposures. Finally, in our microenvironmental models, we assumed that all PM2.5 mass and elemental concentration exposures outside the metro commute were represented by central site ambient monitoring data. This assumption has been made by the large body of research investigating the health effects of PM2.5 which use central site data to represent personal exposures. Indeed, the results of this paper represent an estimate of the exposure misclassification which can result from this assumption. As well, several studies have suggested metro commuting to be a dominant source of PM2.5 elemental exposures. Research investigating changes in personal elemental exposures related to the introduction of the fuel additive MMT (methylcyclopentadienyl manganese tricarbonyl) has found time spent in the metro to contribute substantially to personal exposures of fine fraction metals related to steel. A study assessing personal exposures of a representative sample of Torontonians found time spent in the subway to be the most important predictor of PM2.5 Mn. This was supported by subsequent sampling of six metro platforms which found mean concentrations of PM2.5 (158 μg/m3) PM2.5 Mn (0.428 μg/m3).35,36 These levels are remarkably similar to the levels of this study. A comparison of the personal exposures of taxi cab drivers and office workers in London revealed the highest levels of PM2.5 Mn to be among the subjects within the office worker group who commuted via the metro.35,36 The contribution of subway commuting to personal PM2.5 levels is likely shared for other metal related elemental exposures. In New York City, an exposure study of metro-commuting teenagers revealed personal exposures to be higher than exposures outdoors and indoors at home for Cr, Mn, and Fe.10 In general, findings suggest that particulate air pollution exposures may be significantly increased in metro environments relative to outdoors, particularly for some metallic elements. Moreover, our results suggest that metro system features can significantly influence PM2.5. These findings may be used to

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.6b05775. Summary of data compilation and quality assurance along with estimates of precision and bias (Table S1), data coverage (Table S2), metro exposures by season, riding/waiting, and above/below grade (Table S3), ambient and metro PM2.5 elemental levels and enrichments (Table S4), Spearman’s correlation analysis for PM2.5, UFP, and BC exposures in below-grade sections (Table S5) and above-grade sections of metros (Table S6), an image of the personal sampling setup (Figure S1), gravimetric versus continuous PM2.5 (Figure S2), exposure levels by city, waiting/riding, and season for PM2.5 (Figure S3), UFP (Figure S4), and BC (Figure S5), spatial maps of PM2.5 for Toronto (Figure S6), Montreal (Figure S7), and Vancouver (Figure S8), and correlation plots of Fe and Mn enrichments in PM2.5 samples of Toronto (Figure S9), Montreal (Figure S10), and Vancouver (Figure S11) (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; mailing address: 269 Laurier Ave West, Ottawa, Ontario, Canada, K1A 0K9. ORCID

Keith Van Ryswyk: 0000-0003-3349-0723 Lance Wallace: 0000-0002-6635-2303 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank the Toronto Transit Commission, la Société de transport de Montréal, and Vancouver’s Translink for their support. We would also like to thank the field technicians from Carleton and McGill University and the universities of British Columbia, Toronto, and Montreal for their diligent work. We thank Dr. Carlyn J. Matz, Hongyu You, and Marika Egyed for their contribution to this study. We thank Dr. Phil Hopke and the journal reviewers for their insightful comments. This study was funded by Health Canada.

■

REFERENCES

(1) Miao, Q.; Bouchard, M.; Chen, D.; Rosenberg, M. W.; Aronson, K. J. Commuting behaviors and exposure to air pollution in Montreal, Canada. Sci. Total Environ. 2015, 508, 193−198. (2) Weichenthal, S.; Dufresne, A.; Infante-Rivard, C.; Joseph, L. Determinants of ultrafine particle exposures in transportation

