Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Feb 1.
Published in final edited form as: Transp Res D Transp Environ. 2022 Dec 30;115:103579. doi: 10.1016/j.trd.2022.103579

Investigation of a river-tunnel effect on PM2.5 concentrations in New York City subway stations

David G Luglio 1, Tri Huynh 1, Antonio Saporito 1, Terry Gordon 1
PMCID: PMC9879276  NIHMSID: NIHMS1861972  PMID: 36711192

Abstract

It is well-documented that subway stations exhibit high fine particulate matter (PM2.5) concentrations. Little is known about the potential of river-tunnels to increase PM2.5 concentrations in subways. We hypothesized a “river-tunnel” effect exists: Stations adjacent to poorly ventilated tunnels that travel beneath rivers exhibit higher PM2.5 concentrations than more distant stations. Accordingly, the PM2.5 concentrations were monitored at stations adjacent to and two- and three-stations distant from the river-tunnel. Multivariate linear regression analysis was conducted to disentangle how proximity to a river-tunnel and other factors (e.g., depth) influence concentrations. Stations adjacent to a river-tunnel had 80–130% higher PM2.5 concentrations than more distant stations. Moreover, distance from a river-tunnel was the strongest PM2.5-influencing factor This distance effect was not observed at underground stations adjacent to a river-bridge. The “river-tunnel” effect explains some of the inter-station variability in subway PM2.5 concentrations. These results support the need for improving ventilation systems in subways.

Keywords: Subway, PM2.5, fine particulate matter, urban air pollution, New York City, occupational health

Graphical abstract

graphic file with name nihms-1861972-f0001.jpg

1.1. Introduction

Underground subways provide rapid, inexpensive transit for millions of people globally, but have been found to expose workers and riders to high levels of fine particulate matter (PM2.5). In New York City (NYC) alone, over 4 million people ride the subway system daily and while transit systems, such as NYC’s Metropolitan Transit Authority (MTA), have reduced aboveground vehicle traffic and ambient air pollution (Lu et al., 2018), subways pose considerable air quality problems for transit workers and riders. Most significantly, subway air contains high levels of particulate matter of 2.5 μm or smaller (PM2.5). These PM2.5 levels were observed to be elevated several-fold over ambient levels in over 30 subway stations in NYC (Vilcassim et al., 2014). Similarly, Chillrud and colleagues documented the importance of the metal exposures in NYC subways, finding that the subway air concentrations of Fe, Mn, and Cr were more than 100 times ambient NYC levels (Chillrud et al., 2004, Chillrud et al., 2005). Because of the increased respiratory and cardiovascular morbidity and mortality associated with exposure to ambient PM2.5 (Xu et al., 2020, Li et al., 2018, Wang et al., 2019, Requia et al., 2018, Thurston et al., 2016), a key health burden question for commuters and transit workers is whether subway air is safe to breathe.

Our most recent research indicates this problem is widespread in other U.S. subway systems. At rush hour, subway station PM2.5 in Boston, Washington, DC, and Philadelphia were 33-, 28-, 9-fold higher than ambient PM2.5, respectively, although lower than in NYC subways (Luglio et al., 2021). Elevated PM2.5 air pollution in U.S. subway stations has been corroborated by numerous studies throughout North America (Kam et al., 2011, Van Ryswyk et al., 2017, Figueroa-Lara et al., 2019) and the world (Martins et al., 2016, Xu and Hao, 2017, Smith et al., 2020), although the levels are lower than reported for NYC subways (Luglio et al., 2021). Importantly, underground subway aerosols are rich in carbon and metals – particularly iron – but the health impact burden of these metal and carbon rich PM2.5 mixtures on subway commuters and workers is not known.

The source(s) and the reasons for the elevated levels of subway PM2.5 have been previously investigated. It is likely that suspension of particles, produced by the abrasion of brakes and wheels and rails and electric power conduits, is the source of these iron-rich particles (Lee et al., 2018, Minguillón et al., 2018, Park et al., 2012). Other factors, such as ventilation (Moreno et al., 2017, Moreno et al., 2014), depth below the surface (Figueroa-Lara et al., 2019) , train frequency (Tu and Olofsson, 2021), and frequency of cleaning (Johansson and Johansson, 2003) contribute to varying concentrations of airborne particles reported in underground subway and rail systems. In combination, these factors result in the poor air quality encountered by commuters and workers in underground transit systems. Previously, we observed that the highest PM2.5 levels in the NYC/NJ PATH rail system occurred in the stations close to the tunnel under the Hudson River (Luglio et al., 2021). Based upon this observation, we hypothesized that PM2.5 generated in the poorly ventilated tunnels under the waterways surrounding Manhattan is a significant contributor to airborne PM2.5 concentrations encountered on subway platforms. As trains pass through these tunnels, they can generate, resuspend, and push airborne PM into neighboring stations. The objective of this study was to test this hypothesis, which would add to the body of knowledge on why PM concentrations vary between subway stations. Our results may be applicable to other cities with tunnels that travel beneath waterways such as London and Hong Kong. In this study, we expanded our exposure assessment to measuring PM2.5 concentrations in the 3 stations closest to both sides of all the river tunnels used by the subway/rail lines in NYC.

