The recorded meteorological data highlight the overall climatological stability of the Los Angeles Basin, characterized by moderate differences in terms of temperature (T) and RH percentage between colder and warmer phases (Table 1). These moderate meteorological variations may cause low variability in pollutant levels throughout the year. The average T and RH percentage were lower in Riverside than at the three San Gabriel sites. During the warmer phase of the study, the indoor and outdoor areas were generally characterized by similar T, whereas during the colder phases the average T was ∼10°C higher indoors than outdoors. The higher differences between indoor and outdoor T values in the colder phase can potentially cause higher variability between indoor and outdoor concentrations compared with the warmer phase, especially for volatile (or semi-volatile) components. The AER at different sites were relatively similar to each other throughout the monitoring phases (Table 1), suggesting an overall similarity in home characteristics among different communities (Polidori et al., 2007). The magnitude of the AERs was generally low (0.21–0.4/h), and consistent with the low number of open windows and doors, the presence of central air conditioners, and the overall structural characteristics of the studied retirement homes.
Table 1. Studied sites PM concentrations, meteorology and air exchange rates
| ||Quasi-UF PM (μg/m3)||Temperature (°C)||Outdoor humidity (%)||AER (per h)|
| San Gabriel 1||9.9 ± 2.2||10.3 ± 1.6||25.1 ± 2.2||26.1 ± 1.2||60 ± 6||0.25 ± 0.04|
| San Gabriel 2||9.3 ± 1.8||9.8 ± 1.6||21.5 ± 2.1||23.6 ± 0.9||58 ± 15||0.28 ± 0.06|
| San Gabriel 3||10.4 ± 2.3||6.6 ± 1.3||25.9 ± 2.8||23.3 ± 1.1||58 ± 11||0.40 ± 0.12|
| Riverside||11.5 ± 3.0||8.7 ± 2.3||21.1 ± 4.0||24.9 ± 1.7||53 ± 15||0.21 ± 0.06|
| San Gabriel 1||8.8 ± 1.8||9.6 ± 3.1||15.4 ± 2.8||23.4 ± 1.2||58 ± 19||0.33 ± 0.07|
| San Gabriel 2||10.4 ± 2.6||9.4 ± 2.6||14.9 ± 2.1||23.9 ± 0.8||49 ± 14||0.31 ± 0.10|
| San Gabriel 3||10.7 ± 2.0||7.1 ± 2.6||16.6 ± 3.6||24.7 ± 1.2||55 ± 10||0.26 ± 0.08|
| Riverside||7.2 ± 2.2||6.0 ± 1.5||11.2 ± 2.8||25.4 ± 0.7||42 ± 22||0.31 ± 0.09|
The average outdoor quasi-UF mass concentrations at all sites varied from 9.3 to 11.5 μg/m3 in the warmer phases and from 8.9 to 10.7 μg/m3 in the colder phases, thus indicating relatively low seasonal variability. Similar to outdoor quasi-UF PM levels, indoor levels were consistent throughout the year at all sites. Mean indoor concentrations were generally lower than or similar to the corresponding outdoor concentrations (average indoor levels were 63–107% of their outdoor values).
Outdoor organic species and seasonal variability
As shown in Figure 1a, the outdoor PAHs concentrations were similar in the warmer and the colder phases. However, medium and high molecular weight PAHs levels were slightly higher in the colder months compared with the warmer periods. PAHs are mainly products of incomplete combustion, including vehicular emissions (Manchester-Neesvig et al., 2003). The higher colder phase levels could be attributed to the influence of cold-start spark-ignition from gasoline-powered vehicles, which emit higher amounts of high molecular weight PAHs, such as benzo(ghi)perylene and coronene, than hot-start conditions (Fine et al., 2004; Lough et al., 2007; Miguel et al., 1998).
