PM2.5 exposure assessment of the population in Lower Manhattan area of New York City after the World Trade Center disaster

PM2.5 exposure assessment of the population in Lower Manhattan area of New York City after the World Trade Center disaster

S.P. Nga, , , C. Dimitroulopouloub, A. Grossinhoc, L.C. Chena and M. Kendalla

a -NYU School of Medicine, Department of Environmental Medicine, 57 Old Forge Road, Tuxedo, NY 10987, USA
b -Building Research Establishment Ltd, Garston, Watford, WD25 9XX, UK
c -Imperial College London, South Kensington campus, London SW7 2AZ, UK




Abstract
On 11 September 2001, the explosion and the collapse of the World Trade Center (WTC) Twin Towers in New York City (NYC), USA, generated a massive release of dust and inhalable toxic substances to the atmosphere as a result of the pulverization of various building materials, furniture, and computers. Many concerns were raised as Particulate Matter (PM) levels in Lower Manhattan might not meet the current National Ambient Air Quality Standards (NAAQS) (65 μg m−3). The current study aims to provide a first estimate of the scale of population exposures during this episode. Data collected from existing monitoring stations in September showed the occurrence of a series of high peaks of PM2.5 registered in the Lower Manhattan area after the 11 September. An interpolation technique was used within a Geographical Information Systems (GIS) environment to estimate outdoor PM2.5 concentrations over NYC. Monthly average of 24 h outdoor PM2.5 concentration of Lower Manhattan was 20.2 μg m−3 and did not exceed the NAAQS value. PM2.5 concentrations in indoor micro-environments were simulated by a deterministic micro-environmental model (INTAIR) and linear regression equations. Three typical population groups were identified for the NYC area—home-makers, office/shop-workers, and students/children—and their 12 h nighttime and daytime exposures were estimated from 14 September until the end of September, either as mean exposure (daytime and nighttime) or as exposure weighted by residential population distribution (nighttime only). Average nighttime and daytime exposures of the Lower Manhattan population were calculated to be 37.3 and 23.6 μg m−3, respectively (daily average: 30.45 μg m−3), in which the various group movements and activities, smoking habits of individuals, and special population movements due to access restrictions and risk avoidance behaviors were considered. Within the study period, assuming the real nighttime population distribution followed the residential population pattern, approximate one quarter of the population was exposed to 20–30 μg m−3 PM2.5; one half of the Lower Manhattan population was exposed to 10–20 and 30–60 μg m−3; around one fifth of the population was exposed to over 60 μg m−3 during nighttime. The results indicated that although the outdoor PM2.5 concentration was lower than the NAAQS value, personal exposure levels, which were generally higher than the outdoor PM2.5 concentration, might still be a reason of concern.


1. Introduction
In Manhattan New York City (NYC), on 11 September 2001, the Twin Towers in the World Trade Centre (WTC) complex were hit by two hijacked airplanes and collapsed. Large amounts of gases and particulate matter were released into the atmosphere, exposing the Lower Manhattan population to a large air pollution episode. Particulate matter (PM), consisting of coarse particles (pulverized concrete and other materials) and fine particles (smoke, dust, and soot from combustion) (Claudio, 2001), was considered as causing potentially serious environmental challenges to people living and working in the Lower Manhattan area. Many concerns were raised as PM levels in Lower Manhattan might not meet the current National Ambient Air Quality Standards (NAAQS) (U.S. EPA, 2002a). The aim of this study was to estimate PM2.5 exposures of the populations closest to the WTC site—in Lower Manhattan (below 14th Street)—for the period almost immediately after the event (14 September) until the end of September. This period was selected for reasons discussed later (see Section 3.1).

Exposure assessments are usually carried out by combining directly measured pollutant concentrations with time activities of relevant population groups (Ozkaynak et al., 1996; Abt et al., 2000; Dimitroulopoulou et al., 2001a, b; Burke et al., 2001). In the absence of detailed personal exposure monitoring for this specific period, immediately after the tragic event, modelling may enable us to provide some estimates of mean population exposure. For this purpose, outdoor PM2.5 levels across NYC were interpolated and input to a deterministic micro-environmental model (INTAIR) (Dimitroulopoulou et al., 2001a) to simulate analytically the concentrations in indoor micro-environments (MEs), whereas for some other indoor MEs, linear regression equations (Burke et al., 2001) were used. Estimates began on the 14 September since it was the first day that data points were adequate for interpolating the outdoor PM2.5 levels across NYC. Personal exposures for representative individuals (RIs)—home-maker (HM), office/shop-worker (OW), and student/child (SC)—were estimated, based on indicative activity patterns.

Due to the lack of data to carry out a detailed study using a probabilistic exposure model (INDAIR/EXPAIR, Dimitroulopoulou et al., 2001b, Ashmore et al., 2000), the current work may be considered as a first screening-level estimate of exposure to PM2.5. However, results from this study may be indicative of whether population exposure levels to PM2.5 in Lower Manhattan from 14 to 31 September 2001 are of concern.