2. Methods

2.1. Sampling Sites

This study investigated the relative PM2.5 concentrations at NYC/NJ’s MTA and PATH stations adjacent to subway tunnels that travel beneath the 2 rivers on the East and West sides of Manhattan. More specifically, underground stations in close proximity to seven MTA subway tunnels that travel beneath the East River in NYC were selected for investigation. These tunnels service eleven subways lines (e.g. A, C, E, F, M, 4, 5, R, N, W, and L) that connect the boroughs of Manhattan, Brooklyn, and Queens. In addition, a Hudson River tunnel, which services the 33rd Street-Journal Square line of the PATH rail system between Manhattan and NJ was included. For a comparison, measurements were also conducted at underground stations on a line which crosses the East River by a bridge (i.e. B).

To compare PM2.5 concentrations at underground subway stations at different distances from the tunnels, real-time concentrations were measured on the platform of the three consecutive underground stations closest to the river, on both sides of the tunnel, for a total of six stations per tunnel or bridge. These stations are labeled in this manuscript as M3, M2, and M1 for Manhattan stations located three, two, and one stops from the river-tunnel, respectively. Likewise, the stations in Brooklyn, Queens and Jersey City are labeled BQ1, BQ2, and BQ3 in increasing number of stations from the river. A map of all stations is presented in Figure 1. In addition, all stations are listed in sequential order in the supplementary material (Table A1). All real-time measurements of PM2.5 were collected from 0900 hours to 1300 hours during weekdays in February and March 2022.

Figure 1 –

Figure 1 –

Open in a new tab

Map of sampled stations. The location of the stations are located with circles, squares or triangles upon lines both in the respective color of the designated line (i.e. green for 4,5, yellow for R, gray for L, blue for A,C and E, and orange for the F and B). Orange squares are used for the B to differentiate it from the nearby F. PATH stations are indicated by yellow triangles on a red dotted line. Numbers adjacent to each marker correspond to the station on the list to the right. The map was created on ArcMap. The MTA lines were generated using publicly available data created by Frank Donelly through the Baruch Geospatial portal based on MTA data and published to CCE_NYC AGOL by Glen Johnson, CUNY School of Public Health. The PATH lines were generated using publicly available data provided by Glenn D. Newman PLS, PP, director of GIS-Transportation at NJ transit and processed by the NJ Office of information Technology (NJCIT), Office of Geographic Information Systems (OGIS).

2.2. Real-Time Measurements

Real-time PM2.5 concentrations were measured using nephelometric-based personal DataRAMs (pDR 1500; Thermo Fisher Scientific, Franklin, MA). These instruments were outfitted with 2.5-μm cut-point inlet cyclones, logged concentrations at one-second averaging intervals, and zeroed with HEPA-filtered air before each run. Real-time data from each pDR were adjusted with correction coefficients derived from the calibration curve obtained from gravimetric PM2.5 measurements (see below).

The sampling scheme is detailed in the supplementary section (Figure A1). To summarize, as a means to minimize temporal variability while measuring the targeted spatial differences, three investigators sampled at each of three consecutive stations (e.g. M3, M2, M1) at the same time. Real-time PM2.5 concentrations were collected on subway platforms for a minimum of 5 minutes before all three individuals boarded a train and moved on to the next stop, until all stations were sampled at least twice. The sampling instruments were placed at chest height (approximately 1.4 meters) at the center of each sampling platform. To measure the PM2.5 concentration inside a tunnel itself, an Airbeam 2 (Habitat Map, Brooklyn, NY) was attached to the outside of a train car as it traveled from station M3 to BQ3 through the river-tunnel. The Airbeam 2 concentrations were corrected by comparing it to a calibrated pDR-1500.

2.3. Gravimetric and Elemental Measurements

PM2.5 was collected for 30 to 45 minutes at each of the stations on the 4,5 lines (Table A1) on 37-mm low-trace element Teflon™ (Pall Inc., Ann Arbor, MI) filters with a 2.5-μm cut-point inlet Personal Environmental Monitor (PEM; SKC, Inc., Shoreview, MN), and a calibrated Leland Legacy Pump (SKC, Inc.) at 10 L/min.

Before and after sampling, the filters were conditioned in a temperature- and humidity-regulated weighing chamber (MTL, Minneapolis, MN) at 21 ± 1° C and 35 ± 5% RH for at least 24 hours. After conditioning, the post- and pre-sampling filters were weighed using a microbalance (Model MT5, 1 μg readability; Mettler-Toledo, Columbus, OH). The PM2.5 mass concentrations were derived through standard gravimetric analysis. Laboratory blank samples were used to correct for daily variation in the micro-balance analyses.

All six filters were analyzed by X-ray fluorescence (XRF) (Epsilon 5 ED-XRF, PAN Analytical B.V., Netherlands) for elemental concentration analysis. Elements included in these analyses include the following although detection limits for many elements were not achieved: Na, Mg, Al, Si, P, S, Cl, K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, Ge, As, Se, Br, Rb, Sr, Y, Zr, Nb, Mo, Rh, Pd, Ag, Cd, In, Sn, Sb, Te, I, Cs, Ba, La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Hf, W, Ir, Pt, Au, Hg, Tl, Pb, Bi, and U.

