Trends in the Elemental Composition of Fine Particulate Matter in Santiago, Chile, from 1998 to 2003

Transcription

1 Journal of the Air & Waste Management Association ISSN: (Print) (Online) Journal homepage: Trends in the Elemental Composition of Fine Particulate Matter in Santiago, Chile, from 1998 to 2003 Sonja N. Sax, Petros Koutrakis, Pablo A. Ruiz Rudolph, Francisco Cereceda- Balic, Ernesto Gramsch & Pedro Oyola To cite this article: Sonja N. Sax, Petros Koutrakis, Pablo A. Ruiz Rudolph, Francisco Cereceda- Balic, Ernesto Gramsch & Pedro Oyola (2007) Trends in the Elemental Composition of Fine Particulate Matter in Santiago, Chile, from 1998 to 2003, Journal of the Air & Waste Management Association, 57:7, , DOI: / To link to this article: Published online: 24 Jan Submit your article to this journal Article views: 184 Citing articles: 22 View citing articles Full Terms & Conditions of access and use can be found at

2 TECHNICAL PAPER ISSN: J. Air & Waste Manage. Assoc. 57: DOI: / Copyright 2007 Air & Waste Management Association Trends in the Elemental Composition of Fine Particulate Matter in Santiago, Chile, from 1998 to 2003 Sonja N. Sax Gradient Corporation, Cambridge, MA Petros Koutrakis Harvard School of Public Health, Boston, MA Pablo A. Ruiz Rudolph Centro Mario Molina Chile, Santiago, Chile; and Harvard School of Public Health, Boston, MA Francisco Cereceda-Balic Department of Chemistry, Federico Santa Maria Technical University, Valparaíso, Chile Ernesto Gramsch Department of Physics, University of Santiago, Santiago, Chile Pedro Oyola Centro Mario Molina Chile, Santiago, Chile ABSTRACT Santiago, Chile, is one of the most polluted cities in South America. As a response, over the past 15 yr, numerous pollution reduction programs have been implemented by the environmental authority, Comisión Nacional del Medio Ambiente. This paper assesses the effectiveness of these interventions by examining the trends of fine particulate matter (PM 2.5 ) and its associated elements. Daily fine particle filter samples were collected in Santiago at a downtown location from April 1998 through March Additionally, meteorological variables were measured continuously. Annual concentrations of PM 2.5 decreased only marginally, from 41.8 g/m 3 for the period to 35.4 g/m 3 for the period. PM 2.5 concentrations exceeded the annual U.S. Environmental Protection Agency standard of 15 g/m 3. Also, approximately 20% of the daily samples exceeded the old standard of 65 g/m 3, whereas approximately half IMPLICATIONS This paper reports particle composition measurements taken in Santiago from 1998 to Although PM 2.5 concentrations did not decrease substantially in the 5-yr period of the study, there were large decreases in Pb, Br, and S as a result of the phaseout of leaded gasoline and the introduction of low sulfur fuels. These results show that the pollution reduction programs implemented by the environmental authority have been successful. However, particle mass levels continue to exceed national and international standards, thus, efforts to reduce air pollution should continue. of the samples exceeded the new standard of 35 g/m 3 (effective in 2006). Mean PM 2.5 levels measured during the cold season (April through September) were three times higher than those measured in the warm season (October through March). Particulate mass and elemental concentration trends were investigated using regression models, controlling for year, month, weekday, wind speed, temperature, and relative humidity. The results showed significant decreases for Pb, Br, and S concentrations and minor but still significant decreases for Ni, Al, Si, Ca, and Fe. The larger decreases were associated with specific remediation policies implemented, including the removal of lead from gasoline, the reduction of sulfur levels in diesel fuel, and the introduction of natural gas. These results suggest that the pollution reduction programs, especially the ones related to transport, have been effective in reducing various important components of PM 2.5. However, particle mass and other associated element levels remain high, and it is thus imperative to continue the efforts to improve air quality, particularly focusing on industrial sources. INTRODUCTION In the last decades, the city of Santiago, Chile, has experienced a rapid urban expansion and economic growth, mostly in the industrial sector. The population of Chile is approximately 14 million, of which approximately 40% (6 million) live in the capital, Santiago. The city is located in the central valley of Chile at an elevation of 520 m (0.3 mi) above sea level and is surrounded by two large mountain ranges, the Andes to the east and the Coastal to the west. Summers are hot and dry, whereas winters tend to Volume 57 July 2007 Journal of the Air & Waste Management Association 845