Figure 1. Outdoor concentrations of (a) PAH’s, (b) hopanes and steranes, (c) n-alkanes, and (d) acids. The presented values are average concentrations across all sites, and error bars are standard deviation of these averages at each site
Download figure to PowerPoint
The average seasonal changes in hopanes and steranes were also quite small (Figure 1b), a result that can be explained by the low seasonal variability in the emission rates of their main source (i.e., engine lubricating oil of mobile sources; Rogge et al., 1996; 1993; Schauer et al., 1996), which are independent of the driving conditions (e.g., cold-start, hot-start, or steady state; Schauer et al., 2002). Moreover, this low seasonal variability may reflect the low volatility of these organic species.
A major portion of the analyzed n-alkanes was characterized by substantially higher and more variable concentrations in the colder months over the studied sites (Figure 1c), possibly because of the enhanced condensation of gas phase n-alkanes onto existing particles (Fraser et al., 1997; Kuhn et al., 2005). Conversely, the lower n-alkane concentrations observed in the warmer phases could be due to volatilization of their most volatile fraction (Fraser et al., 1997; Kuhn et al., 2005) and to variations in the emission sources of these compounds.
Hexadecanoic, octadecanoic, and phthalic acids were the most dominant measured acids in quasi-UF PM (Figure 1d). Phthalic acid concentration in the warmer months was on an average more than two times higher than in the colder periods. This variability is probably because of relatively higher photo-oxidation rates of organic gases in warmer conditions (Pandis et al., 1993; Robinson et al., 2007; Rogge et al., 1991).
Figures 2 and 3 present the relationship between indoor and outdoor concentrations for the studied organic compounds in the quasi-UF particle range. Figure 2 shows the average indoor and outdoor levels of PAHs, hopanes and steranes, n-alkanes, and organic acids at each site and phase of the study. Figure 3 shows the indoor/outdoor (IN/OUT) ratios and correlation coefficients of all measured organic species for different phases of the study. The average outdoor level of the sum of all PAHs were lowest in Riverside (0.5 ng/m3) and highest at San Gabriel 2 site (1.5 ng/m3). The Riverside site was the most distant from any primary combustion sources (i.e., freeways and busy roadways), while the San Gabriel 2 site was the closest to a major freeway (within 300 m) among all sites. Typically, the average concentrations of the sum of all measured PAHs were similar, but slightly lower indoors than outdoors. Accordingly, the average IN/OUT ratio of most of the measured PAHs was close to or lower than 1 and correlation coefficients were always positive and generally high for most of the components (median R for all components = 0.60). These results may reflect the possible impact of outdoor sources (e.g., motor-vehicle emissions) on indoor PAHs in the quasi-UF mode, which is consistent with previous studies (Ohura et al., 2004). PAHs generated by tobacco smoke were not expected indoors, as all of the studied retirement communities were non-smoking residences. Few individual PAH components, such as phenanthrene, anthracene, and benz(a)anthracene, showed slightly higher than 1 average IN/OUT ratios and a relatively high standard deviation, hence indicating the possibility of indoor sources (e.g., natural gas appliances) for these species.
Figure 2. Concentration of total (a) PAH’s, (b) hopanes and steranes, (c) n-alkanes, and (d) acids. Dots are average of concentrations across all the sites, and error bars are standard deviation of these averages at each site
Download figure to PowerPoint
Figure 3. Correlation coefficient and indoor and outdoor ratios of (a) PAHs, (b) hopanes and steranes, (c) n-alkanes, and (d) acids, values are averaged over the sites and bars are the standard deviation over the sites
Download figure to PowerPoint
Similarly to PAHs, the sum of all measured hopanes and steranes concentrations was slightly higher outdoors than indoors at all sites. Average IN/OUT ratios were close to 1 (min. = 0.83, max. = 1.31, and median = 0.94), accompanied by relatively high R-values (median of R for all components = 0.74). As in the case of PAHs, these results highlight the possible influence of outdoor sources to the measured indoor concentrations of hopanes and steranes. There were no clear seasonal patterns for the corresponding IN/OUT ratios and R-values. As hopanes and steranes are more stable species compared with PAHs, the effect of temperature differences between indoor and outdoor environments on their indoor and outdoor associations becomes less significant. Similar relationships between the indoor–outdoor concentrations of hopanes and steranes (and also PAHs) were found in a recent study conducted in Tampa, Florida (Olson et al., 2008).
