Risk Assessment for People Exposed to PM 2.5 and Constituents at Di ﬀ erent Vertical Heights in an Urban Area of Taiwan

: Environmental epidemiological studies have consistently reported associations between ambient particulate matter (PM) concentrations and everyday mortality / morbidity. Many urban dwellers in Asia live in high-rise apartment buildings; thus, the pollutant concentrations of their immediate outdoor environments are a ﬀ ected by the vertical distribution of pollutants in the atmosphere. The vertical distributions of pollutants provide unique information about their sources and dynamic transport in urban areas, as well as their relationship to people’s exposure at ground level, while the vertical distributions of pollutants have rarely been considered in exposure assessment. In the current study, PM concentrations (with aerodynamic diameters less than 1.0 µ m (PM 1 ), 2.5 µ m (PM 2.5 ), and 10 µ m (PM 10 )), nanoparticles, black carbon (BC), and particle-bound polycyclic aromatic hydrocarbons (p-PAHs) were measured at di ﬀ erent residential heights—6 m, 16 m, and 27 m—at Feng Chia University near a popular night market in Western Taiwan. PM 2.5 data were further adopted for health risk estimations. In winter, the magnitude of PM 1 , PM 2.5 , and PM 10 concentrations were 16 m > 6 m > 27 m; nanoparticle concentrations were 6 m > 27 m > 16 m; and BC and p-PAHs concentrations were 27 m > 16 m > 6 m. In summer, PM 1 , PM 2.5 , and PM 10 concentrations ranged from 6 m > 16 m > 27 m; nanoparticle concentrations were 6 m > 16 m; and BC and p-PAHs concentrations were from 27 m > 16 m. PM and constituents concentrations during winter were signiﬁcantly higher in the nighttime than those in daytime, and levels of PM 1 , PM 2.5 , and PM 10 increased rapidly on 6 m and 16 m heights (but did not increase at 27 m) after 5 pm, whereas these trends became less signiﬁcant in summer. Health risk analysis for PM 2.5 concentrations showed a decrease in lung cancer mortality rate and an extended lifespan for residents living at 27 m. Overall, the current study investigated the vertical proﬁle of particulate matters and analyzed health impacts of PM 2.5 at di ﬀ erent residential heights in urban area of Taiwan. As the distributions of PM and the constituents varied at di ﬀ erent residential heights, exposure and risk assessment of particle concentrations with multiple sizes and various components at broader vertical heights should be further investigated. polycyclic aromatic