3 Saxetal. be colder and more humid, with frequent thermal inversions. The topography, meteorology, large number of industries (70% of the country s total), high population density, and large vehicle and bus fleet (800,000) all contribute to poor air quality in Santiago. 1 High levels of nitrogen oxides, CO, ozone, particulate matter (PM) less than 2.5 m (PM 2.5 ), PM less than 10 m (PM 10 ), polycyclic aromatic hydrocarbons, and black carbon have been measured in this urban center. 2 9 Previous studies conducted in Chile have found health effects associated with high air pollution levels in Santiago, including daily mortality 10,11 and hospital admissions for respiratory illnesses From a health standpoint, there is evidence that certain chemical components of PM are more likely to be responsible for toxic or adverse health effects. 15 For example, some metals such as iron (Fe), vanadium (V), nickel (Ni), and copper (Cu) can trigger inflammation and cause DNA damage. Organic compounds can act as irritants or can cause mutations, which can result in cancer. Sulfates and nitrates can impair mucociliary clearance and change the bioavailability of metals because of their acidity. Also, elemental carbon can induce lung irritation, enhance epithelial proliferation, and cause fibrosis in the lungs. 15 Furthermore, an epidemiologic study by Laden et al. 16 of mortality in six U.S. cities showed that fine particles from motor vehicle and coal combustion sources were associated with higher mortality outcomes, whereas particles of crustal origin were not associated with increases in mortality. Thus, determining the composition and potential sources of PM in Santiago can help to implement more cost-effective mitigation strategies that target the reduction of the more toxic components of PM. Additionally, a better understanding of the factors that influence air pollution levels, such as meteorology, will also aid policy decisions, because these factors also vary over time. Since the early 1990s, the Chilean government has taken numerous steps to improve the air quality in the Santiago metropolitan area (e.g., see Jorquera et al. 17 ). These measures include car-use bans based on license plate number and mandatory vehicle inspections. Also, new cars are now required to have catalytic converters; emission standards have been set for industrial, nonindustrial, and residential heating sources; and an emissions-trading program has been established. In addition, a large fraction of the old bus fleet has been replaced with cleaner diesel fuel buses, and this effort is expected to continue until the entire fleet is replaced. Recently, two important measures were taken: in 1998, a street cleaning and paving plan was initiated and a tree planting campaign was launched to reduce road dust levels, and in 2001, lead was removed from gasoline and sulfur was reduced from diesel fuel from 1000 parts per million (ppm) to 300 ppm. To assess the effectiveness of these and other measures, the Chilean government established an extensive air pollution monitoring network across the city of Santiago. The air pollution network in Santiago is supported by the Chilean Ministry of Health and has been in operation since the mid-1980s. This network provides the basis for investigating the spatial and temporal profiles of air pollution levels throughout the metropolitan area and aids in the evaluation of the effectiveness of emission control strategies. Using data from this network, our group has examined trends in PM 2.5,PM 10, and PM between 2.5- and 10- m concentrations over a 12-yr period using regression analyses and controlling for corresponding changes in meteorology during that period. 18 The results from these analyses showed a substantial decrease of 52% in PM 2.5 concentrations over the 12-yr period examined. Specific trends in the elemental composition of PM 2.5 were not, however, examined. Thus, in this paper, we focus on the elemental components of PM 2.5 samples collected in downtown Santiago for the years using similar regression analyses. In addition, results from this trends analysis are used to help determine the effectiveness of pollution control measures. EXPERIMENTAL WORK Sampling and Analysis PM 2.5 sampling was conducted at the Parque O Higgins air quality monitoring station in downtown Santiago. This site is located inside the largest park in the Santiago metropolitan area, close to an amusement park, a rollerblading rink, and a University of Chile campus. More importantly, it is located approximately 0.5 km from a major highway (Route 5, also called the Panamericana ) and near minor pollutant sources, such as mechanic shops, metal works, and other small businesses. Twenty-four-hour particle samples (midnight to midnight) were collected from April 1998 through March PM 2.5 filter samples were collected every day during the cold season (April through September) and every other day during the warm season (October through March). Samples were collected using dichotomous samplers (Andersen Instruments Inc.). Details of sampling and filter weighing can be found elsewhere. 