The average indoor concentrations of the sum of all measured n-alkanes were typically higher than the corresponding outdoor levels. Exceptionally, high indoor levels with large standard deviation were shown during the colder phase at San Gabriel 3 and Riverside sites. Most of the IN/OUT ratios of n-alkanes were much higher than 1 (up to 23 during the colder phase). R-values were not always positive and, on an average, were much lower than those found for PAHs, hopanes, and steranes (min. = −0.14, max. = 0.79, and median = 0.41). The high IN/OUT ratios and low R-values are indicative of the significant influence of indoor sources of n-alkanes. Considerably higher indoor n-alkane levels compared with the outdoor levels were also found in a previous study (Olson et al., 2008). A variety of PM sources such as cooking, household products, dust, smoking, and candle burning are known as indoor sources of n-alkanes (Fine et al., 1999; Kleeman et al., 2008; Schauer et al., 1999). The average indoor and outdoor concentrations of n-alkanes at the San Gabriel 3 site were substantially higher than those at the other sites.
Similarly to n-alkanes, average indoor concentrations of the sum of all measured n-alkanoic acids were higher than the corresponding outdoor levels. The indoor concentrations were substantially enriched (i.e., more than three times of outdoor levels) with a large variation in indoor concentrations during both phases at San Gabriel 1 and 2 sites. IN/OUT ratios of measured organic acids were higher than 1 (except for octanoic acid in the colder months) with average IN/OUT ratios of 4.8 (IN/OUT ratios of up to ∼30 were found for oleic acid in the warmer phases). The high IN/OUT ratios were accompanied by low R-values (median R-values = 0.19; lowest R = -0.27 for oleic acid in the warmer phases), indicating the influence of indoor sources for organic acids. Cooking is a major indoor source of these acids, and oleic and palmitoleic acids had often been used as biomarkers of food cooking (Robinson et al., 2006). Other significant sources of organic acids include direct emission from people (e.g., skin oils from heated surfaces). Throughout the study, no specific seasonal trends were observed for the association between indoor and outdoor organic acids. All R-values and slopes of the regressions between indoor and outdoor concentrations and their related intercepts are listed for all individual organic species in Table S2. For any PM species measured, the regression slope and the corresponding intercept are good surrogates for Finf and indoor-generated concentration, respectively, during periods characterized by a relatively high R. A more comprehensive discussion about the outdoor-generated and indoor-infiltrated contributions to the measured indoor concentrations of PM2.5 and its carbonaceous components is presented in Polidori et al., 2007;. Scatter plots showing the indoor and outdoor concentrations of representative organic species measured at all sites and during all phases of CHAPS are shown in Figure S1.
The carbon preference index (CPI), i.e., the ratio of the concentrations of odd-carbon-to-even-carbon n-alkanes, is a parameter used to differentiate between anthropogenic and biogenic source contributions to PM (Simoneit, 1986). Previous studies have shown that n-alkanes originating from anthropogenic sources have a CPI close to 1, whereas the CPI is generally higher than 2 when the biogenic sources are dominant (Simoneit, 1986). The average indoor and outdoor CPIs at sampling sites varied from 0.60 to 0.95, suggesting that anthropogenic emissions (originated from fossil fuel) are the dominating sources in the studied area (see Figure S2).