2. Methods
2.1. Outdoor PM2.5 concentrations
Continuous outdoor PM2.5 monitoring in NYC (Fig. 1) was carried out by New York State Department of Environmental Conservation (DEC) (12 sites, Twaddell, 2002), Environmental Protection Agency (EPA) (2 sites, U.S. EPA, 2002b), and New York University (NYU) (2 sites, Chen et al., 2002). Data was collected and analyzed to investigate if there was any significant elevations in PM2.5 resulting from WTC explosion and collapse, and if so over what geographical area and period of time. Outdoor PM2.5 monitoring data were mapped using ArcView® GIS (ESRITM). These point data were then interpolated by ArcMapTM using ArcGISTM Geo-statistical Analyst to estimate the outdoor PM2.5 concentrations for 12 h intervals (0600–1800 as daytime and 1800–0600 as nighttime) in each of the 63 census tracts (CTs) in Lower Manhattan. Continuous surfaces were created using a radial basis function. A deterministic method was preferred to create a surface to capture global trends and pick up local variations. The used interpolation method was directly based on the surrounding measured values and required that the surface passed through the measured points (Johnston et al., 2001). Although the number of points to base the interpolation upon was not large, big changes were not expected in the surface values within a short horizontal distance and therefore this method was selected as an exact interpolator. As no major outliers were present in the dataset used, this method presented good results. The estimated ambient PM2.5 concentrations presented a smooth surface distribution across the area interpolated. Outdoor PM2.5 concentrations were then input to the deterministic micro-environmental model (INTAIR) and linear regression equations to estimate the indoor PM2.5 concentrations.


(191K)

Fig. 1. Locations of monitoring stations for PM2.5 concentrations, set up by DEC, EPA, and NYU, on a background of census tracts in NYC.



2.2. Micro-environmental PM2.5 concentrations
The deterministic micro-environmental model (INTAIR) (Dimitroulopoulou et al., 2001a) was used to predict the PM2.5 concentrations and the indoor/outdoor (I/O) ratios, of three indoor MEs—home, office/shop, and classroom—by solving the following differential equation:

dCi/dt=-vd(Ai/Vi)Ci+λrf×Co-λrCi+λt(Cj-Ci)+Qi/Vi, (1)


where Ci is the indoor concentration of the pollutant in ME i; Cj is the indoor concentration of the pollutant in ME j; Co is the outdoor concentration of the pollutant; vd is the deposition velocity of the pollutant; Ai is the surface area of ME i; Vi is the volume of ME i; λr is the air exchange rate between indoors and outdoors; λt is the air exchange rate for transport of pollutants between MEs i and j; f is the building fabric filtration factor; and Qi is the indoor emission rate of the pollutant in ME i.
The simulations were carried out for no-source and source scenarios (cooking, smoking, cooking and smoking) in the home environment and the input parameters were tabulated in Table 1. The parameterization of the model for the home environment, regarding air exchange rates and building characteristics, was also discussed elsewhere (Dimitroulopoulou et al., 2001a and Dimitroulopoulou et al., 2001c). The time resolution was 15 min to include the effect of indoor activity. However, outputs were calculated for a 12 h period to be consistent with the outdoor data resolution.

Table 1.

Parameters used for simulation (INTAIR) for home Parameters Value(s) Source
Building fabric filtration factor 1 Ozkaynak et al. (1996); Thatcher and Layton (1995)
Air exchange rate (AER) 0.54 h−1 Murray and Burmaster (1995)
AER between rooms 1 h−1 (to represent the conditions of open doors) Dimitroulopoulou et al. (2001a)
Surface-to-volume ratio 0.8 U.S. EPA (1999); U.S. EPA (1997)
Deposition velocity 0.39 h−1 Ozkaynak et al. (1996)
Smoking 14 mg PM2.5/cigarette/hour; 16 cigarettes per day Wallace (1996); General Household Survey (1994)
Cooking 1.7 mg min−1; 15 mins for breakfast and 30 mins for dinner Wallace (1996); Ashmore et al. (2000)
Other sources 1.1 mg h−1 Burke et al. (2001)




The input values for office/shop and classroom simulations were tabulated in Table 2. In these analytical simulations, it was assumed that there were no important indoor sources in these MEs. PM2.5 concentrations, in office/shop and classroom, together with other two MEs—transport area and store—were also empirically calculated, for validation reasons, using the linear regression equation

(2)


where b0 is the contribution of PM2.5 sources and b1 is the relationship between micro-environmental and outdoor PM2.5 concentrations (Burke et al., 2001). The input values used in the linear regression equations were tabulated in Table 3.
Table 2.

Parameters used for stimulation (INTAIR) for office/shop and classroom Parameters Office/shop Classroom
Value(s) Source Value(s) Source
Building fabric filtration factor 0.5–0.7 BRE (2002); Burke et al. (2001) 1 Thatcher and Layton (1995); Ozkaynak et al. (1996)
Air exchange rate 0.4 h−1 — 2.24 h−1 Lagus Applied Technology (1995); Daisey and Angell (2002)
Surface-to-volume ratio 0.7 U.S. EPA (1997) 0.7 U.S. EPA (1997)
Deposition velocity 0.39 h−1 Ozkaynak et al. (1996) 0.39 h−1 Ozkaynak et al. (1996)




Table 3.
Input values used for simulation (linear regression equation) for office/shop, classroom, transport area, and store Micro-environments b0 b1
Office/shop 3.6 0.18
Classroom 6.8 0.60
Store 9.0 0.74
Transport area 33.0 0.26




2.3. PM2.5 exposures
Representative individuals (RIs) were selected as typical members to represent the Lower Manhattan population. These were a home-maker (HM), an office/shop-worker (OW), and a student/child (SC). In addition to the estimates of daytime and nighttime mean exposure of RIs (see Section 2.3.1), nighttime exposure assessment was also carried out by a method that considered the geographical variations of PM2.5 concentration and residential population among different parts of Lower Manhattan (see Section 2.3.2). However, daytime exposure assessment could not be carried out by this method due to the lack of data on working/student population and movement patterns during daytime. The consistency of results of the nighttime exposures (of population in home), as estimated by these two methods, could further validate the values estimated by RI exposures for daytime exposures.