18 Elemental analysis was conducted using X-ray fluorescence on the PM 2.5 aerosol filter samples (Desert Research Institute). To minimize analytical costs, only a subset of randomly selected filters was selected for elemental analysis. For each month, only six to eight filters were selected, or approximately one sample every fourth day, for a total of 456 filters analyzed across the 5 yr of the study. Blanks represented approximately 10% of the samples. The limit of detection (LOD) was calculated for each element as three times the standard error of the blanks. Only elements that had at least 80% of all reported values above LOD were included in the statistical analysis. A total of 14 elements satisfied these criteria and were, thus, included in the analysis together with PM 2.5. These include aluminum (Al), silicon (Si), calcium (Ca), iron (Fe), potassium (K), manganese (Mn), zinc (Zn), sulfur (S), copper (Cu), nickel (Ni), chromium (Cr), lead (Pb), bromine (Br), and selenium (Se). Additionally, meteorological variables such as wind speed, wind direction, temperature, and relative humidity (RH) were measured at the sampling site during particle sampling. However, to obtain complete 846 Journal of the Air & Waste Management Association Volume 57 July 2007

4 Sax et al. data on meteorological variables, these data were supplemented from the La Paz air quality monitoring station, located a few kilometers north-northwest of the Parque O Higgins station. Data Analysis Relationships of PM 2.5 and elements with meteorological and temporal variables were investigated using linear regression models to identify specific factors influencing particle concentrations and to quantify their relative impact. 18 Models were run in SAS. The regression model included data for year, month, weekday, and meteorology (wind speed, temperature, and RH) as categorical variables. The continuous meteorological data were dichotomized or trichotomized using the following ranges: for wind speed (ws), ws 0.8 m/sec, 0.8 ws 1.6 m/sec, ws 1.6 m/sec; for RH (rh), rh 70% and rh 70%; and for temperature (mp), mp 20 C and mp 20 C. Wind direction data were also modeled but were found to be nonsignificant and, therefore, were not included in the analysis. Because particulate mass and elemental concentrations were log-normally distributed, concentrations were log transformed, and the natural logs (e.g., ln[pm]) were regressed against the categorical values (eq 1): ln PM a yj*yj mj*mj wj*wj wsj*wsj tmpj*tmpj rhj*rhj (1) where, is the regression intercept and yj, mj, wj, wsj, tmpj, and rhj are the regression coefficients of the independent variables year, yj (j 1 5), month, mj (j 1 12), weekday, wj (j 1 7), wind speed, wsj (j 1 3), temperature, tmpj (j 1 2), and RH, rhj (j 1 2). Based on eq 1, particle concentrations can be expressed as the product of multiple exponential terms: PM exp a * exp yj*yj mj*mj wj*wj wsj*wsj tmpj*tmpj rhj*rhj (2) For simplicity, eq 2 was transformed as follows: PM I * f *yearj * f monthj *f weekdayj * f wsj * f tmpj * f rhj (3) where fij exp[ ij * var ij] is the concentration impact factor (CIF) of a variable i (e.g., year) of a category j. For each of the categorical variables, a reference level was set, with subsequent effects calculated relative to that reference variable. For this analysis, Saturday, December, and year 2003 were set as the reference variables for weekday, month, and year, respectively. Each year category was grouped using measurements collected from April of 1 year to March of the following year (e.g., April 1999 to March 2000, named year 1999/2000), yielding 5 full years of data. For the meteorological variables, the highest ranges were used as the reference for the corresponding group (ws 1.6 m/sec; rh 70%; and tmp 20 C). The CIF for the intercept I e corresponds with the concentration at the reference level (e.g., year 2003, December, Saturday, ws 1.6 m/sec, rh 70%, and tmp 20 C). For the other variables, a regression coefficient of 0 equals a CIF of 1, meaning that the variable has no effect. Conversely, a CIF greater than 1 represents a greater concentration of PM or element, relative to the reference point, and a value less than 1 represents a concentration that is lower relative to the reference. The concept of impact factors has been used previously. 19 It is worth noting, however, that the effects of the different parameters are multiplicative rather than additive. The significance of the association was determined by the P values of the estimates. Regression results are discussed in terms of their corresponding CIFs to facilitate comparisons in concentration levels across different parameters. RESULTS AND DISCUSSION Mass and Elemental Concentration Levels Average concentrations of PM 2.5 and trace elements were determined for each season and year (Table 1), and the overall annual mean was estimated by averaging the means of the two seasons (Table 1). As indicated by Table 1, PM 2.5 concentrations have decreased only moderately in the period between 1998 and 2003 with the highest annual in 1998/1999 of 41.8 g/m 3 and the lowest annual in 2002/2003 of 35 g/ m 3. As Figure 1 shows, however, these concentrations are substantially lower compared with levels in the early 1990s. Previous analysis showed a decrease in PM 2.5 concentrations above 50% between