Introduction
As economic development increases, urban air quality decreases because of the increasing air pollution produced by industries, automobiles, and other human activity-based emission sources. In 2013, outdoor air pollution, especially outdoor particulate matter (PM), was classified as carcinogenic to humans (Group 1) by the International Agency for Research on Cancer (IARC) of the World Health Organization (WHO). Global air pollution reports showed that over 95% of the world's population lived in areas that exceeded the air quality guideline of PM with aerodynamic diameter less than 2.5 µm (PM 2.5 ; fine particulate matter), and that nearly 60% lived in areas that did not meet even the least stringent air quality target from WHO [1]. Air pollution from PM, ozone (O 3 ), and household burning of solid fuels might have contributed to 6.1 million deaths (11% of the global total) in 2016 [1].
In addition to the attention to the health hazards of particles with different sizes, the composition of PM varies; it is formed through complex chemical reactions by different pollutants emitted from fossil fuel power plants, industries, and automobiles. Black carbon (BC), polycyclic aromatic hydrocarbons (PAHs), and heavy metals are also constituents of PM, which are mainly derived from traffic emissions, activities of combustion, and anthropogenic as well as biomass burning emissions [2][3][4].
The health effects of exposure to PM have been getting great attention. A time-series analysis showed that, in China, a 10 µg/m 3 increase in PM 2.5 levels was associated with 0.65%, 0.63%, and 1.38% increases in non-accidental mortality, respiratory mortality, and circulatory mortality, respectively, during 2005-2009 [5]. In Spain, the relative risks (RR) for an increase level of 25 µg/m 3 in PM 2.5 were 1.057 for all-cause mortality, 1.088 for circulatory mortality, and 1.122 for respiratory mortality [6]. Pope et al. reported that a 10 µg/m 3 elevation of fine particulate in the United States was associated with approximately 4%, 6%, and 8% increases in the risk of all-cause, cardiopulmonary, and lung cancer mortality, respectively [7]. The toxicity potential of PM on respiratory health are in fact based on their residence time in the different regions of the airway [8]. Pan et al. reported that long-term exposure to PM 2.5 and low-grade inflammation might lead to hepatic tumorigenesis and increase the risk of liver cancer [9]. Moreover, exposure to PAHs would cause different impacts on human health [10]. With long-term exposure, effects included skin, lung, bladder and gastrointestinal cancers, DNA damage, cataracts, kidney and liver damage, gene mutation cell damaging, and cardiopulmonary mortality [11][12][13][14][15][16][17][18][19]. As a result, the risks of exposure to PM 2.5 and PAHs should be of great concern.
On the other hand, the vertical comparison of PM 2.5 and PM 10 (particulate matter with an aerodynamic diameter less than 10 µm) concentrations were evaluated in previously reports [20][21][22]. Kalaiarasan et al. measured PM 2.5 levels at various heights of two typical types of buildings, point block configuration and slab block configuration, near a busy expressway in Singapore. For both buildings, the highest concentration of PM 2.5 was at the mid-floors, and the lowest was at upper floors, which might be due to interception by tree leaves or the mixing effect by the inflow of clean air from higher altitude. For point block and slab block, the health risk values at the mid and lower floors suggested that occupants living in these floors experienced 1.62-1.81 and 1.28-1.34 times of risk, respectively, compared to those living at the upper floors [21]. The vertical comparison of PM 2.5 and PM 10 concentrations was also studied in Osaka, Japan [22]. The concentrations of samples collected at 200 m were lower than those collected on the ground. In summer, the high concentration on the ground might be because particles absorbed water when relative humidity (RH) was high. In winter, the accumulation of particles might be due to stable meteorological conditions and vehicle exhaust on the ground. Despite these findings, the previous studies were conducted in relatively high-rise buildings, while most people in Taiwan lived at the height of life circumstances relatively lower than the heights in the aforementioned studies. In fact, more than 52% of the buildings in Taiwan were less than 30 m height [23]. Therefore, it can be suspected that PM concentrations at different vertical heights might vary in different seasons. The objectives of this study are to investigate the particle distribution at different residential heights in different seasons, where the results will be further taken into consideration for health risk assessment at different residential heights.

Field Measurement
Taichung is a special municipality located in central-western Taiwan. Metropolitan Taichung covers an area of 2215 km 2 and hosts a population of over 2.8 million people, making it the second largest city on the island [24]. In recent years, the city ran into further climatic anomalies, reducing the number of rainy days and increasing temperatures, providing the appropriate conditions for a chain reaction to occur that contributes to a worsening of the air quality [25]. The field sampling (including instrument collocation) was carried out during the summer and winter of 2014. Outdoor air pollutant concentrations were measured on different floors in a building at Feng Chia University in Taichung. There was a big night market about 100 m west of the building. Vertical sampling at different residential heights was done simultaneously for 10 days from 23 February to 6 March, as well as from 18 to 28 July 2014. Sampling was conducted from the balconies of 2nd floor (6 m), the 5th floor (16 m), and the 8th floor (27 m). Before the vertical sampling began, three sets of identical instruments were set up on the top floor for simultaneous measurement in order to compare and adjust the data collected from the three identical instruments. Instrument collocation was performed for three days in winter and in summer separately.

Quality Control
Quality control procedures were performed each sampling day. In addition to the routine calibration and maintenance of the instruments used in this study, the research staff performed essential calibration for the instrument readings and flows before and after each monitoring. To ensure the quality of data processing, Microsoft Excel (2010) was used for data management, descriptive analysis, and to calculate health risks of pollutant exposures. Zero, negative, missing, and unreasonably high and low values were excluded, as were continuous values in a range more than ten folds.