2.4. Calibration, regression, and statistical analysis

For calibration, all pDR-1500s were co-located with a PM2.5 filter sampling system for gravimetric analysis in the six different stations most proximal to each tunnel. A linear curve was developed comparing the average pDR-1500 concentrations and the gravimetrically-determined concentration for each of the six stations. The y-intercept of the curve was set as 0. Subsequently, the coefficient of the independent value was used as the correction factor for pDR-1500 values. Correction of these pDR-1500 values were done on an individual instrument basis.

The PM2.5 concentrations were averaged for each platform across the time period that each investigator was present on the platform. Each of these instances counted as a measurement for that platform (i.e., n = 2 for M3 and BQ3, and 3 for all others). All of these samples were averaged together for each platform independently for each line (n=59) for presentation in this manuscript. The A,C and F-downtown lines, E and R-uptown lines, and B and F-downtown lines share the same platforms at Jay Street-MetroTech, 36th Street, and Dekalb Avenues, respectively. The data for these readings are considered as separate samples as they are temporally concurrent within a single line but distinct from all others. In addition, the distance (i.e., in meters following the subway line) to the river from these platforms differs depending on the line considered. Some samples are excluded from the regression model as noted below.

To account for potential confounding factors, a multivariate regression analysis was performed with the PM2.5 concentrations at all “river-tunnel” stations as the dependent variable. The distance from the river, station age, station depth/number of steps, and number of tracks were included as independent variables. Spearman-rank correlation tests for the relationship between each of the factors and PM2.5 concentration were also performed. In addition, a correlation matrix was constructed showing the relationships of each independent variable to each other (Table A3). Highly correlated variables (r>0.7) were excluded from the regression model.

All multivariate regression analyses were conducted with R (version, R version 3.6.1; R Core Team, 2019) using the agricolae (version 1.3−1; de Mendiburu and Yaseen, 2019) package. The PM2.5 concentration was natural log transformed for each underground station to attain a normal distribution, and was included as the dependent variable. All independent variables are as listed: absolute number of stations from river (i.e. 1, 2, 3), distance from the river in kilometers, age of station in years, number of steps from surface to platform sampled, number of tracks, number of platforms, type of platform (i.e. island (1), side (2), both (3)), and number of trains per minute passing through the station. The distance from the river was measured on ArcMap; the coordinates of the stations were overlaid onto a map of the New York Metropolitan area with the subway tunnels with publicly available data (i.e., NYC Open Data for the coordinates; ArcMap Online maps for the MTA and PATH tunnel lines) and the distance to the river was measured by tracing the tunnels using the measure distance tool. The number of steps was selected as a proxy for the depth of a station since the depth data were available for only a few stations. The typical step height in the stations (i.e., MTA or PATH) is 18 cm. The type of platform present in a station is coded as: 1 for island, 2 for side, and 3 for both. The number of trains per minute, or frequency, was determined by counting the number of trains passing through a station in a fifteen-minute interval for all stations in a single line. This specified time period was between 0900 and 1200 hours. All the data for each of these factors are included in Table A2. Data for the stations solely on the “bridge” line (n=6) are excluded from the ‘tunnel’ analysis. Roosevelt Island (n=1) is also excluded from the analysis because it does not have a proper position relative to a river tunnel. Journal Square (n=1) is also excluded because it is not an underground station. Accordingly, the total number of samples included in the model is 51. All Spearman-rank correlation tests were conducted with the same variables in the same codification.

Analysis of variance (ANOVA) was performed to compare the PM2.5 concentrations of stations at positions 1, 2, and 3, regardless of location in Manhattan, Brooklyn, Queens or Jersey City. A post-hoc Tukey test was performed to show the pairwise significant differences between groups.

All statistical analyses were conducted in R (R version 3.6.1) (R Core Team, 2019) with the agricolae package (version 1.3−1) (de Mendiburu and Yaseen, 2019). The stations only located on the “bridge” line were excluded from the ANOVA and regression analysis.

3. Results and Discussion

3.1. Evaluation of PM2.5 concentrations

The mean PM2.5 concentrations were the highest in stations adjacent to each river-tunnel connecting Manhattan to Queens, Brooklyn, or NJ (Figure 2). The stations closest to the river-tunnel (i.e., M1 and BQ1 in Figure 2) had 80 to 130% higher PM2.5 concentrations than in the stations 2 or 3 stops away from the river-tunnel. A further distance from the river past the 2nd stop, however, did not appear to result in lower PM2.5 concentrations. Statistical tests confirmed that stations at position 1 (i.e., M1 or BQ1) was significantly different than at positions 2 and 3, whereas 2 and 3 were not different from each other (Table A4). In some cases, however, the stations at position M3 or BQ3 had poorer air quality than at positions M2 or BQ2 (Table A5). Nevertheless, the hypothesis that the stations closest to the river-tunnels would have higher PM2.5 concentrations was substantiated by this study. A similar finding was observed in the unique Roosevelt Island Station (F-uptown Line) which, because it is situated in the middle of the East River, has trains that enter from river-tunnels in both the Manhattan- and Queens-bound directions (Table A5).