the years 1989 and Over this time period, there were several periods of reduced PM concentration levels. These reductions correlate well with the implementation of pollution control strategies. For example, in , catalytic converters were required for all light vehicles, and street-paving programs were established. In 1997, street-cleaning programs were initiated, and the use of natural gas as an alternative fuel was introduced. Nevertheless, mean concentration levels during the period remained high, exceeding U.S. Environmental Protection Agency (EPA) annual PM 2.5 standard of 15 g/m 3. In addition, approximately 20% of the daily PM 2.5 concentrations exceeded the old EPA daily limit (65 g/m 3 ), and approximately 50% of the days exceeded the new standard of 35 g/m 3. In Table 2, concentrations of particle mass and various elements measured in Santiago in recent years are compared with concentrations in other Chilean cities and in major urban centers in the United States. The Santiago results show decreasing mass and element concentrations. The observed mean PM 2.5 concentration levels in recent years (2002) are similar to annual mean levels measured in other Chilean cities such as Volume 57 July 2007 Journal of the Air & Waste Management Association 847

5 Saxetal. Table 1. Mean concentrations of PM 2.5 ( g/m 3 ) and associated elements (ng/m 3 ) for the years and for two seasons, cold (April to September) and warm (October to March), as well as the yearly. 1998/ / / / /2003 Analyte Cold Warm Cold Warm Cold Warm Cold Warm Cold Warm PM Al Br Ca Cr Cu Fe K Mn Ni Pb S Se Si Zn Rancagua (42.6 g/m 3 ), Temuco (35.2 g/m 3 ), and Valparaiso (35.7 g/m 3 ). 20 Al and Si levels are also comparable to those reported for other Chilean cities for 1998, except for Rancagua. Sulfur levels in Santiago for 1996 and 1999/2000 were similar to those reported for other Chilean cities except for Temuco, which had much lower levels. More recent S levels (2002/2003) in Santiago have decreased by a factor of two. In 1996 and 1999, Zn and Pb levels were similar to those observed in other Chilean cities. In , however, Zn and Pb levels were approximately 4 and 12 times lower, respectively, than the highest levels observed in other Chilean cities. Previously reported Cu levels in Santiago for 1996 and were similar to those reported for Rancagua and Valparaiso but were much higher than those in Temuco. In , Cu concentrations in Santiago were reduced substantially from previous years. Santiago levels of PM 2.5 and elements were also compared with levels of other major cities across the world. Compared with concentrations in U.S. cities, the current Santiago PM 2.5 levels are similar to those in Riverside, CA, but about twice as high as in other cities (Table 2). 21 Al and Si levels were also similar to levels found in Riverside and approximately 50% higher than in other U.S. cities. In addition, S levels in Santiago 848 Journal of the Air & Waste Management Association Volume 57 July 2007

6 Sax et al. Figure 1. Monthly concentrations of PM 2.5 from 1989 to 2003 at the Parque O Higgins site in Santiago. were similar to those found in Western U.S. cities (e.g., Houston, TX, and Riverside, CA) but lower than concentrations in cities along the U.S. East Coast (Philadelphia, PA; Atlanta, GA; and Chicago, IL), which have large regional sources of sulfur dioxide. In contrast, Cu, Zn, and Pb levels were substantially higher (four to five times) in Santiago compared with concentrations observed in U.S. cities. Compared with urban centers in Europe, PM 2.5 levels in Santiago were substantially higher. For example, Querol et al. 22 report the following examples of PM 2.5 concentrations in various European countries: Spain and Germany have annual mean concentrations ranging from 20 to 30 g/m 3, whereas in Austria, the Netherlands, United Kingdom, and Switzerland, levels range from 15 to 20 g/m 3. Regression Model Results and Trends Regression analyses were conducted using the PM 2.5 and 14 element data measured in downtown Santiago for a 5-yr period between 1998 and Results are presented in Table 3 for PM 2.5 and for the two elements with the most dramatic trends, Pb and S, in Tables 4 and 5, respectively. These tables report the intercept, regression slopes (estimates), standard error values, P values, and corresponding CIFs for each categorical variable (year, month, weekday, wind speed, temperature, and RH). Intercept. The intercepts for PM 2.5, Pb, and S were 13.1 g/m 3, 6.8 ng/m 3, and 595 ng/m 3, respectively. These values correspond with the expected concentration for the reference day (i.e., Saturdays of December 2003 with a ws above 1.6 m/sec, rh 70%, and tmp 20 C). Trend and the Effect of Control Measures. This study sought to determine whether source emission control strategies implemented in Santiago (as described in the Introduction) have resulted in decreased particle mass concentrations and decreased concentrations of associated elements over the 5 yr of the study. By using a regression model, we assessed the effect of year (and, hence, interventions) while