Data Analysis
Simple linear regression analysis was performed to compare concentrations obtained in field collocation and vertical sampling. For a given pollutant, the coefficient of determination (R 2 ) of the identical instruments was determined. The ratio of means between collocating instruments was applied as an adjustment factor to correct for the measured levels. After corrections were made for the testing methods, intraclass correlation coefficients (ICC) were computed to verify the agreement between the collocating measurements. The value of ICC > 0.75 indicates excellent reliability, 0.4 < ICC < 0.75 indicates good reliability, and ICC < 0.4 indicates poor reliability for the comparison among different instruments [26]. Additionally, paired t-tests were also used to investigate the concentration differences between different residential heights during different seasons. Statistical analyses were performed using JMP (version 5.0) and STATISTICA (version 7.1) software.
Atmosphere 2020, 11, 1145 4 of 16 In the current study, the elevated risks suggested by previous studies were adopted for estimating the excess risk of lung cancer mortality rate for residents living on different floors with different PM 2.5 exposures [7,[27][28][29][30][31]. The PM 2.5 levels measured at the height of 6 m were set as the baseline to calculate the deviation of the mortality rate of lung cancer caused by PM 2.5 exposures at different residential heights. Excess lung cancer mortality rate and life expectancy change were calculated by Equations (1) and (2), respectively: Excess lung cancer mortality rate = Vertical deviation of PM 2.5 concentration * 0.1 * Elevated lung cancer risk per 10 µg/m 3 increase of PM 2.5 * Lung cancer mortality rate, where the vertical deviation of PM 2.5 concentrations (16 m, 27 m) was compared with the levels measured at 6 m, the elevating lung cancer risk of 12% per 10 µg/m 3 increase of PM 2.5 was adopted from Huang et al., and the lung cancer mortality rate in Taiwan was 60.59 per 100,000 in 2017 [30,32].
Life expectancy change = Vertical deviation of PM 2.5 concentration * 0.1 * Loss of life year per 10 µg/m 3 increase of PM 2.5 * 365, where the vertical deviation of PM 2.5 concentrations (16 m, 27 m) was compared with the levels measured at 6 m, the loss of 0.61 life year per 10 µg/m 3 increase of PM 2.5 was adopted from Pope et al. [7,33].
As the excess mortality rate of lung cancer was estimated mainly by two factors-vertical PM 2.5 concentrations and the RRs of lung cancer mortality in different studies-the software @Risk (version 7.6) was applied for evaluating the sensitivity of the risk estimation. PM 2.5 concentrations in different seasons (winter and summer) were combined together for annual analysis at different residential heights, and the extreme high or low values out of 1% (±2.576 standard deviation) were excluded for truncating the reliable results of PM data [34].

Collocation Comparison
Results of the collocation comparison were summarized in Table 1. For PM 1 , PM 2.5 , and PM 10 measurements, R-squared and ICC values were all greater than 0.9 and 0.85, respectively, indicating the measurement agreement between the identical sampling instruments were in high reliability. The concentrations of nanoparticle, temperature, and RH also showed high ICCs in both winter and summer. Meanwhile, for BC, one of the instruments presented poor reliability in summer (ICC less than 0.4 for serial No. 580), as well as one p-PAHs instrument showed a poor ICC (serial No. 210 in summer), as a result, the corresponding sampling data were excluded from further analyses.

PM Concentrations at Different Residential Heights in Different Seasons
The vertical concentrations of PM pollutants at different residential heights in winter and summer are shown in Table 2 and Figure A1 in Appendix A. According to the Taiwan Environmental Protection Administration's (Taiwan EPA) Ambient Air Quality Standards, the 24 h standard value of PM 10 is 100 µg/m 3 and the 24 h standard value of PM 2.5 is 35 µg/m 3 [35]. Given this, the average concentrations of PM 2.5 at 6 m and 16 m in the winter exceeded the standard value in the current study for administrative purposes. Current results of PM 2.5 concentrations were comparable to the findings of previous research in Taichung [36]. Comparing the monitoring data in different seasons, the levels of PM 1 , PM 2.5 , PM 10 , nanoparticle, and BC were significantly higher in winter than those in summer. In winter, the vertical concentrations of PM 1 , PM 2.5 and PM 10 were 16 m > 6 m> 27 m, and nanoparticle concentrations were 6 m > 27 m > 16 m. For BC and p-PAHs concentrations, both were in range as 27 m > 16 m > 6 m, results of which were consistent as the finding in Beijing, China where the concentrations of BC significantly correlated with those of PAHs [37]. In summer, the concentrations of PM 1 , PM 2.5 , and PM 10 were in range as 6 m > 16 m > 27 m, where the result of a decreasing trend with increasing height was consistent with the concentration distribution of PM 2.5 and its chemical compositions in Tianjin, China [38]. Finally, BC and p-PAHs concentrations ranged from 27 m > 16 m.
The proportion of PM 1 (or PM 2.5 ) in PM 10 at different heights were presented in Figure 1. The similar ranges were found in both seasons for PM 1 /PM 10 (between 45% to 65%) and PM 2.5 /PM 10 (66% to 77%), the results of which were comparable with the daytime averaged PM 1 and PM 2.5 contributing 51-60% and 66-67%, respectively, of PM 10 at the roadside in Macao, China [39]. Furthermore, current results indicated that the proportions of different size PM did not vary so much at different residential heights, therefore the components of PM should play a critical role for health risk estimation.

