Weekday/Weekend Differences in Ambient Air Pollutant Concentrations in Atlanta and the Southeastern United States

Transcription

1 TECHNICAL PAPER ISSN J. Air & Waste Manage. Assoc. 56: Weekday/Weekend Differences in Ambient Air Pollutant Concentrations in Atlanta and the Southeastern United States Charles L. Blanchard and Shelley Tanenbaum Envair, Albany, CA Copyright 2006 Air & Waste Management Association ABSTRACT The authors quantified changes between mean weekday and weekend ambient concentrations of ozone (O 3 ) precursors (volatile organic compounds [VOC], carbon monoxide [CO], nitric oxide, and oxides of nitrogen [NO x ]) in Atlanta and surrounding areas to observe how weekend precursor emission levels influenced ambient O 3 levels. The authors analyzed CO, nitric oxide (NO), and NO x measurements from 1998 to 2002 and speciated VOC from 1996 to They observed a strong weekend effect in the Atlanta region, with median daytime (6:00 a.m. to 3:00 p.m. Eastern Standard Time) decreases of 62%, 57%, and 31%, respectively, in the ambient levels of NO, NO x, and CO from Wednesdays to Sundays, during the ozone season (March to October). They also observed significant decreases in ambient VOC levels between Wednesdays and Sundays, with decreases of 28% for the sum of aromatic compounds and 19% for the sum of Photochemical Assessment Monitoring Stations target compounds. Despite large reductions in O 3 precursor levels on weekends, day-of-week differences in O 3 mixing ratios in and near Atlanta were much smaller. Averaging overall O 3 -season days, the 1-hr and 8-hr mean peak daily O 3 maxima on Sundays were 4.5% and 2.3% lower, respectively, than their mean levels on Wednesdays (median of 14 site differences), with no sites showing statistically significant Wednesday-to-Sunday differences. When restricted to high-o 3 days (highest 3 peak O 3 days per day of week per site per year), the 1-hr and 8-hr Sunday O 3 mixing ratios were 11% and 10% lower, respectively, than their mean peak levels on Wednesdays (median of 14 site differences), with 6 of 14 sites showing statistically significant Wednesday-to-Sunday differences. The analyses of weekday/weekend differences in O 3 precursor concentrations show that different emission reductions than normally take place each weekend will be required to achieve major IMPLICATIONS The analyses of weekday/weekend differences in O 3 and O 3 precursor levels show that different emission reductions of O 3 precursors than normally take place each weekend will be required before major reductions in ambient O 3 can be achieved in the Atlanta area. Although regional-scale changes in O 3 levels might result from future control programs, it is sobering that Atlanta-area sites exhibited only minor weekend O 3 reductions in spite of the large and statistically significant weekend reductions of O 3 precursor concentrations. reductions in ambient ozone levels in the Atlanta area. INTRODUCTION The occurrence of generally comparable, or even higher, ambient concentrations of ozone (O 3 ) on Saturdays and Sundays than on other days of the week is commonly known as the weekend O 3 effect. 1 Because emissions of O 3 precursors, including volatile organic compounds (VOCs), oxides of nitrogen (NO x ), and carbon monoxide (CO), are lower on weekends than on weekdays, the weekend O 3 effect is counterintuitive. A weekend effect has been noted for many years and in a variety of locations. 2 6 It has been proposed that areas where O 3 formation is VOC limited tend to exhibit higher peak O 3 on weekends, because of a more pronounced weekend reduction of NO x than VOC emissions. 7 Other explanations for higher weekend peak O 3 values include possible higher precursor emissions on Friday evenings with carryover to Saturday, increased emissions of VOC in some areas on weekends (including possibly higher ratios of VOC/NO x ), changes in the times of occurrence of emissions on weekends, or increased reactivity of VOC emissions on weekends. The weekend effect in Southern California has been studied at length, 8 14 although differences of interpretation remain. Fujita et al. 10 and Lawson 1 conclude that weekend reductions of NO x concentrations allow O 3 to accumulate earlier in the day and to reach higher concentrations compared with weekdays and that proposed alternative hypotheses are not supported by ambient data and do not explain the weekend effect in Southern California. In contrast, Croes et al. 15 considered the available air quality data and photochemical models inadequate to conclusively determine the causes of the weekend O 3 effect in Southern California. Different day-of-week patterns of O 3 concentrations occur in Chicago, Philadelphia, and Atlanta. 16 In Atlanta, mean, median, and 90th percentile daily 1-hr O 3 maxima generally increased from Mondays or Tuesdays to higher levels on Fridays or Saturdays. 16 Other differences between Atlanta and Los Angeles that are of potential significance to the weekend effect in each area are the levels of NO x emissions from electric generating units throughout the Southeastern United States and Atlanta s location relative to other major cities. This paper provides additional analysis of the O 3 weekend effect in Atlanta and its surroundings. The authors quantify the changes between weekday and weekend ambient concentrations of O 3 and O 3 precursors, Volume 56 March 2006 Journal of the Air & Waste Management Association 271