controlling for other parameters, such as meteorology and weather. This is important in a study of trends, especially over a relatively short period of time, because a few atypical years can make it difficult to compare annual mean concentrations. Table 2. Comparison of PM 2.5 and elemental concentrations among Santiago, other Chilean cities, and U.S. cities. Location Analyte PM 2.5 ( g/m 3 ) Al (ng/m 3 ) Si (ng/m 3 ) S (ng/m 3 ) Cu (ng/m 3 ) Zn (ng/m 3 ) Pb (ng/m 3 ) Santiago 1996 a /2000 b /2003 b Other Chilean Cities (1998) c Rancagua Temuco Valparaiso U.S. Cities (2003) d Philadelphia, PA Atlanta, GA Chicago, IL Houston, TX Riverside, CA Notes: a Artaxo et al. 6 ; b this paper (Table 1); c Kavouras et al. 20 ; d EPA. 21 Volume 57 July 2007 Journal of the Air & Waste Management Association 849

7 Saxetal. Table 3. PM 2.5 model results. Effect Estimate Standard Error P Value CIF Intercept Year 1998/ / / / / January February March April May June July August September October November December Sunday Monday Tuesday Wednesday Thursday Friday Saturday Wind speed ws 0.8 m/sec ws 1.6 m/sec) ws 1.6 m/sec Temperature tmp 20 C tmp 20 C RH rh 70% rh 70% The regression analysis results for PM 2.5 show that mass concentrations did not significantly change over time (Tables 3). As discussed above, it appears that, in recent years, PM 2.5 concentrations have stabilized. In contrast, several elements have shown dramatic decreases during the study period. Figure 2 shows a graph of CIFs for all of the elements by year. For most elements, the trend is clearly downward. In particular, Pb levels have decreased from a yearly mean of 359 ng/m 3 for the 1998/ 1999 period to a more modest level of 23.7 ng/m 3 for the Figure 2. CIFs for each element. 850 Journal of the Air & Waste Management Association Volume 57 July 2007

8 Sax et al. Table 4. Lead model results. Effect Estimate Standard Error P Value CIF Intercept Year 1998/ / / / / January February March April May June July August September October November December Sunday Monday Tuesday Wednesday Thursday Friday Saturday Wind speed ws 0.8 m/sec ws 1.6 m/sec) ws 1.6 m/sec Temperature tmp 20 C tmp 20 C RH rh 70% rh 70% /2003 period (Table 1), with a CIF of when comparing years 1998/1999 with 2002/2003 (Table 4). Decreases in Pb levels between 1998 and 2003 occurred in two stages: the first, between 1998/1999 and 1999/2000 (CIF of and 7.00, respectively), and the second, between 2000/2001 and 2001/2002 (CIF of 5.55 and 1.82, respectively; Table 4). These decreases were primarily the result of the removal of lead in gasoline and possibly also because of street-paving and street-cleaning campaigns. Bromine, which was added to gasoline together with Pb, showed similar decreases in concentrations (Figure 3). Sulfur also showed a considerable decrease in concentrations during the study period, with concentrations of 1588 ng/m 3 in 1998/1999 that were reduced to 919 ng/m 3 by 2002/2003. These decreases correspond well with the implementation of reduced S in car fuels in Also, reductions were more evident in the cold season. As shown in Figure 3, the cold-season peak was eliminated, whereas the warm season levels remain relatively stable. This is probably because of formation of secondary sulfate particles from regional sources. Large reductions in other elements were also observed, though not as large as those for Pb and S. For example, Ni showed a CIF of approximately 4 when comparing the years 1998/1999 with 2002/2003, with annual means of 2.6 ng/m 3 and 0.8 ng/m 3, respectively. Reductions above 2-fold were observed for Al for the same period, whereas a little less than a 50% reduction was observed for other elements such as Ca, Cu, Fe, K, Mn, Si, and Zn. These reductions may be related to reduction in dust resuspension as a result of the street-cleaning programs. Although some of these elements are not of crustal origin, resuspended urban dust can be contaminated with elements of industrial or of transport-related origin. 23 In contrast, Cr concentrations increased in the first few periods but then the trend reverted in later years, whereas Se concentrations increased across the 5-yr period. Despite these reductions, levels of Pb and other elements remain high compared with other cities worldwide (Table 2). This may be indicative of additional sources of these air pollutants. For example, traffic-related sources for Ba, Cu, Cr, Fe, Pb, Sb, Si, and Zn may include tire and brake wear. 24,25 Therefore, further traffic emission-control programs are necessary to achieve reduced levels. Moreover, additional programs could help address both direct emissions from vehicles, Volume 57 July 2007 Journal of the Air & Waste Management Association 851