2.3.1. Daily mean exposures of representative individuals
The daily mean exposure of RIs were represented as a linear combination of concentrations in distinct MEs (simulated by INTAIR model and linear regression equations), weighted by the time spent in each of these MEs (Duan, 1982). In the lack of detailed U.S. hourly data, Canadian activity pattern were used to estimate the amount of time that RIs spent in each micro-environment (Statistics Canada, 1998), since Americans and Canadians were reported to spend similar amounts of time indoors and outdoors (Klepeis et al., 2001). The indicative activity patterns employed in the simulations (Table 4) were used to estimate average exposures for the RIs and subsequently the Lower Manhattan population (see Section 2.3.3). PM2.5 exposures for the RIs were presented as 12 h averages.

Table 4.

Assumed time-activity pattern of representative individuals (Klepeis et al. 2001; Statistics Canada 1998) Half day Time Home-maker Office/shop-worker Student/child
Night 18:00–06:00 Home Home Home
Day 06:00–08:00 Home Home Home
08:00–09:00 Home Transport area Transport area
09:00–14:00 Home Office Classroom
14:00 Store Office Classroom
15:00 Store Office Outdoor
16:00 Transport area Office Outdoor
17:00 Home Transport area Transport area




2.3.2. Exposure weighted by residential population distribution (nighttime)
PM2.5 concentrations in homes (with cooking activities) with or without smoker(s), in each of the 63 CTs every day, were estimated analytically as discussed before. Residential populations (U.S. Census Bureau, 2000a. and U.S. Census Bureau, 2000b.) in some of CTs were adjusted by an estimated occupancy rate (Ng, 2002) (Table 5), to estimate the real number of occupants in the restricted zones (FEMA, 2001a; City of New York Emergency Mapping Center, 2001) and areas with facility outage (FEMA, 2001b). The restricted zone size varied in different periods and some residents/workers stayed away longer. Therefore, estimates had to be made in order to calculate the total population in Lower Manhattan for different days. Population weighted average exposure (Expp) in homes, over the 12 h nighttime period, was calculated by

(3)


where PFi=Population in CTi/Total population. Since the majority of population was assumed to stay at home in the nighttime, no time factor was included in Eqs. (3) and (4). Probabilities of living in a home with or without smoker(s), reported by Health Canada in 1994 (Ozkaynak, 1998), were used as population factors to calculate the percentage of population exposed or not exposed to smoke in home in each CT. Population weighted average exposure with smoking habit being considered (Expp,s) was calculated by
(4)


where PF1i=People live in a home with smoker(s) in CTi/Total population; PF2i=People live in a home without smoker in CTi/Total population. Time factors were not shown in Eqs. (3) and (4) as it was assumed that the majority of population was in home at night.
Table 5.

Occupancy rates (%) in and out of the restricted areas in different time periods Restricted period and area Sept 11–13 Sept 14–18 Sept 19–26 Sept 27 and afterwards
South of 14 street South of Canal street WTC and nearby WTC site
Day Night Day Night Day Night Day Night
In restricted areas Home 40 20 60 60 100 100
Office/shopa 0 — 0 — 0 —
Classrooma 0 — 0 — 0 —

Out of restricted area Home 100 100 100 100 100 100 100 100
Office/shopa 0 — 50 — 100 — 100 —
Classrooma 0 — 0 — 50 — 50 —
a Assuming the majority staying at home during nighttime.



PM2.5 concentrations in homes (with or without smoker(s)) in 63 CTs were sequenced and put into incremental categories of 10 μg m−3. Population numbers exposed to the level within each category were summed up to represent the numbers of individuals exposed to that concentration range. It is emphasized again that this method was applied to the nighttime exposures only since there was no data on the population distribution during daytime.

2.3.3. Exposure weighted by population represented by RIs (daytime and nighttime)
Numbers of people in Lower Manhattan falling into each of the RI categories were used to estimate average daily exposures of the Lower Manhattan population. Based on census data (U.S. Census Bureau, 2000a. and U.S. Census Bureau, 2000b.) and surveys reporting transport use in NYC (City of New York DOT, 2001; RTHIS, 2000), RIs could be sub-classified to 7 categories: HM, OW (living and working in Lower Manhattan), OW2 (living but not working in Lower Manhattan), OW3 (working but not living in Lower Manhattan), SC (living and studying in Lower Manhattan), SC2 (living but not studying in Lower Manhattan), and SC3 (studying but not living in Lower Manhattan). Special population movements, including restricted access (FEMA, 2001a; City of New York Emergency Mapping Center, 2001), facility outage (FEMA, 2001b), and possible personal preference, were considered in this estimation (Table 5). Although these figures should be treated with caution, they may be used as an indication of population proportions in each category. 12 h (nighttime or daytime) population weighted average exposure of Lower Manhattan population was calculated by