Figure 2 –

Figure 2 –

Open in a new tab

Average PM2.5 concentrations according to location of station. M = Manhattan subway stations and BQ = Brooklyn/Queens subway stations. The number following M or BQ indicates the relative location of the station relative to the tunnel with ‘1’ being the closest station. Data from the B-bridge line and the F-uptown line are excluded from this chart. The sample number for each of locations is 6.

To more closely examine the river-tunnel as a source of PM2.5, the PM2.5 concentration in a tunnel was directly measured using a calibrated low-cost PM sensor (Airbeam) attached to the outside of a train-car as it passed from station BQ3 to M3 on the L-line (Figure 3). The highest PM2.5 concentrations were measured while entering the BQ1 and M1 stations. The river-tunnel had slightly lower PM2.5 concentrations than in the two adjacent stations, suggesting that the source of airborne PM2.5 in the tunnel-adjacent stations included PM from the tunnel as well as station-specific sources such as human activity and outdoor PM.

Figure 3 –

Figure 3 –

Open in a new tab

Time series of PM2.5 concentration outside of a train during ride from Graham Avenue (BQ3) to 14th Street-Union Square (M3) on the L-line. The river-tunnel is indicated as the tunnel between Bedford Avenue (BQ1) and 1st Avenue (M1). PM2.5 concentration steadily rises as the train approaches the BQ1 and M1 stations from either side. The concentration remains high in the river-tunnel. M = Manhattan subway stations and BQ = Brooklyn/Queens subway stations.

To further assess the specific role of the river-tunnel as a PM2.5 source, the tunnel sampling procedure was mirrored on a subway line (B) which crosses the East River via a bridge. In this scenario, subway trains would only push ambient air into the river-adjacent underground stations. As these stations were not subjected to the piston effect of a train passing through an underground tunnel, these bridge-associated stations were not expected to have elevated PM2.5 concentrations. The PM2.5 concentrations at the bridge-associated M1 and BQ1 underground stations were 125.7 +/− 48.6 and 99.7 +/− 41.7 μg/m3, respectively, which were lower than the concentrations at M2 and M3 (i.e., the opposite of the effect seen in the river-tunnel scenario; Table A5).

All-in-all, these results are indicative of a “river-tunnel effect” contributing to the PM2.5 concentrations in the underground subway stations adjacent to the river-tunnel. Most of the NYC subway system ventilates by the piston effect, which requires a direct passageway from the tunnels to the surface. This is impossible for river-tunnels which are underwater. Instead they have large towers on either side of the river in an attempt to compensate for the lack of tunnel adjacent grates. There may also be operable fans at the ends of the river-tunnel, which may aid in air exchange. The lack of access to the surface, however, remains an issue. Because of low ventilation in the tunnels, we speculate that the PM2.5 generated in the river-tunnels has little to no exchange with ambient air and thus airborne and settled particles may accumulate over time. Thus, as trains pass through the long river-tunnels, they push highly concentrated PM2.5-laden air forward into the subway stations at either end of the tunnel.

3.2. Influence of station characteristics on PM2.5 variability

There appeared to be considerable variability in PM2.5 concentrations among stations across the different positions in the targeted subway lines (Table A5). Therefore, other factors, such as station depth, age of station, or train frequency, were expected to be influential contributors to subway station PM2.5 levels. Depth, in particular, could be a major factor because as stations get closer to the river-tunnel, they are expected to increase in depth to accommodate passage into the sub-seabed tunnels. In fact, there appears to be a weak correlation between station depth and distance to a river (Table A3). Yet, based upon the station depth in this study (i.e., counting the number of steps from street level to each platform), there was no significant correlation between depth and PM2.5 (Table 1). The Bowling Green station, for example, is adjacent to the river and has some of the highest concentrations (233.0 +/− 86.7 μg/m3) observed in this study, but appears to be the shallowest of the sampled stations on the line (Table A2 and A4). In addition, train frequency was expected to be an influential factor with increases in PM2.5 concentrations via increasing frequency of particle resuspension and generation and the piston effect of trains. An example of this is the 36th Street Station on the R-uptown and E Lines which had higher train frequencies and PM2.5 concentrations than at (the river-tunnel-adjacent) Queens Plaza station despite being shallower and further from the river (Table A2–A3). Overall, however, there was no significant correlation between train frequency (i.e., # of trains/minute) and PM2.5 concentration (r = −0.051, p-value = 0.720; Table 1). The variability of many other factors may have occluded any observable relationship of PM2.5 and train frequency.

Table 1 –

Spearman-Rank correlation coefficients and p-values for relationships between station characteristic and PM2.5 concentration.

Characteristic r p-value
Location of station 0.403 0.003
Distance from river-tunnel (meters) −0.443 0.001
Age (years) 0.174 0.221
# of tracks −0.225 0.112
# of platforms −0.289 0.0400
# of steps 0.216 0.127
Train frequency (# of trains/minute) −0.0512 0.720
Open in a new tab

The # of steps is a proxy for station depth underground.