G


Article

Environmental Science & Technology environments: findings of an 8-month survey conducted in Montreal, Canada. J. Exposure Sci. Environ. Epidemiol. 2008, 18, 551−563. (3) Loxham, M.; Cooper, M. J.; Gerlofs-Nijland, M. E.; Cassee, F. R.; Davies, D. E.; Palmer, M. R.; Teagle, D. A. H. Physicochemical Characterization of Airborne Particulate Matter at a Mainline Underground Railway Station. Environ. Sci. Technol. 2013, 47, 3614− 3622. (4) Park, D.; Lee, T.; Hwang, D.; Jung, W.; Lee, Y.; Cho, K.; Kim, D.; Lee, K. Identification of the sources of PM10 in a subway tunnel using positive matrix factorization. J. Air Waste Manage. Assoc. 2014, 64, 1361. (5) Park, D.; Park, E.; Oh, M.; Yoon, Y.; Lee, K. Source identification of PM10 pollution in subway passenger cabins using positive matrix factorization. Atmos. Environ. 2012, 49, 180−185. (6) Karanasiou, A.; Viana, M.; Querol, X.; Moreno, T.; de Leeuw, F. Assessment of personal exposure to particulate air pollution during commuting in European cities–recommendations and policy implications. Sci. Total Environ. 2014, 490, 785. (7) Kaur, S.; Nieuwenhuijsen, M. J.; Colvile, R. N. Fine particulate matter and carbon monoxide exposure concentrations in urban street transport microenvironments. Atmos. Environ. 2007, 41, 4781−4810. (8) Nieuwenhuijsen, M. J.; Gómez-Perales, J. E.; Colvile, R. N. Levels of particulate air pollution, its elemental composition, determinants and health effects in metro systems. Atmos. Environ. 2007, 41, 7995− 8006. (9) Aarnio, P.; Yli-Tuomi, T.; Kousa, A.; Mäkelä, T.; Hirsikko, A.; Hämeri, K.; Räisänen, M.; Hillamo, R.; Koskentalo, T.; Jantunen, M. The concentrations and composition of and exposure to fine particles (PM 2.5) in the Helsinki subway system. Atmos. Environ. 2005, 39, 5059−5066. (10) Chillrud, S. N.; Epstein, D.; Ross, J. M.; Sax, S. N.; Pederson, D.; Spengler, J. D.; Kinney, P. L. Elevated airborne exposures of teenagers to manganese, chromium, and iron from steel dust and New York City’s subway system. Environ. Sci. Technol. 2004, 38, 732−737. (11) Knibbs, L. D.; de Dear, R. J. Exposure to ultrafine particles and PM2.5 in four Sydney transport modes. Atmos. Environ. 2010, 44, 3224−3227. (12) Weichenthal, S.; Van Ryswyk, K.; Kulka, R.; Sun, L.; Wallace, L.; Joseph, L. In-vehicle exposures to particulate air pollution in Canadian metropolitan areas: the urban transportation exposure study. Environ. Sci. Technol. 2015, 49, 597−605. (13) American Public Transport Association Ridership Report. http://www.apta.com/resources/statistics/Pages/ridershipreport.aspx. (14) Hämeri, K.; Koponen, I. K.; Aalto, P. P.; Kulmala, M. The particle detection efficiency of the TSI-3007 condensation particle counter. J. Aerosol Sci. 2002, 33, 1463−1469. (15) Demokritou, P.; Kavouras, I. G.; Ferguson, S. T.; Koutrakis, P. Development and Laboratory Performance Evaluation of a Personal Multipollutant Sampler for Simultaneous Measurements of Particulate and Gaseous Pollutants. Aerosol Sci. Technol. 2001, 35, 741−752. (16) United States Environmental Protection Agency, National Exposure Research Laboratory Exposure Methods and Measurement Division, Research Triangle Park, NC; List of Designated Reference and Equivalent Methods, 2016. (17) Wheeler, A. J.; Xu, X.; Kulka, R.; You, H.; Wallace, L.; Mallach, G.; Van Ryswyk, K.; MacNeill, M.; Kearney, J.; Dabek-Zlotorzynska, E.; Wang, D.; Poon, R.; Williams, R.; Stocco, C.; Anastassopoulos, A.; Miller, J. D.; Dales, R.; Brook, J. R. Windsor, Ontario exposure assessment study: design and methods validation of personal, indoor, and outdoor air pollution monitoring. J. Air Waste Manage. Assoc. 2011, 61, 142−156. (18) Matz, C. J.; Stieb, D. M.; Davis, K.; Egyed, M.; Rose, A.; Chou, B.; Brion, O. Effects of Age, Season, Gender and Urban-Rural Status on Time-Activity: Canadian Human Activity Pattern Survey 2 (CHAPS 2). Int. J. Environ. Res. Public Health 2014, 11, 2108−2124. (19) Bukowiecki, N.; Gehrig, R.; Hill, M.; Lienemann, P.; Zwicky, C. N.; Buchmann, B.; Weingartner, E.; Baltensperger, U. Iron, manganese and copper emitted by cargo and passenger trains in Zürich