(5)


where PFi=Population represented by RIi/Total population.
3. Results and discussion
3.1. Outdoor PM2.5 concentrations in Lower Manhattan September–December 2001
Fig. 2 showed the hourly PM2.5 concentration monitored in NYC from September to November in 2001. Significant peaks attributable to the WTC collapse were observed from 14 to 18 September. Peaks in PM2.5 concentrations after the end of September were comparable with PM2.5 in previous years and coincided with regional pollution events. Results showed that the impact in September 2001 was not caused by regular regional or seasonal episodes, as the mean outdoor PM2.5 levels in fall 2000 was very similar to that in fall 2001 (Twaddell, 2002) and regional levels outside Lower Manhattan were much lower. The main area of impact was, therefore, in Lower Manhattan, as monitoring instruments in the sites in this area (PS 64, Hunter College, NYU Hospital, and Pace University) recorded higher readings than surrounding monitors (Freshkills, IS 143, and MLib). On the basis of these results, the exposure assessment only included the September period. Outdoor PM2.5 concentrations for the first few days after the WTC disaster (11–13 September 2001) were unknown because some of the monitors had not been set up. Therefore, we only estimated exposures since 14 September. This did not include the highest exposures during the event or the highest exposure groups such as the rescue team and clean-up workers. These exposures did not affect much on the exposure of the whole population of the Lower Manhattan of the month as on 11–14 September, residents/workers/students left the Lower Manhattan area and were not exposed to those high levels of PM2.5.


(89K)

Fig. 2. Hourly PM2.5 concentrations (from selected sites) monitored in New York City from September to November 2001.



Prediction maps for daytime and nighttime outdoor PM2.5 concentrations in NYC from 14 to 30 September 2001 were constructed and presented elsewhere (Ng, 2002). Here, we present, as an illustrative example, the map for 15 September (Fig. 3). This showed that the highest concentrations usually occurred in Lower Manhattan. Mean daytime and nighttime outdoor PM2.5 concentrations of Lower Manhattan (i.e. mean of 63 CTs) on each day during 14–30 September were presented in Fig. 4. Daytime and nighttime concentrations differed considerably during mid September, probably depending on the WTC site activities, including demolition and material removal by trucks. Monthly average of daytime and nighttime outdoor PM2.5 concentrations of Lower Manhattan was 19.3 μg m−3 (SD: 7.7 μg m−3, max: 36.5 μg m−3, min: 6.3 μg m−3) and 21.1 μg m−3 (SD: 7.6 μg m−3, max: 36.9 μg m−3, min: 6.3 μg m−3). Monthly average of 24 h outdoor PM2.5 concentration of Lower Manhattan was 20.2 μg m−3 (SD: 7.1 μg m−3, max: 33.5 μg m−3, min: 6.3 μg m−3). The variations in the above results were among days and the concentrations did not exceed the NAAQS value of 65 μg m−3.


(77K)

Fig. 3. Predicted nighttime 12 h mean outdoor PM2.5 concentrations in NYC on 15 September.




(23K)
Fig. 4. Mean daytime and nighttime outdoor PM2.5 concentrations in Lower Manhattan from 14 to 30 September 2001.



3.2. Simulations of micro-environmental PM2.5 concentrations
Results from the analytical and empirical simulations of the PM2.5 indoor MEs were consistent with values reported in literatures. The 12 h mean PM2.5 concentration and mean I/O ratio in home, derived from the analytical simulations with the INTAIR model, were tabulated in Table 6. The outdoor mean concentrations in the same period were 19.3 μg m−3 (daytime) and 21.1 μg m−3 (nighttime). The no source scenario gave an I/O ratio lower than that (0.7) reported by the PTEAM study for California (Ozkaynak et al., 1996) due to the lower air exchange rate reported in NY. In other scenarios, the smoking activity increased the 12 h home PM2.5 concentrations by 27.0 μg m−3. Each cigarette contributed 1.7 μg m−3 to the 12 h averages. This was similar to the value of 1.5 μg m−3/cigarette reported by the PTEAM study (Wallace, 1996). In the same study, it was reported that cooking contributed between 10 and 20 μg m−3 PM10 to 12 h averages and smaller amounts of PM2.5 (Wallace, 1996). The estimate from this study (14.9 μg m−3) fell into this reported range.

Table 6.

Simulated 12 h mean PM2.5 home concentrations, mean concentration differences from the no source scenario, and mean simulated indoor/outdoor ratios Scenarios Daytime or Nighttime Mean PM2.5 conc. (μg m−3) Concentration differences from the no source scenario (μg m−3) I/O
No source Nighttime 12.24 0.0 0.6
Daytime 11.30
Smoking Nighttime 34.57 27.0 1.9
Daytime 42.92
Cooking Nighttime 26.73 14.9 1.3
Daytime 19.98
Smoking & cooking Nighttime 47.37 35.6 2.3
Daytime 49.32




The 12 h mean PM2.5 concentrations and I/O ratios of all indoor MEs, derived from both analytical and empirical simulations, were presented in Table 7. Concentrations and I/O ratios of office/shop and classroom simulated by INTAIR were very similar to those calculated by linear regression equations used in the Philadelphia study (Burke et al., 2001). I/O ratio of office/shop was also near to I/O ratio of 0.5 recorded in BASE data set (Ligman et al., 2002). I/O ratios for transport area and store were consistent to the mean values considered in INDAIR/EXPAIR modeling framework (Ashmore et al., 2000), which were based on a detailed literature review of European and US studies. In that study, mean travel I/O was 1.5 (range: 0.8 – 5.5), whereas store I/O was 0.9 (range: 0.5–1.3).

Table 7.