The coefficient of variance, also called coefficient of variation (CV = standard deviation/mean), was determined for several measured components to investigate their spatial variation over the studied area (Figure 4). Most of the uncertainty in the calculated coefficient of variations is due to the fact that sampling at the different sites was not concurrent, which can affect the calculated CVs to some extent. However, CV was calculated separately for colder and warmer phases to reduce this effect. In general, colder phases were characterized by higher CV values compared with the warmer phases, probably because of the lower regional atmospheric mixing and increased atmospheric stability during cold meteorological conditions. Quasi-UF mass concentrations showed relatively low variability over the studied sites in both indoor and outdoor environments (CV = 0.09–0.22). Outdoor WSOC was less variable in the warmer phases compared with colder phases, probably because of the seasonal variability of the main WSOC sources (i.e., secondary atmospheric processes and biomass burning; Weber et al., 2007). Secondary quasi-UF OC formation is expected to be higher in warmer periods because of higher photochemical activity, whereas higher biomass burning emission of OC is expected in colder conditions in the absence of wildfire. Biomass burning has localized effects on OC compared with secondary photo-oxidation formation of OC, which can influence the concentrations of several carbonaceous compounds on a regional scale. Although particulate OC formation by biomass burning is generally not significant (at least in the absence of wildfires) in the polluted Los Angeles Basin (Minguillon et al., 2008), it may still affect the spatial variability of WSOC levels. Regarding indoor CVs of WSOC, the relatively high values in the warmer months may be due to the high levels of indoor WSOC at the San Gabriel 1 and 2 sites during these periods. The spatial variance (CV) of outdoor hopanes and steranes was 0.83 and 0.97, respectively, one of the highest among all organic species. This variability originates from differences in the local traffic sources and their influence on each sampling site. Spatial variability in indoor and outdoor PAHs was similar, because of the significant contribution of outdoor originated PAHs to indoor PM.
Figure 4. Coefficient of variance for indoor and outdoor organic groups for (a) warmer and (b) colder period of the study. SOA, secondary organic aerosol; rs-Dust, resuspended dust; nss-Sulfate, non-sea salt sulfate; SS, sea salt; BB, biomass burning; LDV, light-duty vehicles; HDV, heavy-duty vehicles
Download figure to PowerPoint
Source contribution estimations
The results of the source apportionment of quasi-UF particle mass are presented in Figure 5. Vehicular sources, including both HDV and LDV, showed the highest contribution for both indoor and outdoor particles at all sites (on an average 1.67–4.86 μg/m3 or 24–47% of the quasi-UF mass). Estimations of the HDV contribution were higher than those from LDV. This could be due to the location of the study sites, all situated in the eastern region of Los Angeles, where the traffic fleet has a higher fraction of HDV compared with the typical urban area of Los Angeles. The average percentage contribution of HDV to the total vehicle fleet (HDV + LDV) at the I-10, I-210, and I-60 freeways in eastern Los Angeles counties (San Bernardino and Riverside counties), where most of the sites were located, was ∼10–15%, which is approximately two to three times higher than the corresponding percentage in the Los Angeles urban area (values were estimated from data obtained from the Caltrans website; http://traffic-counts.dot.ca.gov). As we stated in the methods section, the LDV source profile used for the CMB analysis is from a study conducted near the I-110 freeway, between downtown Los Angeles and Pasadena, CA (Phuleria et al., 2007). However, in the study areas, the emissions of EC from older vehicles can be higher than that at the I-110 (Schauer et al., 2002). This could bias the results obtained by the CMB analysis by underestimating the LDV contribution, because in this model, EC is a key determinant for discriminating between HDV and LDV. Therefore, higher EC levels emitted from old LDV can artificially increase the source contribution estimations for HDV.
Figure 5. Contribution of different sources to quasi-ultrafine PM in the four sites and during the two sampling periods
Download figure to PowerPoint
The relative contribution of biomass burning to the measured indoor and outdoor quasi-UF mass was low (on an average, from 0.06 to 0.87 μg/m3, or about 1–9% of the measured mass, across all sites and phases of the study, with the exception of San Gabriel 2 site). The differences between indoor and outdoor contributions were low, except at the San Gabriel 2 site in the warmer phase. The average outdoor biomass burning contribution was 1.4–3.8 times higher in the colder phases compared with the warmer phase. Sea spray and ship emission contributions to PM were negligible, as it is expected for sites located far away from the ocean (median sea spray and ship contributions were 0.12 and 0.08 μg/m3, respectively, over all sites and phases, which corresponds to around 1% of the measured mass). Estimated indoor and outdoor non-sea salt sulfate contributions tracked each other at most of sites, suggesting that a substantial portion of indoor sulfate originates from outdoors (median = 0.72 μg/m3 indoors and 0.74 μg/m3 outdoors). Outdoor non-sea salt sulfate was, on an average, 33% higher in warmer phases compared with colder phases, confirming the secondary origin of this pollutant (Rodhe, 1999).