To assess the contribution of multiple station characteristics, such as depth, a multivariate regression analysis was performed (Table 2). In this analysis, subway station distance from the river-tunnel had the most statistically significant association with PM2.5 concentrations when considering all lines together (Table 2). The type and number of platforms were the next most influential factors. The best fit model additionally included the number of tracks (r2=0.367) (Table 2). Other factors, however, not accounted for here, may be involved in determining air pollution levels in underground subway stations.

Table 2 –

Multivariate regression results. The effect of each factor is indicated in the left-hand columns on the natural log of the PM2.5 concentrations is displayed.

Ln(PM2.5 concentration (μg/m3))
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7
Train frequency (# of trains/min) 0.0014
# of steps 0.00083 0.00083
Age (years) 0.0020 0.0023 0.0022
# of tracks 0.28* 0.27 0.27 0.27
# of platforms −0.35* −0.75** −0.74** −0.73* −0.73*
Type of platform (side platforms, “2”) −0.33 −0.17 0.43* 0.20 0.22 0.22
Type of platform (both island and side, “3”) 0.60 1.14** 1.86** 1.79** 1.78** 1.78**
Distance from river (km) −0.36*** −0.32** −0.26* −0.35** −0.34** −0.34** −0.34**
Constant 5.17*** 5.18*** 5.57*** 5.51*** 5.35*** 5.24*** 5.24***
Observations 51 51 51 51 51 51 51
R2 0.193 0.282 0.329 0.367 0.357 0.343 0.328
Residual Std. Error 0.655 (df = 49) 0.618 (df = 47) 0.597 (df = 46) 0.580 (df = 45) 0.585 (df = 44) 0.591 (df = 43) 0.598 (df = 42)
F Statistic 12.9*** (df = 1; 49) 7.56** (df = 2; 48) 7.12*** (df = 4; 46) 6.79** (df = 5; 45) 5.63*** (df = 6; 44) 4.73*** (df = 7; 43) 4.04** (df = 8; 42)
Open in a new tab

There are seven models constructed as indicated by the numbers on top, reducing in complexity from 7 to 1, as the factors with the highest p-values are subsequently removed. The distance from the river-tunnel of the station is the only constant. The ß coefficients for each factor are listed for each model alongside indications of their significance with * in the top-half of the table. Below each model are model characteristics, such as its R2 and F-statistic values. The significance of the F-statistic is indicated with *. df = degrees of freedom. Observations refer to the number of samples included in each model.

*

(p-value) < 0.05;

**

< 0.01;

***

< 0.001

Previous studies have investigated the effect of various station characteristics on PM2.5 concentration with varying results. Figueroa-Lara et al. (Figueroa-Lara et al., 2019), Vilcassim et al. (2014), and Ma et al. (Ma et al., 2014) previously found that deeper subway transfer lines exhibited higher PM2.5 concentrations for the same station. The results presented here show the same directionality of correlation, albeit weakly and non-significantly (Tables 1 & 2). In this study, PM2.5 concentrations were positively correlated with station age. This matches well with previous findings in Shanghai and Barcelona (Zhao et al., 2017, Querol et al., 2012), although this age-effect was found to be not a factor in other systems (Cha et al., 2019). The number of platforms at a station can serve as a proxy for station size, which was shown here to be negatively correlated with PM2.5 concentrations, which was one of the more influential regression factors (Tables 1 & 2). Larger volumes of air-space likely dilute the mass concentration of suspended PM entering a station. Few studies, however, have investigated the size of a station’s effect on PM levels. Interestingly, the type of platform played a major role in explaining the PM level variability. PM2.5 concentrations were the highest in stations with both island and side platform(s). These stations represent a complex situation, in which these stations are larger but have more tracks and high train passing frequencies. To tease-out the mechanistic influence of this factor requires more work.

3.3. Compositional analysis

Compositional analysis was also performed on the six river-tunnel stations sampled on the 4/5 subway line. As expected, based on previous findings (Lu et al., 2015, Lee et al., 2018, Minguillón et al., 2018, Smith et al., 2020, Luglio et al., 2021), iron was the dominant element, accounting for 36 to 72% of total PM2.5 mass, although the relative concentration did vary among the stations (Figure 4). Other elements such as silicon, chromium, copper, sulfur and barium were present but comprised less than 3% of the total mass. Thus, overall, fresh generation and resuspension of the friction products of the wheels, rails, and electrical conduction elements are the likely source of airborne PM2.5 in these subway tunnels. The unknown component(s) for the PM at some of these stations, however, is relatively large. This can be attributed in part to the limitations of the analytical techniques, but also to the fact that a limited number of elements were analyzed. For example, we did not analyze for carbon and oxygen, and carbon may have contributed a significant mass to the PM2.5 as was seen in previous work (Luglio et al., 2021). Similarly, the large iron content of subway PM2.5 was present as rust (unpublished observations) and thus oxygen may also have been an important contributor.

Figure 4 –

Figure 4 –

Open in a new tab

PM2.5 composition for the six sampled stations on the 4,5 line. Elements in the “other” category include: Mg, K, Ti, Ni, Cu, Zn, As, Se, Br, Rb, Sr, Y, Cd, Sn, Ce, Pr, Eu, Er, Lu, Ir, Au, Tl, and Pb. M = Manhattan subway stations and BQ = Brooklyn/Queens subway stations.