12 h PM2.5 concentrations and I/O ratios of micro-environments simulated by analytical and empirical methods Micro-environments Micro-environmental
Nighttime or Daytime INTAIR Linear regression equations
PM2.5 conc. (μg m−3) I/O PM2.5 conc. (μg m−3) I/O
Home (smoking, cooking, other sources) Nighttime 47.37 2.3 — —
Daytime 49.32 2.3 — —
Office/shop (no indoor source) Daytime 7.42 0.4 7.1 0.4
Classroom (no indoor source) Daytime 16.91 0.8 18.6 0.9
Transport area (no indoor source) Daytime — — 38.0 1.9
Store (no indoor source) Daytime — — 23.3 1.2




The lowest I/O ratio of office/shop, among all the MEs, indicated that outdoor pollutants might not be easy to infiltrate through HVAC system and minimal PM2.5 sources were present in non-smoking office buildings (Burke et al., 2001). Store and classroom had higher I/O ratios because outdoor PM2.5 did contribute significantly to indoor PM2.5 concentrations (Burke et al., 2001). The regression equation for transport area incorporated measurements from roadsides and inside vehicles. High I/O ratio of transport area was shown because of the high b0 (contribution of PM2.5 sources) used, which accounted for the elevated concentrations often measured in roadways used (Burke et al., 2001).

3.3. Estimates of PM2.5 exposures
3.3.1. Daily mean exposures of representative individuals
The mean, SD, and minimum/maximum values (among days) of 12 h (daytime and nighttime) outdoor PM2.5 concentrations and exposures of RIs, living in homes with smoker(s) or without smoker, from 14 to 30 September were tabulated in Table 8. Cooking activities were considered in simulations of exposures in home. All representative individuals (RIs) living in homes with smoker(s) were exposed to the same level of PM2.5 during nighttime, as they were assumed to spend time in home, no matter which RI category they were in. This applied to all RIs in homes without smoker. At daytime, home-makers (HM) had higher exposures than students/children (SC) and office/shop-workers (OW), mainly due to the higher PM2.5 concentrations in home than in classroom and office/shop. Exposure of HM in homes with smoker(s) was 20 μg m−3 higher than in homes without smoker and higher than those of OW and SC during daytime because smoking in office/shop and classroom is banned in the US. Nighttime exposure of HM was slightly higher than daytime exposure due to the emissions from cooking. Nighttime exposure of OW and SC was much higher than daytime exposure because ME concentration in home was higher than in office/shop and classroom.

Table 8.

12 h nighttime and daytime outdoor PM2.5 concentrations and exposures of representative individuals in Lower Manhattan, from 14 to 30 September 2001 Outdoor PM2.5 conc. (μg m−3) 12 h PM2.5 Exposures (μg m−3)
HM, NS HM, S OW, NS OW, S SC, NS SC, S
Average 21.1 26.7 47.4
Night time SD 7.6 4.3 4.5
Min 6.3 18.2 39.1
Max 36.9 35.5 56.4

Average 19.3 23.3 43.2 13.3 20.0
Day time SD 7.7 4.5 4.5 3.3 5.5
Min 6.3 15.8 35.8 8.0 10.6
Max 36.5 33.0 52.9 20.0 31.3




The variations of outdoor PM2.5 concentrations, and nighttime and daytime exposures of RIs, living in homes with or without smoker(s), in September were shown in Fig. 5 and Fig. 6, respectively. Nighttime exposure followed more closely the variation of outdoor PM2.5 concentrations than daytime exposure, since people were assumed to be only in the home environment.


(30K)

Fig. 5. Nighttime exposure of all RIs living in homes with or without smoker(s) in Lower Manhattan from 14 to 30 September 2001.




(37K)
Fig. 6. Daytime exposure of Home-makers/workers, Office/shop-workers, and Students/children, living in home with or without smoker(s), in Lower Manhattan from 14 to 30 September 2001.



3.3.2. Exposure weighted by residential population distribution (nighttime)
Outdoor and indoor home PM2.5 concentrations, and the nighttime exposures of population living with or without smoker(s), from 14 to 30 September 2001 were tabulated in Table 9. Mean nighttime exposures of people living in homes with smoker(s) (49.2 μg m−3) were slightly different from PM2.5 concentrations of home with smoker(s) (51.0 μg m−3) (although it was assumed people stayed in home during nighttime) because the exposures were weighted by population variation in 63 CTs while PM2.5 concentrations were not. Mean exposures calculated for the Lower Manhattan residents were roughly the mean of the exposures for the population living in homes with smoker(s) and the exposures for population living in homes without smoker, since roughly half of the population lived with smoker(s).

Table 9.

12 h nighttime outdoor and indoor (homes with or without smoker(s)) PM2.5 concentrations and exposures calculated by the residential population distribution in Lower Manhattan in September 2001 Nighttime PM2.5 concentrations in Mean⁎⁎⁎ exposures of
Outdoor environment Home with smoker(s) Home without smoker Population Smokers⁎ Non-smokers⁎⁎
Mean (μg m−3) 21.1 51.0 29.4 39.0 49.2 28.4
SD 7.6 22.8 14.8 18.8 22.7 14.7
⁎ Smokers=people living in homes with smoker(s).
⁎⁎ Non-smokers=people living in homes without smoker.
⁎⁎⁎ Weighted by variation of outdoor PM2.5 concentration and residential population in 63 CTs.



Fig. 7 showed the variation in nighttime outdoor PM2.5 concentrations, and PM2.5 concentrations and exposures in homes across CTs with smoker(s) and without smoker, during September 2001. As expected, indoors PM2.5 concentrations reflected the variation of outdoor PM2.5 concentrations and were higher than outdoor PM2.5 concentrations due to the presence of indoor sources (cooking and smoking).


(72K)

Fig. 7. Variation of nighttime outdoor and indoor (homes with and without smoker(s)) PM2.5 concentrations and exposures of average population, smokers, and non-smokers across CTs in Lower Manhattam from 14 to 30 September 2001.