Temporal Patterns of PM Concentrations at Different Residential Heights
In order to investigate the temporal variation of PM concentrations, the hourly averaged concentrations within a day in winter and summer were calculated (as shown in Figures 2 and 3, respectively). In winter, the levels of PM 1 , PM 2.5 , and PM 10 at 27 m height were apparently lower than those of the other floors, and the PM 2.5 concentrations in the nighttime were higher than those measured at daytime. Furthermore, after 5 pm, levels of PM 1 , PM 2.5 , and PM 10 increased rapidly on 6 m and 16 m heights, but did not increase at 27 m. These results indicated that people living above 27 m might be exposed to less PM from local emission sources. Likewise, the concentrations of nanoparticle presented a similar temporal distribution at different heights, suggesting a well-mixed finer particle size distribution in vertical diffusion [40,41]. For p-PAHs, the concentrations rose at 8 am, but did not considerably increase after 5 pm, results of which showed that emissions from vehicles might be the major source for p-PAHs while the night market activities might be minor in this study. Quang et al. have reported that particle number (PN) and PM 2.5 concentration was influenced by vehicle emissions and new particle formation around the building [42]. Concentrations of PN in size range of < 30 nm and total PN increased (7-65% and 5-46%, respectively), while PM 2.5 concentration decreased (36-52%) with height.
The proportion of PM1 (or PM2.5) in PM10 at different heights were presented in Figure 1. The similar ranges were found in both seasons for PM1/PM10 (between 45% to 65%) and PM2.5/PM10 (66% to 77%), the results of which were comparable with the daytime averaged PM1 and PM2.5 contributing 51-60% and 66-67%, respectively, of PM10 at the roadside in Macao, China [39]. Furthermore, current results indicated that the proportions of different size PM did not vary so much at different residential heights, therefore the components of PM should play a critical role for health risk estimation.