Candle and cigarette smoke sources were also used as an input for the CMB model, but the resulting contributions were negligible (no more than 0.02 μg/m3, or less that 1% of the measured mass, at all sites and during all phases of the study). Our CMB model was not able to apportion the contribution of cooking using common meat cooking source profiles (Kleeman et al., 2008; Schauer et al., 1999). This does not necessarily imply that the contribution of cooking to the quasi-UF particle mass was negligible, but it can indicate that the cooking source profile used is not representative of the specific food cooking emissions at our study sites. The influence of food cooking emissions on indoor PM is evidenced by the elevated levels of indoor organic acids such as oleic acid and palmitoleic acid, frequently used as biomarkers of food cooking (Robinson et al., 2006). The same study also showed that significant inconsistencies exist between ambient data and published source profiles for cooking, which makes it difficult to obtain reliable estimates of the relative contribution of cooking to ambient PM. Resuspended dust contributions were 0.02–1.66 μg/m3 (or approximately 0–22% of the measured mass) to indoor, and 0.27–2.17 μg/m3 to outdoor quasi-UF PM across all sites. Road dust and indoor activities are, respectively, the main outdoor and indoor sources of resuspended dust. Warmer phase resuspended dust was on average more than 100% higher than the corresponding colder phase levels across all outdoor stations.
Estimated SOA accounted for 0.23–1.62 μg/m3 (or around 3–19% of the measured mass) of the indoor and outdoor quasi-UF PM at all sites and during all phases. At some sites (e.g., San Gabriel 2, colder phase), estimated SOA concentrations were higher indoors than outdoors (up to approximately three times), which can be partially because of the formation of secondary particles in indoor environments from reactions of household products with ozone and to a lesser extent hydroxyl radicals (Destaillats et al., 2006; Weschler and Nazaroff, 2008); also, as SOA was estimated here from WSOC, it can be influenced by indoor emissions of water-soluble organics (such as those from cooking). The average SOA contribution in the warmer phase was about two times higher than during the colder phase at all outdoor sites, which highlights the important role of photo-oxidation in the formation of SOA. The average un-apportioned fraction of quasi-UF PM was 33 ± 15% among all sites and phases. A fraction of this un-apportioned mass could be attributed to ammonium nitrate, which was not measured, but could account for as much as 2–3 μg/m3 of the PM0.25 mass concentrations, especially in the Riverside area (Hughes et al., 2002; Kleeman et al., 1999; Sardar et al., 2005). Moreover, there are uncertainties associated with the calculations of SOA. Part of this uncertainty originates from the multiplication factor used to convert WSOC to SOA and from the assumed fraction of WISOC in SOA (20%), as they both can vary with time and location; (Docherty et al., 2008). A study carried out in Tokyo showed that 6–26% of summer oxygenated OC was water-insoluble (Kondo et al., 2007), whereas water-insoluble SOA fractions as large as 60% have recently been reported for urban environments (Favez et al., 2008). In a recent study by Docherty et al. (2008), SOA was estimated to comprise 45–90% of the organic fine aerosol mass in the Los Angeles Basin.
Lastly, the estimated indoor LDV and HDV source contributions were similar to those calculated outdoors during both phases and at all retirement communities. This is probably because of the significant role of outdoor mobile sources on indoor environments and illustrates the high indoor infiltration of particles generated by mobile sources. This finding has important exposure and health implications considering that in an earlier publication (also part of the CHAPS study) we found that traffic-related particles had much stronger associations with adverse health effects in the elderly retirees of the studied communities compared with uncharacterized indoor particles (Delfino et al., 2008).
To the best of our knowledge, this work and the one conducted by Minguillon et al. (2008) are the only source apportionment studies on quasi-UF PM conducted in the LA area. The results from both investigations are very consistent.