9 Saxetal. Table 5. Sulfur model results. Effect Estimate Standard Error P Value CIF Intercept Year 1998/ / / / / January February March April May June July August September October November December Sunday Monday Tuesday Wednesday Thursday Friday Saturday Wind speed ws 0.8 m/sec ws 1.6 m/sec) ws 1.6 m/sec Temperature tmp 20 C tmp 20 C RH rh 70% rh 70% as well as indirect resuspension of elements that were originally emitted by other sources. The high levels of some elements related to industry (Cu, Fe, Mn, Ni, and Zn), together with the increasing levels of Cr and Se, suggest that further controls of industrial sources are advisable. Levels of Ca and K, which are typically associated with wood burning, are discussed in the following section. Month Effect: Seasonal Variability. As observed previously, PM 2.5 concentrations exhibited a strong seasonal pattern with mean concentrations approximately three times higher during the cold season (April to September). 18,26 The regression analysis and the differences in concentrations by season shown in Tables 3 and 1, respectively, confirm that PM 2.5 monthly concentrations exhibit pronounced seasonal variability, with CIF varying from unity in December to 3.69 in July. These differences are because of frequent thermal inversions within the Santiago basin typically encountered in the cold season. In the winter, a coastal low-pressure system frequently forms between two high-pressure systems, the semipermanent Pacific high and the migratory high in Central/North Argentina. An upsurge of the midtroposphere creates a warm ridge above Central Chile, resulting in stable conditions, a reduced mixing layer height, and, thus, poor ventilation of the Santiago basin. 26 This seasonal variability was also observed for several elements. For example, Pb had a maximum CIF in July of 3.49, whereas a maximum CIF of 1.95 was seen for S in August. These results agree with the trends depicted in Figure 3. Other trace elements, such as Cr, Mn, Ni, Cu, and Zn, also had strong seasonal trends, with winter month CIFs of 3 4. Poor air mass mixing favors the accumulation of combustion-generated PM 2.5, which can remain airborne for hours to days. Crustal elements (Al, Si, Ca, and K) had more modest seasonal variability compared with the other elements. For example, concentrations of Al and Si were lower in the cold/humid season, with CIF in July of 0.75 and 0.80, respectively. A possible explanation is that crustal material is more easily resuspended in the drier seasons compared with the more humid seasons. For K and Ca, 852 Journal of the Air & Waste Management Association Volume 57 July 2007

10 Sax et al. Figure 3. Concentrations of Pb, Br, and S for the years Lines represent 30-day moving s. a small seasonal trend was observed. Ca showed a maximum CIF of 1.25 in June, whereas K showed a maximum of 1.53 in April. Both K and Ca not only originate from crustal material but are also a result of biomass burning used in winter for heating. It is worth noting that the month effect was significant even when controlling for meteorological factors. If these factors are excluded from the regression analysis, the month effects are increased, indicating that these variables explain some of the variability in PM 2.5 and elemental concentrations. However, because there is still a significant amount of unexplained variation, additional meteorological parameters may also be of importance, including mixing height and synoptic air mass movements. The addition of these variables and the refinement of the regression model, for instance, by smoothing the meteorological variables, could further clarify the effect of year and, thus, of the control measures, while also controlling for meteorological variations. Effects of Wind Speed, Temperature, and RH. The regression analysis results suggest strong associations between particle concentrations and wind speed, and, to a lesser extent, temperature and RH (Table 3). PM 2.5 concentrations were approximately 82% and 31% higher for wind speed values, ws 0.8 m/sec and 0.8 ws 1.6 m/sec, respectively, compared with higher wind speeds (ws 1.6 m/sec). Figure 4 shows that wind speed and PM 2.5 concentrations have an inverse relationship, that is, when the wind speed is high, PM 2.5 concentrations are low. Most elements showed a similar effect of wind speed. In particular Cr, Mn, Fe, Cu, and Zn had CIFs above 2 for the low wind speed compared with high wind speeds (data not shown). As reported previously, air pollutant concentrations are considerably higher during the cold season because of the frequent stagnation of air masses in the Santiago basin. 4 Particulate levels decreased only during strong wind events that managed to push the air pollution away from the city. Similar relationships between pollutant concentrations and weather conditions in Figure 4. Concentrations of PM 2.5 and wind speeds for the years Lines represent 30-day moving s. Volume 57 July 2007 Journal of the Air & Waste Management Association 853