Temporal Patterns of PM Concentrations at Different Residential Heights
In order to investigate the temporal variation of PM concentrations, the hourly averaged concentrations within a day in winter and summer were calculated (as shown in Figures 2 and 3, respectively). In winter, the levels of PM1, PM2.5, and PM10 at 27 m height were apparently lower than those of the other floors, and the PM2.5 concentrations in the nighttime were higher than those measured at daytime. Furthermore, after 5 pm, levels of PM1, PM2.5, and PM10 increased rapidly on 6 m and 16 m heights, but did not increase at 27 m. These results indicated that people living above 27 m might be exposed to less PM from local emission sources. Likewise, the concentrations of nanoparticle presented a similar temporal distribution at different heights, suggesting a well-mixed Atmosphere 2020, 11, x FOR PEER REVIEW 7 of 16 finer particle size distribution in vertical diffusion [40,41]. For p-PAHs, the concentrations rose at 8 am, but did not considerably increase after 5 pm, results of which showed that emissions from vehicles might be the major source for p-PAHs while the night market activities might be minor in this study. Quang et al. have reported that particle number (PN) and PM2.5 concentration was influenced by vehicle emissions and new particle formation around the building [42]. Concentrations of PN in size range of < 30 nm and total PN increased (7-65% and 5-46%, respectively), while PM2.5 concentration decreased (36-52%) with height.
In summer, PM2.5 concentrations in the daytime were more variable and higher than those at night, which might be due to intense human activities, like vehicle emission, and suitable meteorological condition in the daytime [38]. The concentrations of PM2.5 and PM10 were relatively high at 6 pm and 10 pm on the 6 m height, while the pattern did not hold at the higher floors, which implied that some coarser PM sources might exist near the ground. In addition, the temporal trend and levels of nanoparticle were similar at different heights, except for those at 27 m in the night. BC and p-PAHs concentrations showed varying fluctuations at different heights, especially for those at 27 m, which in part might be attributed to the semivolatile characteristic of p-PAHs.    Figure 4 presented diurnal averaged concentrations of PM at different heights. In winter, PM1, PM2.5, PM10, nanoparticle, BC, and p-PAHs were significantly higher in the nighttime than those in daytime, whereas these trends became less apparent or even opposite in summer. As wintertime temperature was lower than that in summer, especially at night, the height of atmospheric mixing layer was closer to the ground surface, which resulted in a poor diffusion vertically and accompanied with the accumulation of air pollutants in the night [43][44][45]. In contrast, the temperature variation was less noticeable in summer; hence the diurnal differences of PM pollutants did not show statistical significance. Using an unmanned aerial vehicle, Lu et al. showed that the PM2.5 concentrations at 300-1000 m altitude in morning flights were generally lower than those in afternoon ones, while the variations of PM2.5 in morning flights were much higher than those in afternoon ones [46]. Besides, in a study of 31 Chinese provincial capital cities, the correlation coefficients between PM2.5 and nitrogen dioxide (NO2) (or sulfur dioxide (SO2)) were either high or moderate based on data from 286 monitoring sites [47]. Previous studies have also reported that PM2.5 was associated with PAHs, heavy metals, and other gaseous pollutants, which implied that if the residents were exposed to high levels of PM2.5, they might be exposed to other pollutants, such as NO2 and SO2, at the same time [1,20,48]. In summer, PM 2.5 concentrations in the daytime were more variable and higher than those at night, which might be due to intense human activities, like vehicle emission, and suitable meteorological condition in the daytime [38]. The concentrations of PM 2.5 and PM 10 were relatively high at 6 pm and 10 pm on the 6 m height, while the pattern did not hold at the higher floors, which implied that some coarser PM sources might exist near the ground. In addition, the temporal trend and levels of nanoparticle were similar at different heights, except for those at 27 m in the night. BC and p-PAHs concentrations showed varying fluctuations at different heights, especially for those at 27 m, which in part might be attributed to the semivolatile characteristic of p-PAHs. Figure 4 presented diurnal averaged concentrations of PM at different heights. In winter, PM 1 , PM 2.5 , PM 10 , nanoparticle, BC, and p-PAHs were significantly higher in the nighttime than those in daytime, whereas these trends became less apparent or even opposite in summer. As wintertime temperature was lower than that in summer, especially at night, the height of atmospheric mixing layer was closer to the ground surface, which resulted in a poor diffusion vertically and accompanied with the accumulation of air pollutants in the night [43][44][45]. In contrast, the temperature variation was less noticeable in summer; hence the diurnal differences of PM pollutants did not show statistical significance. Using an unmanned aerial vehicle, Lu et al. showed that the PM 2.5 concentrations at 300-1000 m altitude in morning flights were generally lower than those in afternoon ones, while the Atmosphere 2020, 11, 1145 9 of 16 variations of PM 2.5 in morning flights were much higher than those in afternoon ones [46]. Besides, in a study of 31 Chinese provincial capital cities, the correlation coefficients between PM 2.5 and nitrogen dioxide (NO 2 ) (or sulfur dioxide (SO 2 )) were either high or moderate based on data from 286 monitoring sites [47]. Previous studies have also reported that PM 2.5 was associated with PAHs, heavy metals, and other gaseous pollutants, which implied that if the residents were exposed to high levels of PM 2.5 , they might be exposed to other pollutants, such as NO 2 and SO 2 , at the same time [1,20,48].

Excess Lung Cancer Risk and Loss of Life by PM2.5 Exposure
As of today, the impact of season on the associations between air pollution and mortality were found to vary in different researches [49,50]. We therefore combined all of the winter and summer data for excess risk estimation. In the current study, the elevated risks reported by Pope et al. were first adopted for estimating the excess risk, with the PM2.5 level at 6 m height serving as the baseline to estimate life expectancy changes due to PM2.5 exposure at different heights [7,33]. As shown in Table 3  The relationship between exposure to PM2.5 and lung cancer mortality had been documented worldwide. For qualitative sensitivity analysis in discussing whether the quantitative relative risk or