Table 10 showed the proportion of Lower Manhattan population exposed to twelve ranges of PM2.5 concentration (with 10 μg m−3 intervals) during the nighttime on 14–30 September 2001. During 14–18 September, the estimated total population in Lower Manhattan was 259,963 while during 19–30 September, the estimated total population in Lower Manhattan was 291,944 (U.S. Census Bureau, 2000a. and U.S. Census Bureau, 2000b.). The largest proportion of the population was exposed to 20–30 μg m−3 PM2.5, followed by 10–20 μg m−3, 30–40 μg m−3, 40–50 μg m−3, 50–60 μg m−3, and 60–70 μg m−3. This even distribution of population exposed to different PM2.5 concentration ranges could be explained by the geographical and temporal variation of micro-environmental PM2.5 among CTs and over time. The variation of average exposures among nights could be explained by the variation of outdoor (and so micro-environmental) PM2.5 concentration and/or the shifting of the highest PM2.5 concentration between highly populated areas and sparsely populated areas.

Table 10.

Percentage of Population distributed into twelve PM2.5 nighttime home concentration ranges PM2.5 Conc. Ranges (μg m−3) Population percentage exposed to relevant PM2.5 concentration range
Date in Sept. Sept
14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 14–30
0–10 11 1 49 3.6
10–20 4 46 16 6 44 27 41 49 51 16.6
20–30 44 27 30 31 9 1 49 37 10 21 24 34 42 51 24.0
30–40 39 10 20 24 21 42 12 26 24 21 10 10 15.2
40–50 13 3 19 24 20 7 51 5 18 21 19 9 35 4 14.6
50–60 3 8 9 19 21 2 21 19 17 7 9 1 7.9
60–70 1 2 3 13 24 15 12 49 9 5 4 0 8.1
70–80 2 6 10 7 3 18 3 3 4 1 3.2
80–90 3 1 4 5 14 5 1 1 1 2.0
90–100 2 2 8 11 0 3 1 1.6
100–110 2 5 40 2 0 2.8
>110 0 4 0 2 2 0.4




3.3.3. Exposure weighted by population represented by RIs (daytime and nighttime)
Nighttime and daytime average exposures of Lower Manhattan population, from 14 to 18 and from 19 to 30 September, were tabulated in Table 11 and Table 12, respectively. Nighttime exposure of the period 14–18 was similar to that during 19–30 September as the PM2.5 concentrations in those two periods were nearly the same. Daytime exposure during 14–18 September was higher than those during 19–30 of the same month as office/shop and classroom were closed and people tended to stay in home where there were higher concentrations. Nighttime exposure for all days was 37.3 μg m−3 that was higher than the daytime exposures of 23.6 μg m−3, because of the cooking and smoking activities in home during nighttime.

Table 11.

Average exposures of population subgroups from 14 to 18 September 2001 RIs 12 h Night 12 h Day
Nighttime exposures (μg m−3) of RIs Exposure×Populations factor Daytime exposures (μg m−3) of RIs Exposure×Population factor
HM (ns) 26.6 9.4 21.8 8.1
HM (s) 46.5 17.1 41.7 16.1
OW (ns) 26.6 2.2 11.7 1.0
OW (s) 46.5 4.0 11.7 1.1
OW2 (ns) (only live in LM) 26.6 1.0 0.0
OW2 (s) (only live in LM) 46.5 1.8 11.7 0.0
OW3 (ns) (only work in LM) 0.0 0.4
OW3 (s) (only work in LM) 0.0 11.7 0.4
SC2 (ns) (only live in LM) 26.6 0.5 0.0
SC2 (s) only live in LM) 46.5 0.9 0.0

Total 36.8 27.0




Table 12.
Average exposure of population subgroups from 19 to 30 September 2001 12 h Night 12 h Day
RIs Nighttime exposures (μg m−3) of RIs Exposure×Population factor Daytime exposures (μg m−3) of RIs Exposure×Population factor
HM (ns) 26.8 5.2 23.9 4.3
HM (s) 47.7 9.6 43.8 8.2
OW (ns) 26.8 5.3 14.0 2.6
OW (s) 47.7 9.8 14.0 2.7
SC (ns) 26.8 1.2 20.9 0.9
SC (s) 47.7 2.3 20.9 0.9
OW2 (ns) (only live in LM) 26.8 0.9 0.0
OW2 (s) (only live in LM) 47.7 1.7 0.0
OW3 (ns) (only work in LM) 0.0 14.0 0.9
OW3 (s) (only work in LM) 0.0 14.0 1.0
SC2 (ns) (only live in LM) 26.8 0.5 0.0
SC2 (s) only live in LM) 47.7 0.9 0.0
SC3 (ns) (only study in LM) 0.0 20.9 0.3
SC3 (s) only study in LM) 0.0 20.9 0.4

Total 37.5 22.2




3.4. Comparison of two methods employed in the exposure assessment
Average and SDs (among days in September) of nighttime PM2.5 exposures from 14 to 30 September estimated by two methods (3.3.2 and 3.3.3) were tabulated in Table 13. The mean exposures estimated by two methods were very close to each other, implying that the latter method (presented in Section 3.3.3) could be a substitute of the former one (presented in Section 3.3.2) when there was an information gap on population distribution, that controlled the number of people exposed to each concentration ranges, during daytime. Higher SDs in the first method indicated that it was more eligible to detect the variation among different days when PM2.5 concentrations contours shifted among CTs, affecting the number of people exposed to different PM2.5 levels and thus the average exposure. The second method was unable to reflect the geographical variation of PM2.5 concentrations and residential population. Even during a coincidence of high PM2.5 concentrations falling into highly populated areas, it gave an ‘averaging’ effect and lowered the peaks because population was assumed to be evenly distributed and outdoor PM2.5 concentration among CTs in Lower Manhattan was averaged before they were input to the INTAIR model and linear regression equations. On the other hand, the second method was able to demonstrate the pattern of the exposures within a day. Although sharp exposures for the Lower Manhattan population on certain days could not be detected, it was still reliable to find the average exposures over a period of time and detailed enough to describe the activities causing higher exposures.