3.4. Potential solutions

In order to reduce river-tunnel effects on PM2.5 concentrations at station platforms, innovative solutions are needed to reduce particle sources and improve ventilation to protect commuters and transit workers. Platform screen doors (PSD) are being implemented around the world, and have been shown to effectively reduce concentrations in subway stations (Kim et al., 2012, Zhang et al., 2019, Han et al., 2015). Yet, this suggestion must be met with caution as it may trap subway PM at the tracks and tunnels, increasing concentrations in those areas (Son et al., 2013). Accordingly, in-train PM levels have been demonstrated to have increased, as train cars draw air from these tunnels (Son et al., 2014). Implementation of PSD should be coupled with improved filtration of in-train air to maintain lower PM concentrations. Providing active ventilation to stations, in general and especially at river-tunnel adjacent stations, is an additional mitigation strategy (Moreno et al., 2017, Tu and Olofsson, 2021). Reducing the PM levels in the tunnels could have, presumably, the most powerful, positive impact on subway station air quality.

4. Conclusions

It has been well-documented that PM concentrations on underground subway platforms are much higher than in aboveground settings. Yet, the reasons why these concentrations are so high and why they vary among stations, subway lines, and transit systems is less clear. The results from this study have revealed that stations adjacent to river-tunnels have generally greater PM2.5 concentrations than in more distant stations. This metric of a station’s distance to a river-tunnel proved to be the most influential when considering all factors. As so, we have ascribed there to be an impactful river-tunnel effect on PM2.5 concentrations in a subway system. Efforts to reduce PM concentrations on subway platforms should focus on reducing the PM concentrations in the tunnels themselves or the ability of the particles to travel between the tunnels and the stations. One limitation in this study was our inability to collect direct information about the depth and dimensions of each station, which would have provided more accurate data for our models. Another limitation was the inability to track whether and relative proportion of particles had traveled from the tunnels and ambient air into the stations directly. This information could substantially support this river-tunnel hypothesis and could be done in future studies with tracer molecules. This river-tunnel effect is likely not unique to NYC, but to our knowledge, it has yet to be studied in other cities and would warrant investigation.

Supplementary Material

1

Highlights:

  • A “river-tunnel” effect explains some of the variability in NYC subway station PM2.5 concentrations.

  • Stations next to river-tunnels have higher PM2.5 concentrations than those 2 or 3 stops away. These results were not observed between stations on a line which crosses the river over a bridge.

  • Other potential factors such as station age, depth, or size could not explain the variability as well.

Acknowledgements:

The authors thank Dr. George Thurston for providing feedback on the statistical analyses.

NJ Transit, GIS-Transportation and NJ Office of Information Technology (NJOIT), Office of Geographic Information Systems (OGIS) for the baselayer map of the PATH transit lines utilized in Figure 1.

Funding sources:

This work was supported by NIH/NIEHS Core Center grant (ES000260), T32 Training Grant (ES007324), and NY/NJ Occupational Safety and Health Center ERC Pilot Project Award (T42 OH008422).

Glossary:

PM

Particulate matter, solid and liquid particles in the air

PM2.5

Particulate matter whose particles have an aerodynamic diameter of 2.5 micrometers or less

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References:

  1. CHA Y, TU M, ELMGREN M, SILVERGREN S & OLOFSSON U 2019. Variation in Airborne Particulate Levels at a Newly Opened Underground Railway Station. Aerosol and Air Quality Research, 19, 737–748. doi: 10.4209/aaqr.2018.06.0225 [DOI] [Google Scholar]
  2. CHILLRUD S, GRASS D, ROSS JM, COULIBALY D, SLAVKOVICH V, EPSTEIN D, SAX SN, PEDERSON D, JOHNSON D, SPENGLER JD, KINNEY PL, SIMPSON HJ & BRANDT-RAUF P 2005. Steel dust in the New York City subway system as a source of manganese, chromium, and iron exposures for transit workers. Journal of Urban Health, 82, 33–42. doi: 10.1093/jubran/jti006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. CHILLRUD SN, EPSTEIN D, ROSS JM, SPENGLER JD & KINNEY PL 2004. Elevated airborne exposures of teenagers to manganese, chromium, and iron from steel dust and New York City’s subway system. Environmental Science & Technology, 38, 732–737. doi: 10.1021/es034734y [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. DE MENDIBURU F & YASEEN M 2019. agricolae: Statistical Procedures for Agricultural Research. R package version 1.4.0 [Google Scholar]
  5. FIGUEROA-LARA JJ, MURCIA-GONZÁLEZ JM, GARCÍA-MARTÍNEZ R, ROMERO-ROMO M, TORRES RODRÍGUEZ M & MUGICA-ÁLVAREZ V 2019. Effect of platform subway depth on the presence of Airborne PM2.5, metals, and toxic organic species. Journal of Hazardous Materials, 377, 427–436. doi: 10.1016/j.jhazmat.2019.05.091 [DOI] [PubMed] [Google Scholar]
  6. HAN H, LEE J-Y, JANG K-JJI & ENVIRONMENT B 2015. Effect of platform screen doors on the indoor air environment of an underground subway station. 24, 672–681. doi: [Google Scholar]
  7. JOHANSSON C & JOHANSSON P-Å 2003. Particulate matter in the underground of Stockholm. Atmospheric Environment, 37, 3–9. doi: 10.1016/s1352-2310(02)00833-6 [DOI] [Google Scholar]
  8. KAM W, CHEUNG K, DAHER N & SIOUTAS C 2011. Particulate matter (PM) concentrations in undergound and ground-level rail systems of the Los Angeles Metro. Atmospheric Environment, 45, 1506–1516. doi: 10.1016/j.atmosenv.2010.12.049 [DOI] [PubMed] [Google Scholar]
  9. KIM K-H, HO DX, JEON J-S & KIM J-C 2012. A noticeable shift in particulate matter levels after platform screen door installation in a Korean subway station. Atmospheric Environment, 49, 219–223. doi: 10.1016/j.atmosenv.2011.11.058 [DOI] [Google Scholar]
  10. LEE Y, LEE Y-C, KIM T, CHOI JS & PARK D 2018. Sources and characteristics of particulate matter in subway tunnels in Seoul, Korea. International Journal of Environmenta Research and Public Health, 15. doi: 10.3390/ijerph15112534 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. LI T, ZHANG Y, WANG J, XU D, YIN Z, CHEN H, LV Y, LUO J, ZENG Y, LIU Y, KINNEY PL & SHI X 2018. All-cause mortality risk associated with long-term exposure to ambient PM2.5 in China: A cohort study. Lancet Public Health, 3, e470–477. doi: 10.1016/S2468-2667(18)30144-0 [DOI] [PubMed] [Google Scholar]
  12. LU H, ZHU Y, YU Q & YU J 2018. Do urban subway openings reduce PM2.5 concentrations? Evidence from China. Sustainability, 10. doi: 10.3390/su10114147 [DOI] [Google Scholar]
  13. LU S, LIU D, ZHANG W, LIU P, FEI Y, GU Y, WU M, YU S, YONEMOCHI S, WANG X & WANG Q 2015. Physico-chemical characterizastion of PM2.5 in the microenvironment of Shanghai subway. Atmospheric Research, 153, 543–552. doi: 10.1016/j.atmosres.2014.10.006 [DOI] [Google Scholar]
  14. LUGLIO DG, KATSIGEORGIS M, HESS J, KIM R, ADRAGNA J, RAJA A, GORDON C, FINE J, THURSTON G, GORDON T & VILCASSIM MJR 2021. PM2.5 concentration and composition in subway systems in the Northeastern United States. Environmental Health Perspectives, 129. doi: 10.1289/EHP7202 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. MA H, SHEN H, LIANG Z, ZHANG L & XIA C Passengers’ Exposure to PM2.5, PM10, and CO2 in Typical Underground Subway Platforms in Shanghai. In: LI A, ZHU Y & LI Y, eds. Proceedings of the 8th International Symposium on Heating, Ventilation and Air Conditioning, 2014//2014 Berlin, Heidelberg. Springer; Berlin Heidelberg, 237–245. [Google Scholar]
  16. MARTINS V, MORENO T, MENDES L, ELEFTHERIADIS K, DIAPOULI E, ALVES CA, DUARTE M, MIGUEL ED, CAPDEVILA M, QUEROL X & MINGUILLÓN MC 2016. Factors controlling air quality in different European subway systems. Environmental Research, 146, 35–46. doi: 10.1016/j.envres.2015.12.007 [DOI] [PubMed] [Google Scholar]
  17. MINGUILLÓN MC, RECHE C, MARTINS V, AMATO F, MIGUEL ED, CAPDEVILA M, CENTELLES S, QUEROL X & MORENO T 2018. Aerosol sources in subway environments. Environmental International, 167, 314–328. doi: 10.1016/j.envres.2018.07.034 [DOI] [PubMed] [Google Scholar]
  18. MORENO T, PÉREZ N, RECHE C, MARTINS V, DE MIGUEL E, CAPDEVILA M, CENTELLES S, MINGUILLÓN MC, AMATO F, ALASTUEY A, QUEROL X & GIBBONS W 2014. Subway platform air quality: Assessing the influences of tunnel ventilation, train piston effect and station design. Atmospheric Environment, 92, 461–468. doi: 10.1016/j.atmosenv.2014.04.043 [DOI] [Google Scholar]