Excess Lung Cancer Risk and Loss of Life by PM 2.5 Exposure
As of today, the impact of season on the associations between air pollution and mortality were found to vary in different researches [49,50]. We therefore combined all of the winter and summer data for excess risk estimation. In the current study, the elevated risks reported by Pope et al. were first adopted for estimating the excess risk, with the PM 2.5 level at 6 m height serving as the baseline to estimate life expectancy changes due to PM 2.5 exposure at different heights [7,33]. As shown in Table 3, people living at 16 m height might lose 4.5 days of their life as comparing with the life expectancy of the residents living at 6 m height, and might also increase 0.15 per 100,000 for mortality rate of lung cancer. Additionally, the residents living at 27 m might extend 86.8 days of their life and decrease 2.84 per 100,000 in mortality rate of lung cancer due to less PM 2.5 exposures. Even though people in Taiwan have relatively long lifespan, with a life expectancy of 80.39 years of age and the standardized mortality rate of lung cancer 60.59 per 100,000 in 2017, the potential health risk from air pollution still could not be neglected [32,51]. The relationship between exposure to PM 2.5 and lung cancer mortality had been documented worldwide. For qualitative sensitivity analysis in discussing whether the quantitative relative risk or PM 2.5 levels is the more important factor in influencing PM 2.5 exposure on lung cancer risk at different vertical heights, various associations between PM 2.5 and lung cancer mortality were adopted from results of previous studies [7,[27][28][29][30][31]. In the United States (U.S.), Pope et al. have shown that the elevation of 10 µg/m 3 in PM 2.5 concentration was associated with approximately an 8% increase for lung cancer mortality, and a decrease of 10 µg/m 3 of fine particulate matter was associated with an estimated increase life expectancy of 0.61 ± 0.20 year [7,33]. Besides, the RRs of lung cancer mortality with a 10 µg/m 3 increase in PM 2.5 concentrations were 1.20 (95% confidence interval (CI): 0.79, 1.82) in Los Angeles, U.S. and 1.08 (95% CI: 1.07, 1.09) in China [29,52] A meta-analysis of 17 studies showed that the RR of lung cancer per 10 µg/m 3 increase in PM 2.5 exposure was 1.09 (95% CI: 1.03, 1.15) in Asia; in subgroup analyses for country, the mortality meta-estimate for developed countries was 1.14 (95% CI: 1.06, 1.23), and for developing countries was 1.03 (95% CI: 1.00, 1.07) [30]. Other meta-analysis studies also demonstrated comparable RRs of 1.09 (95% CI: 1.04, 1.14), 1.09 (95% CI: 1.06, 1.11), and 1.14 (95% CI: 1.07, 1.21) for lung cancer associated with PM 2.5 [27,28,31]. Lately, the higher risk of lung cancer with increased PM 2.5 exposure was found to be more significant for female, and the carcinogenic effects of PM 2.5 have urban/rural differences as well as sex differences [53,54].
Results of qualitative sensitivity analysis showed that variations of vertical PM 2.5 concentrations were dominant for excess lung cancer risk estimation in different scenarios, which might be attributed to the wide seasonal variation of vertical PM 2.5 levels. Variation of PM 2.5 at 16 m height as compared to those at 6 m drove the final distribution of excess lung cancer risk at 16 m, accounting for 74.5% of the variance, while the RRs of lung cancer from different studies accounted for 25.5% (Figure 5a). At 27 m height, deviation of PM 2.5 concentrations and the lung cancer RRs from different studies explained 88.6% and 12.8% of the variance, respectively (Figure 5b). Findings of the current study demonstrated that the representativeness of PM 2.5 monitoring was crucial for estimating the relative risk in alternative heights, and the lung cancer slope factor only had a relatively small influence on the results. Therefore, how to acquire more detailed and representative concentrations at different residential heights warrants further studies.
Results of qualitative sensitivity analysis showed that variations of vertical PM2.5 concentrations were dominant for excess lung cancer risk estimation in different scenarios, which might be attributed to the wide seasonal variation of vertical PM2.5 levels. Variation of PM2.5 at 16 m height as compared to those at 6 m drove the final distribution of excess lung cancer risk at 16 m, accounting for 74.5% of the variance, while the RRs of lung cancer from different studies accounted for 25.5% (Figure 5a). At 27 m height, deviation of PM2.5 concentrations and the lung cancer RRs from different studies explained 88.6% and 12.8% of the variance, respectively (Figure 5b). Findings of the current study demonstrated that the representativeness of PM2.5 monitoring was crucial for estimating the relative risk in alternative heights, and the lung cancer slope factor only had a relatively small influence on the results. Therefore, how to acquire more detailed and representative concentrations at different residential heights warrants further studies.  As there is no known level of pollution exposure that is harmless, strategies to minimize daily exposure are often impactful across a good range of settings. In Toronto, Canada, Zuraimi and Tan reported that tighter building envelope, use of mechanical ventilation systems, and improved filtration are key factors that reduce PM2.5 exposure and its impacts on population health [55]. In fact, the indoor/outdoor ratios of PM concentration vary considerably due to the different-sized particles with different emission rates, the cracks and leaks in the building envelope, and the air exchange rates [56]. A study in Durbin, Ireland suggested that increasing air exchange between the outdoors and indoors during nighttime periods could flush out air pollutants [57]. Meanwhile, in the current study, as the levels of PM pollutants were higher at night than in the day, the proposal might not be applicable in Taiwan. Some qualified evidence supports individual actions to reduce cardiopulmonary health risks from exposures to ambient air pollutants by staying indoors and limiting physical exertion when air pollutant concentrations exceed levels associated with increased risk of acute and chronic health problems [58,59]. Overall, based on the results of the current study, to minimize PM exposures, personal strategies for people residing in urban areas below 16 m are to keep the windows closed and stay indoors during the traffic rush hour of 8 am as well as the business hour of night market after 5 pm. Besides, keeping away from mobile sources during the traffic jam is advised in avoiding exposure to p-PAHs.