Table 13.

Nighttime exposure assessments, done by two methods Method Relevant section Nighttime PM2.5 exposures
Smokers Non-smokers Population
Employing RIs to represent typical population groups 3.3.3 Mean 47.4 26.7 37.3
SD (among days) 4.5 4.3 —⁎
Consideration of geographical variation of PM2.5 concentrations and population among CTs 3.3.2 Mean 49.2 28.4 39.0
SD (among days) 22.7 14.7 18.8
⁎ SD unavailable as mean exposure of the population was derived by the exposures of RIs weighted by number of people represented by each RI.



4. Conclusions
During the current modeling work, representative individuals were used to represent the Lower Manhattan population and their PM2.5 exposures were estimated during daytime and nighttime, either as mean or as exposure weighted by residential population distribution. Although this study may be described as a screening-level approximation, it provided a first estimate of the exposure to PM2.5 levels in the Lower Manhattan, almost immediately after the disaster of the World Trade Center.

Monthly average of 24 h outdoor PM2.5 concentration of Lower Manhattan was 20.2 μg m−3 (SD: 7.1 μg m−3, max: 33.5 μg m−3, min: 6.3 μg m−3) and did not exceed the NAAQS value of 65 μg m−3. The 12 h daytime exposure was estimated to be high (>40 μg m−3) for the home-makers when smoking activity was assumed in the home environment. The other two representative individuals (office/shop-workers and students/children) had 12 h daytime exposures in the range of 10–20 μg m−3. During nighttime and assuming being at home, the largest population was exposed to PM2.5 levels in the range 20–30 μg m−3, depending on the smoking activity. Only a small part of the population (13%) during this period was exposed to concentrations greater than 65 μg m−3.

The results indicated that although the outdoor PM2.5 concentration was lower than the NAAQS value, personal exposure levels, which were generally higher than the outdoor PM2.5 concentration, might still be a reason of concern. Further work, using probabilistic models properly parameterized to include variability in activity patterns and the other input parameters, is required to validate the above results.


Acknowledgements

The authors wish to thank the NIEHS and Dr. Kenneth Olden for the supplemental funds provided to our NIEHS Center at the NYU School of Medicine (ES00260) to complete these analyses. This study was also funded in part by a U.S. EPA PM Center Grant (R827351).


References
Abt et al., 2000 E. Abt, H.H. Suh, G. Allen and P. Koutrakis, Characterization of indoor particle sources: a study conducted in the metropolitan Boston area, Environmental Health Perspectives 108 (2000) (1), pp. 35–44. Abstract-EMBASE Abstract-Elsevier BIOBASE Abstract-MEDLINE

Ashmore et al., 2000. Ashmore, M.R., Dimitroulopoulou, C., Hill, M.T.R., Byrne, M.A., Kinnersley, R.P., 2000. Modeling of indoor and personal exposures to air pollutants: final contract report to DETR. Contract no. EPG 1/3/110.

BRE, 2002. BRE, 2002. Personal communication regarding the building fabric filtration factor for commercial buildings.

Burke et al., 2001 J.M. Burke, M.J. Zufall and H. Ozkaynak, A population exposure model for particulate matter: case study results for PM2.5 in Philadelphia, PA, Journal of Exposure Analysis and Environmental Epidemiology 11 (2001), pp. 470–489. Abstract-EMBASE Abstract-Elsevier BIOBASE Abstract-MEDLINE Full Text via CrossRef

Chen et al., 2002. Chen, L.C., Thurston, G., Gorczynski, J., Blaustein, M., 2002. Personal communication regarding data collected from NYU monitoring stations.

City of New York Department of Transport, 2001. City of New York Department of Transport, 2001. Manhattan River Crossings. NYC DOT Press Office.

City of New York and Emergency Mapping Center, 2001. City of New York, Emergency Mapping Center, 2001. Prohibited zone shrinkages. Available: http://www.nyc.gov/html/oem/pdf/prohibited_zones/full_view_update_4-1-02.pdf (cited on 15 August 2002).

Claudio, 2001 L. Claudio, Environmental aftermath, Environmental Health Perspectives 109 (2001) (11), pp. A529–A536.

Daisey and Angell, 2002. Daisey, J., Angell, W., 2002. Indoor air quality, ventilation and health symptoms in schools: an analysis of existing information. Indoor Air Conference 2002.

Dimitroulopoulou et al., 2001a C. Dimitroulopoulou, M.R. Ashmore, M.A. Byrne and R.P. Kinnersley, Modeling of indoor exposure to nitrogen dioxide in the UK, Atmospheric Environment 35 (2001), pp. 269–279. Abstract-EMBASE Abstract-Compendex

Dimitroulopoulou et al., 2001b. Dimitroulopoulou, C., Ashmore, M.R., Byrne, M.A., Hill, M.T.R., 2001b. Modeling of indoor and personal exposures to air pollutants. Proceedings of Urban Air Quality Third International Conference, Loutraki, Greece, 19–23 March 2001.