11 Saxetal. Santiago have been reported previously using more comprehensive weather measurements, such as wind trajectory analysis and boundary layer height. 26 The effects of both RH and temperature on both PM 2.5 and most elements were minimal and nonsignificant. The only elements that had significant temperature and humidity effects were the crustal elements. In particular, Al and Si had 50% higher concentrations for low humidity (dry) days, possibly because of increased particle resuspension. Weekday Effect. As shown in Tables 3 and 4, weekday is an important determinant for PM 2.5 and Pb concentrations. PM 2.5 concentration impact factors were 0.84, 0.89, 0.96, 1.03, 0.98, 1.03, and 1.00 for Sunday through Saturday, respectively. These results are expected, because traffic and other human particle-generating activities are reduced on weekends. Furthermore, our analysis suggests that there is a lag in particle levels, with Mondays having lower particle concentrations compared with the rest of the workdays and Saturdays having higher concentrations than Sundays. A very similar pattern was observed for all of the trace elements, in particular Pb (Table 4), and to a lesser extent S (Table 5). This lag indicates that pollution levels reasonably accumulate from one day to the other. CONCLUSIONS In this paper we examine the trends of PM 2.5 mass and elemental concentrations to assess the effectiveness of air pollution reduction interventions in Santiago. Concentration trends were investigated using regression models, while controlling for the effect of weekday, month, wind speed, temperature, and RH on particle and trace element concentrations. PM 2.5 concentrations and many elements exhibited strong seasonal patterns because of the distinct differences in climatologic conditions during the cold (April through September) and warm (October through March) seasons. Mean PM 2.5 and most of the elemental concentrations during the cold season were about three times higher than those observed during the warm season. The crustal elements, on the other hand, were found to have less seasonal variability, with slightly lower concentrations in the cold (wet) season. The results of the regression analyses also suggest that, for PM 2.5 and most elements, wind speed had a significant influence on concentrations. Most of the highest particle levels occur during low wind speed events, when air masses stagnate for several days over the metropolitan region and air pollution emissions are concentrated in the valley. Levels decrease when wind velocity increases to move pollution away from the city. Lastly, model results showed that particle and elemental levels are lower on Sundays and, to a lesser extent, Saturdays and Mondays. Further refinement of the model, including additional meteorological variables and/or improving the regression model, for instance by smoothing the variables, could clarify the contribution of control measures on PM 2.5 levels. We found significant decreases for important elements, such as Pb, Br, and S, after controlling for meteorological and other factors, though PM 2.5 levels have not decreased substantially in the time period studied. These reductions in PM 2.5 components demonstrate that pollution reduction programs, which included paving and cleaning streets and the use of unleaded gasoline and low-s diesel, among others, have been effective. Future analysis should assess new measures and ongoing emission trends. For example, in 2005, fuel S was further reduced from 300 to 50 ppm. In addition, the industrial sector has increased during recent years because of Chile s economic development, which can yield higher emissions of PM 2.5 and particularly of elements associated with these sources. Despite concentration decreases over the study period, levels of Pb and other toxic elements exceed typical concentrations found in the United States. Also, PM 2.5 concentrations remain high, exceeding EPA ambient air quality standards, and thus justifying continuing efforts to improve air quality. To achieve this, we suggest that further measures focus on control of both transport and industrial emissions. ACKNOWLEDGMENTS The filter samples were collected by the Servicio de Salud Metropolitano del Ambiente of the Ministry of Health. The data analysis was supported by the Comisión Nacional del Medio Ambiente. The authors thank Yolanda Silva Cerna and Ignacio Olaeta Undabarrena for their assistance in the collection and analysis of filter samples. REFERENCES 1. Romero, H.; Ihl, M.; Rivera, A.; Zalazar, P.; Azocar, P. Rapid Urban Growth, Land-Use Changes and Air Pollution in Santiago, Chile; Atmos. Environ. 1999, 33, Tsapakis, M.; Lagoudaki, E.; Stephanou, E.G.; Kavouras, I.G.; Koutrakis, P.; Oyola, P;. Von Baer, D. The Composition and Sources of PM 2.5 Organic Aerosol in Two Urban Areas of Chile; Atmos. Environ. 2002, 36, Jorquera, H. Air Quality at Santiago, Chile: a Box Modeling Approach: I Carbon Monoxide, Nitrogen Oxides and Sulfur Dioxide; Atmos. Environ. 2002, 36, Jorquera, H. Air Quality at Santiago, Chile: a Box Modeling Approach: II PM 2.5, Coarse and PM 10 Particulate Matter Fractions; Atmos. Environ. 2002, 36, Adonis, M.; Gil, L. Polycyclic Aromatic Hydrocarbon Levels and Mutagenicity of Inhalable Particulate Matter in Santiago, Chile; Inhal. Toxicol. 2000, 12, Artaxo, P.; Oyola, P.; Martinez, R. Aerosol Composition and Source Apportionment in Santiago de Chile; Nucl. Instrum. Meth. B. 1999, 150, Gil, L.; King, L.; Adonis, M. Trends of Polycyclic Aromatic Hydrocarbon Levels and Mutagenicity in Santiago s Inhalable Airborne Particles in the Period ; Inhal. Toxicol. 