Conclusions
The present study analyzed vertical distribution and health impacts of PM in urban residential area of Taiwan. For vertical PM2.5 concentrations, the lowest levels were found at the height of 27 m as compared to those at 6 m and 16 m. In winter, PM and the constituents concentrations were significantly higher in the nighttime than those in daytime, and levels of PM1, PM2.5, and PM10 increased rapidly on 6 m and 16 m heights (but did not increase at 27 m) after 5 pm, whereas these trends became less significant in summer. Health risk analysis for PM2.5 concentrations showed a decrease in lung cancer mortality rate and an extended lifespan for residents living at 27 m. Findings in the current study highlight the needs to integrate more measurement data at broader residential heights, exposure scenarios, and the behaviors of deploying window/door for future investigation.  As there is no known level of pollution exposure that is harmless, strategies to minimize daily exposure are often impactful across a good range of settings. In Toronto, Canada, Zuraimi and Tan reported that tighter building envelope, use of mechanical ventilation systems, and improved filtration are key factors that reduce PM 2.5 exposure and its impacts on population health [55]. In fact, the indoor/outdoor ratios of PM concentration vary considerably due to the different-sized particles with different emission rates, the cracks and leaks in the building envelope, and the air exchange rates [56]. A study in Durbin, Ireland suggested that increasing air exchange between the outdoors and indoors during nighttime periods could flush out air pollutants [57]. Meanwhile, in the current study, as the levels of PM pollutants were higher at night than in the day, the proposal might not be applicable in Taiwan. Some qualified evidence supports individual actions to reduce cardiopulmonary health risks from exposures to ambient air pollutants by staying indoors and limiting physical exertion when air pollutant concentrations exceed levels associated with increased risk of acute and chronic health problems [58,59]. Overall, based on the results of the current study, to minimize PM exposures, personal strategies for people residing in urban areas below 16 m are to keep the windows closed and stay indoors during the traffic rush hour of 8 am as well as the business hour of night market after 5 pm. Besides, keeping away from mobile sources during the traffic jam is advised in avoiding exposure to p-PAHs.

Conclusions
The present study analyzed vertical distribution and health impacts of PM in urban residential area of Taiwan. For vertical PM 2.5 concentrations, the lowest levels were found at the height of 27 m as compared to those at 6 m and 16 m. In winter, PM and the constituents concentrations were significantly higher in the nighttime than those in daytime, and levels of PM 1 , PM 2.5 , and PM 10 increased rapidly on 6 m and 16 m heights (but did not increase at 27 m) after 5 pm, whereas these trends became less significant in summer. Health risk analysis for PM 2.5 concentrations showed a decrease in lung cancer mortality rate and an extended lifespan for residents living at 27 m. Findings in the current study highlight the needs to integrate more measurement data at broader residential heights, exposure scenarios, and the behaviors of deploying window/door for future investigation.