Dimitroulopoulou et al., 2001c C. Dimitroulopoulou, M.R. Ashmore and M.A. Byrne, Modeling the contribution of passive smoking to PM10 in UK homes, Indoor and Built Environment 10 (2001) (3–4), pp. 209–213. Abstract-EMBASE Full Text via CrossRef

Duan, 1982 N. Duan, Micro-environment types: a model for human exposure to air pollution, Environment International 8 (1982), pp. 305–309. SummaryPlus Full Text + Links PDF (460 K)

Federal Emergency Management Agencies and ITS Mapping and Analysis Center, 2001a. Federal Emergency Management Agencies, ITS Mapping and Analysis Center, 2001a. Major crossing status. Available: http://www.gismaps.fema.gov/2001graphics/dr1391/travel_restrictions09_19.jpg (cited on 15 August 2002).

Federal Emergency Management Agencies and ITS Mapping and Analysis Center, 2001b. Federal Emergency Management Agencies, ITS Mapping and Analysis Center, 2001b. Utility outages. Available: http://www.gismaps.fema.gov/2001graphics/dr1391/power_outages09_19.jpg (cited on 15 August 2002).

General Household Survey, 1994. General Household Survey, 1994. Living in Britain. OPCS.

Johnston et al., 2001. Johnston, K., Ver Hoef, J.M., Krivoruchko, K., Lucas, N., 2001. Using Arc GISTM Geostatistical Analyst. ESRITM.

Klepeis et al., 2001 N.E. Klepeis, W.C. Nelson, W.R. Ott, J.P. Robinson, A.M. Tsang, P. Switzer, J.V. Behar, S.C. Hern and W.H. Engelmann, The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants, Journal of Exposure Assessment and Environmental Epidemiology 11 (2001), pp. 231–252. Abstract-EMBASE Abstract-Elsevier BIOBASE Abstract-MEDLINE Full Text via CrossRef

Lagus Applied Technology, 1995. Lagus Applied Technology, 1995. Air change rates in non-residential buildings in California. California Energy Commission, P400-91-034BCN.

Ligman et al., 2002. Ligman, B., Casey, M., Braganza, E., Coy, A., Redding, Y., Womble, S., 2002. Airborne particulate matter within school environment in the US. Indoor Air Conference 2002.

Murray and Burmaster, 1995 D.M. Murray and D.E. Burmaster, Residential air exchange rates in the US: empirical and estimated parametric distributions by season and climatic region, Risk Analysis 15 (1995) (4), pp. 459–465. Abstract-EMBASE Abstract-Compendex

, 2002. Ng, S.P., 2002. Exposure Assessment of the population of New York City to Fine Particulate Matter after the Twin Towers disaster. Dissertation for completion of a master degree in Imperial College London.

Ozkaynak, 1998 H. Ozkaynak, Modeling population exposures to particulate matter. Presented at Particulate methodology workshop, National research center for statistics and the environment, Seattle Washington (1998).

Ozkaynak et al., 1996 H. Ozkaynak, J. Xue, J. Spengler, L. Wallace, E. Pellizzari and P. Jenkins, Personal exposure to airborne particles and metals: results from the Particle TEAM study in Riverside, California, Journal of Exposure Analysis and Environmental Epidemiology 6 (1996) (1), pp. 57–78. Abstract-EMBASE Abstract-MEDLINE

Regional Travel—Household Interview Survey, 2000. Regional Travel—Household Interview Survey, 2000. Executive summary: general final report for the RT-HIS. Parsons Brinckerhoff Quade & Douglas, Inc. in association with Cambridge Systematics, Inc. and NuStats International.

Statistics Canada, 1998. Statistics Canada, 1998. Overview of the time use of Canadians in 1998. Statistics Canada.

Thatcher and Layton, 1995 T.L. Thatcher and D.W. Layton, Deposition, resuspension, and penetration of particles within a residence, Atmospheric Environment 29 (1995) (13), pp. 1487–1497. SummaryPlus Full Text + Links PDF (904 K)

Twaddell, 2002. Twaddell, R., 2002. Personal communication regarding data collected from DEC monitoring stations.

U.S. Census Bureau, 2000a. U.S. Census Bureau, 2000a. 2000 Census DP-2, DP-3, DP-4 Profiles. Population Division—NYC Department of City Planning.

U.S. Census Bureau, 2000b. U.S. Census Bureau, 2000b. 2000 Census SF 1. US Census Bureau.

EPA, 1997. U.S. EPA, 1997. Exposure Factors Handbook. National Center for Environmental Assessment. Available: http://www.epa.gov/ncea/pdfs/edfs/efh/sect17.pdf (cited on 19 August 1999).

EPA, 1999. U.S. EPA, 1999. THERdbASE software system on the internet. Available: http://www.epa.gov/nerl/heasd/therdbase.htm (cited on 19 August 1999).

EPA, 2002a. U.S. EPA, 2002a. National ambient air quality standards. Available: http://www.epa.gov/airs/criteria.html (cited on 10 September 2002).

EPA, 2002b. U.S. EPA, 2002b. PM2.5 in air. Available: http://www.epa.gov/wtc/pm25/index.html (cited on 09 August 2002).

Wallace, 1996 L. Wallace, Indoor particle: a review, Air and Waste Management Association 46 (1996), pp. 98–126.





Atmospheric Environment
Volume 39, Issue 11 , April 2005, Pages 1979-1992