  19. MORENO T, RECHE C, MINGUILLÓN MC, CAPDEVILA M, DE MIGUEL E & QUEROL X 2017. The effect of ventilation protocols on airborne particulate matter in subway systems. Science of The Total Environment, 584–585, 1317–1323. doi: 10.1016/j.scitotenv.2017.02.003 [DOI] [PubMed] [Google Scholar]
  20. PARK D, OH M, YOON Y, PARK E & LEE K 2012. Source identification of PM10 pollution in subway passenger cabins using positive matrix factorization. Atmospheric Environment, 49, 180–185. doi: 10.1016/j.atmosenv.2011.11.064 [DOI] [Google Scholar]
  21. QUEROL X, MORENO T, KARANASION A, RECHE C, ALASTUEY A, VIANA M, FONT O, GIL J, DE MIGUEL E & CAPDEVILA M 2012. Variability of levels and composition of PM10 and PM2.5 in the Barcelona metro system. Atmospheric Chemistry and Physics, 12, 5055–5076. doi: 10.5194/acp-12-5055-2012 [DOI] [Google Scholar]
  22. R CORE TEAM 2019. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. [Google Scholar]
  23. REQUIA WJ, ADAMS MD, ARAIN A, PAPTHEODOROU S, KOUTRAKIS P & MAHMOUD M 2018. Global association of air pollution and cardiorespiratory diseases: A systematic review, meta-analysis, and investigation of modifier variables. American Journal of Public Health, 108, S123–S130. doi: 10.2105/AJPH.2017.303839 [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. SMITH JD, BARRATT BM, FULLER GW, KELLY FJ, LOXHAM M, NICOLOSI E, PRIESTMAN M, TREMPER AH & GREEN DC 2020. PM2.5 on the London Underground. Environment International, 134. doi: 10.1016/j.envint.2019.105188 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. SON Y-S, SALAMA A, JEONG H-S, JEONG JY, LEE J, SUNWOO Y & KIM J-C 2013. The Effect of Platform Screen Doors on PM10 Levels in a Subway Station and a Trial to Reduce PM10 in Tunnels. Asian Journal of Atmospheric Environment, 7, 38–47. doi: 10.5572/ajae.2013.7.1.038 [DOI] [Google Scholar]
  26. SON YS, JEON JS, LEE HJ, RYU IC & KIM JC 2014. Installation of platform screen doors and their impact on indoor air quality: Seoul subway trains. J Air Waste Manag Assoc, 64, 1054–61. doi: 10.1080/10962247.2014.923350 [DOI] [PubMed] [Google Scholar]
  27. THURSTON G, BURNETT RT, TURNER MC, SHI Y, KREWSKI D, LALL R, ITO K, JARRETT M, GAPSTUR SM, DIVER WR & POPE CA III 2016. Ischemic heart disease mortality and long-term exposure to source-related components of U.S. fine particle air pollution. Environmental Health Perspectives, 124, 785–794. doi: 10.1289/ehp.1509777 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. TU M & OLOFSSON U 2021. PM levels on an underground metro platform: A study of the train, passenger flow, urban background, ventilation, and night maintenance effects. Atmospheric Environment: X, 12, 100134. doi: 10.1016/j.aeaoa.2021.100134 [DOI] [Google Scholar]
  29. VAN RYSWYK K, ANASTASOPOLOS AT, EVANS G, SUN L, SABALIAUSKAS K, KULKA R, WALLACE L & WEICHENTHAL S 2017. Metro commuter exposures to particulate air pollution and PM2.5-associated elements in three Canadian cities: The urban transportation exposure study. Environmental Science & Technology, 51, 5713–5720. doi: 10.1021/acs.est.6b05775 [DOI] [PubMed] [Google Scholar]
  30. VILCASSIM MJR, THURSTON GD, PELTIER RE & GORDON T 2014. Black carbon and particulate matter (PM2.5) concentrations in New York City’s subway stations. Environmental Science & Technology, 48, 14738–14745. doi: 10.1021/es50429hl [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. WANG M, AARON CP, MADRIGANO J, HOFFMAN EA, ANGELINI E, YANG J, LAINE A, VETTERLI T, KINNEY PL, SAMPSON PD, SHEPPARD LE, SZPIRO AA, ADAR SD, KIRWA K, SMITH B, LEDERER DJ, DIEZ-ROUX AV, VEDAL S, KAUFMAN JD & BARR RG 2019. Association between long-term exposure to ambient air pollution and change in quantitatively assessed emphysema and lung function. JAMA, 322, 546–556. doi: 10.1001/jama.2019.10255 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. XU B & HAO J 2017. Air quality inside subway metro indoor environment worldwide: A review. Environment International, 2017, 33–46. doi: 10.1016/j.envint.2017.06.016 [DOI] [PubMed] [Google Scholar]
  33. XU D, CHEN Y, WU L, HE S, XU P, ZHANG Y, LUO J, YE X, CHEN Z, WANG X & LOU X 2020. Acute effects of ambient PM2.5 on lung function among schoolchildren. Scientific Reports, 10. doi: 10.1038/s41598-020-61003-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. ZHANG W, ZHAO H, ZHAO A, LIN J & ZHOU R 2019. Current status, challenges and resilient respose to air pollution in urban subway. Atmosphere, 10. doi: 10.3390/atmos10080472 [DOI] [Google Scholar]
  35. ZHAO L, WANG J, GAO HO, XIE Y, JIANG R, HU Q & SUN Y 2017. Evaluation of particulate matter concentration in Shanghai’s metro system and strategy for improvement. Transportation Research Part D: Transport and Environment, 53, 115–127. doi: 10.1016/j.trd.2017.04.010 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1861972-supplement-1.docx (93.2KB, docx)

RESOURCES