2000, 12, Kavouras, I.G.; Lawrence, J.; Koutrakis, P.; Stephanou, E.G.; Oyola, P. Measurement of Particulate Aliphatic and Polynuclear Aromatic Hydrocarbons in Santiago de Chile: Source Reconciliation and Evaluation of Sampling Artifacts; Atmos. Environ. 1999, 33, Rappengluck, B.; Oyola, P.; Olaeta, I.; Fabian, P. The Evolution of Photochemical Smog in the Metropolitan Area of Santiago de Chile; J. Appl. Meteorol. 2000, 39, Sanhueza, P.; Vargas, C.; Jimenez, J. Daily Mortality in Santiago and its Relationship with Air Pollution; Rev. Med. Chile 1999, 127, Cifuentes, L.A.; Vega, J.; Kopfer, K.; Lava, L.B. Effect of the Fine Fraction of Particulate Matter Versus the Coarse Mass and Other Pollutants on Daily Mortality in Santiago, Chile; J. Air & Waste Manage. Assoc. 2000, 50, Lee, S.A.; Hastie, T.; Mancilla, P.F.; Astudillo, P.O.; Kuschner, W.G. Fine Particulate Air Pollution (PM 2.5 ) and Medical Visits for Lower Respiratory Tract Illness among Children in Santiago, Chile; J. Invest. Med. 2000, 48, 96A-96A. 13. Ilabaca, M.; Olaeta, I.; Campos, E.; Villaire, J.; Tellez-Rojo, M.M.; Romieu, I. Association between Levels of Fine Particulate and Emergency Visits for Pneumonia and Other Respiratory Illnesses among Children in Santiago, Chile; J. Air & Waste Manage. Assoc. 1999, 49, Journal of the Air & Waste Management Association Volume 57 July 2007

12 Sax et al. 14. Ostro, B.D.; Eskeland, G.S.; Sanchez, J.M.; Feyzioglu, T. Air Pollution and Health Effects: a Study of Medical Visits among Children in Santiago, Chile; Environ. Health Perspect. 1999, 107, Health Effects Institute. Understanding the Health Effects of Components of the Particulate Matter Mix. Progress and Next Steps in HEI Perspectives. Health Effects Institute: Cambridge, MA, Laden, F.; Neas, L.M.; Dockery, D.W.; Schwartz, J. Association of Fine Particulate Matter from Different Sources with Daily Mortality in Six US Cities; Environ. Health Perspect. 2000, 108, Jorquera, H.; Palma, W.; Tapia, J. An Intervention Analysis of Air Quality Data at Santiago, Chile; Atmos. Environ. 2000, 34, Koutrakis, P.; Sax, S.N.; Sarnat, J.A.; Coull, B.; Demokritou, P.; Oyola, P.; Garcia, J.; Gramsch, E. Analysis of PM 10, PM 2.5, and PM Concentrations in Santiago, Chile, from 1989 to 2001; J. Air & Waste Manage. Assoc. 2005, 55, Chaloulakou, A.; Kassomenos, P.; Spyrellis, N.; Demokritou, P.; Koutrakis, P. Measurements of PM 10 and PM 2.5 Particle Concentrations in Athens, Greece; Atmos. Environ. 2003, 37, Kavouras, I.G.; Koutrakis, P.; Cereceda-Balic, F.; Oyola, P. Source Apportionment of PM 10 and PM 2.5 in Five Chilean Cities Using Factor Analysis; J. Air & Waste Manage. Assoc. 2001, 51, U.S. Environmental Protection Agency. Air Quality Criteria for Particulate Matter, Vols. 1, 2 and 3. EPA 600/P-99/002aF-bF; U.S. Government Printing Office: Washington, DC, Querol, X.; Alastuey, A.; Ruiz, C.R.; Artinano, B.; Hansson, H.C.; Harrison, R.M.; Buringh, E.; Ten Brink, H.M.; Lutz, M.; Bruckmann, P.; Straehl, P.; Schneider, J. Speciation and Origin of PM 10 and PM 2.5 in Selected European Cities; Atmos. Environ. 2004, 38, Young, T.M.; Heeraman, D.A.; Sirin, G.; Ashbaugh, L.L. Resuspension of Soil as a Source of Airborne Lead near Industrial Facilities and Highways; Environ. Sci. Technol. 2002, 36, Wahlin, P.; Berkowicz, R.; Palmgren, F. Characterisation of Traffic- Generated Particulate Matter in Copenhagen; Atmos. Environ. 2006, 40, Lough, G.C.; Schauer, J.J.; Park, J.S.; Shafer, M.M.; Deminter, J.T.; Weinstein, J.P. Emissions of Metals Associated with Motor Vehicle Roadways; Environ. Sci. Technol. 2005, 39, Rutllant, J.; Garreaud, R. Meteorological Air-Pollution Potential for Santiago, Chile-Towards an Objective Episode Forecasting; Environ. Monit. Assess. 1995, 34, About the Authors Sonja Sax is an associate at Gradient Corporation. Petros Koutrakis is a professor of environmental sciences in the Department of Environmental Health at the Harvard School of Public Health. He is also the director of the Exposure, Epidemiology and Risk Program and the director of U.S. Environmental Protection Agency/Harvard University Center for Ambient Particle Health Effects. Pablo A. Ruiz Rudolph is a research associate at the Mario Molina Center Chile and a research fellow at Harvard School of Public Health. Francisco Cereceda is professor in the Department of Chemistry at the Federico Santa Maria Technical University. Ernesto Gramsch is associate professor in the Department of Physics at the University of Santiago. Pedro Oyola is director at the Mario Molina Center Chile, he is also a visiting professor at the University of São Paulo, Brazil, in the Faculty of Public Health and a researcher at the Federico Santa María Technical University. Please address correspondence to: Pablo A. Ruiz Rudolph, Centro Mario Molina Chile, Avenida del Valle 662, Office 202, Huechuraba, Santiago, Chile; phone: ; fax: ; pruiz@hsph.harvard.edu. Volume 57 July 2007 Journal of the Air & Waste Management Association 855