(Switzerland): Size-segregated mass concentrations in ambient air. Atmos. Environ. 2007, 41, 878−889. (20) Burkhardt, M.; Rossi, L.; Boller, M. Diffuse release of environmental hazards by railways. Desalination 2008, 226, 106−113. (21) Fridell, E.; Ferm, M.; Ekberg, A. Chalmers University of Technology; Department of Applied Mechanics, D.; Chalmers tekniska högskola; School of Mechanical Engineering; Institutionen för tillämpad mekanik, D. Emissions of particulate matters from railways − Emission factors and condition monitoring. Transportation Research Part D 2010, 15, 240−245. (22) Lorenzo, R.; Kaegi, R.; Gehrig, R.; Grobéty, B. Particle emissions of a railway line determined by detailed single particle analysis. Atmos. Environ. 2006, 40, 7831−7841. (23) Moreno, T.; Martins, V.; Querol, X.; Jones, T.; BéruBé, K.; Minguillón, M. C.; Amato, F.; Capdevila, M.; de Miguel, E.; Centelles, S.; Gibbons, W. A new look at inhalable metalliferous airborne particles on rail subway platforms. Sci. Total Environ. 2015, 505, 367− 375. (24) Moreno, T.; Reche, C.; Rivas, I.; Cruz Minguillon, M.; Martins, V.; Vargas, C.; Buonanno, G.; Parga, J.; Pandolfi, M.; Brines, M.; Ealo, M.; Sofia Fonseca, A.; Amato, F.; Sosa, G.; Capdevila, M.; de Miguel, E.; Querol, X.; Gibbons, W. Urban air quality comparison for bus, tram, subway and pedestrian commutes in Barcelona. Environ. Res. 2015, 142, 495−510. (25) Martins, V.; Moreno, T.; Mendes, L.; Eleftheriadis, K.; Diapouli, E.; Alves, C. A.; Duarte, M.; de Miguel, E.; Capdevila, M.; Querol, X.; Minguillón, M. C. Factors controlling air quality in different European subway systems. Environ. Res. 2016, 146, 35−46. (26) Kim, K. Y.; Kim, Y. S.; Roh, Y. M.; Lee, C. M.; Kim, C. N. Spatial distribution of particulate matter (PM10 and PM2.5) in Seoul Metropolitan Subway stations. J. Hazard. Mater. 2008, 154, 440−443. (27) Vilcassim, M. J.; Thurston, G. D.; Peltier, R. E.; Gordon, T. Black carbon and particulate matter (PM2.5) concentrations in New York City’s subway stations. Environ. Sci. Technol. 2014, 48, 14738− 14745. (28) Zhu, Z.; Hinds, W.; Shen, S.; Sioutas, C. Seasonal Trends of Concentration and Size Distribution of Ultrafine Particles Near Major Highways in Los Angeles. Aerosol Sci. Technol. 2004, 38, 5. (29) Bukowiecki, N.; Dommen, J.; Prévôt, A. S. H.; Weingartner, E.; Baltensperger, U. Fine and ultrafine particles in the Zü rich (Switzerland) area measured with a mobile laboratory: an assessment of the seasonal and regional variation throughout a year. Atmos. Chem. Phys. 2003, 3, 1477−1494. (30) Sabaliauskas, K.; Jeong, C.; Yao, X.; Jun, Y.; Jadidian, P.; Evans, G. J. Five-year roadside measurements of ultrafine particles in a major Canadian city. Atmos. Environ. 2012, 49, 245−256. (31) Kam, W.; Cheung, K.; Daher, N.; Sioutas, C. Particulate matter (PM) concentrations in underground and ground-level rail systems of the Los Angeles Metro. Atmos. Environ. 2011, 45, 1506−1516. (32) Park, D.; Ha, K. Characteristics of PM10, PM2.5, CO2 and CO monitored in interiors and platforms of subway train in Seoul, Korea. Environ. Int. 2008, 34, 629−634. (33) Martins, V.; Moreno, T.; Minguillón, M. C.; van Drooge, B. L.; Reche, C.; Amato, F.; de Miguel, E.; Capdevila, M.; Centelles, S.; Querol, X. Origin of inorganic and organic components of PM2.5 in subway stations of Barcelona, Spain. Environ. Pollut. 2016, 208, 125− 136. (34) Martins, V.; Moreno, T.; Minguillón, M. C.; Amato, F.; de Miguel, E.; Capdevila, M.; Querol, X. Exposure to airborne particulate matter in the subway system. Sci. Total Environ. 2015, 511, 711−722. (35) Pfeifer, G. D.; Harrison, R. M.; Lynam, D. R. Personal exposures to airborne metals in London taxi drivers and office workers in 1995 and 1996. Sci. Total Environ. 1999, 235, 253−260. (36) Pellizzari, E. D.; Mason, R. E.; Clayton, C. A.; Thomas, K. W.; Cooper, S.; Piper, L.; Rodes, C.; Goldberg, M.; Roberds, J.; Michael, L. Manganese Exposure Study (Toronto); Research Triangle Institute: Research Triangle Park, NC, 1998; Vol. RTI/6312/02-01 F.

H


Metro Commuter Exposures to Particulate Air ... - ACS Publications

Recommend Documents