2 the extension of the database to sites in Northern Alabama and Northern Georgia was intended to provide information on day-of-week concentration variations in areas potentially upwind of Atlanta on many days. Figure 1. Locations of monitoring sites. describe the effects of weekend precursor changes on ambient O 3 and particulate nitrate levels in Atlanta and surrounding areas, and evaluate possible explanations for the observed differences in the weekend behavior of O 3 and O 3 precursors. For most states, photochemical grid modeling is the primary planning tool for managing O 3 levels. In some cases, modeling will be supplemented by a weight-ofevidence demonstration, which might include analyses of trends or application of other analyses of observations to assess the effectiveness of possible emission control programs. The day-of-week variation in ambient concentrations is one of type of analysis that could be performed, along with others, as part of a complete weight-of-evidence demonstration. EXPERIMENTAL WORK Data Sources Data were obtained for the years to provide a multiyear record with focus on near-current conditions. U.S. Environmental Protection Agency (EPA) Aerometric Information Retrieval System (AIRS) (now Air Quality System) data were obtained for sites in Atlanta and in Fulton County and for sites in the neighboring counties of Douglas, Cobb, Forsyth, DeKalb, Clayton, Fayette, Coweta, Henry, Rockdale, Gwinnett, and Dawson (Figure 1). These data included measurements of O 3, nitric oxide (NO), NO x, and CO. O 3 data were also acquired for the nearest Clean Air Status and Trends Network sites (Georgia Station, GA, Sand Mountain, AL, and Coweeta, NC). A variety of gas-phase and particle measurements were obtained from four sites in the Southeastern Aerosol Research and Characterization (SEARCH) network 17 : Jefferson Street (Atlanta), Yorkville (GA), Centreville (AL), and Birmingham (AL). Particulate matter (PM) measurements were obtained from EPA Speciated Trends Network for all of the sites in Northern Georgia and Alabama. VOC data were available from Photochemical Assessment Monitoring Station (PAMS) measurements at four sites (Yorkville, South DeKalb, Tucker, and Conyers) for the years For both gas-phase and condensed-phase species, Statistical Comparisons Day-of-week averages were determined by hour of the day for hourly data (including CO, NO, and NO x ) and for 24-hr measurements (including particulate nitrate and sulfate). For O 3, day-of-week averages were determined for both peak 1-hr and peak 8-hr concentrations. For the hourly data, statistical comparisons were made for the hours of 6:00 a.m. Eastern Standard Time (EST; indicative of fresh emissions, typically having maximum morning concentrations) and 12:00 p.m. EST (midday, higher photochemical activity), for 3-hr intervals (6:00 9:00 a.m., 9:00 a.m. 12:00 p.m., and 12:00 p.m. 3:00 p.m. EST) and for daytime (6:00 a.m. 3:00 p.m. EST) averages. The O 3 comparisons were made using all of the O 3 - season days and for high-o 3 days, the latter facilitating a focus on conditions of greatest regulatory interest. As defined by EPA, the O 3 season in Georgia extends from March to October. 18 Twenty-one high-o 3 days were selected from each year by determining the top three O 3 days for each day of the week and each year. This procedure represents each day of the week and each year equally and provides a set of 9% of the 245 days from March to October. For each monitor, separate peak 1-hr and peak 8-hr high-o 3 data subsets were determined. NO x, VOC, nitric acid (HNO 3 ), and CO measurements were also restricted to the months of March to October, although in the case of VOCs, measurements were available only for the months of June to September. Differences in weekday and weekend concentrations of these compounds were computed using all of the days within the selected months to provide as large a dataset as possible for quantifying the changes in precursor levels on weekends. For NO, NO x, and reactive odd nitrogen (NO y ), a second set of day-of-week averages was also computed for the high 8-hr O 3 data subsets to determine whether day-of-week precursor differences occurred on high-o 3 days, as well as on average days. PM measurements were analyzed for all days of the year, because high-pm days often occur during winter months, and because PM samples are collected less often (typically, once every third or sixth day). All of the statistical tests refer to t tests of the differences between means. The data points constituting each mean were separated in time (e.g., Sunday noon compared with Wednesday noon) so that statistical assumptions of independence were not violated. Statistical methods exist for testing the significance of multiple comparisons so that the joint significance probability of all tests is controlled at a specified level (e.g., P 0.05) One such approach to simultaneous statistical inference assigns each of m comparisons an error rate of /m, which gives an experiment-wise error rate of. 20,21 Depending on the compound, the authors tested differences at 4 14 sites. For consistency, they used a fixed comparison-wise significance P value (P 0.01) to evaluate statistical significance of the differences in site means. 272 Journal of the Air & Waste Management Association Volume 56 March 2006

3 Using P 0.01, 1 statistical test in 100 is expected to yield an apparently significant result when no statistical difference actually exists. To facilitate comparisons, for each monitoring site and each specified pollutant and time of day, the differences between mean Wednesday and mean Sunday concentrations were expressed as the percentage decrease from the Wednesday mean: difference 100% (Wednesday mean Sunday mean)/(wednesday mean). Statistical Characterization of Weekday/Weekend Differences in Pollutant Levels Primary Pollutants. As determined from all of the O 3 -season days, the median daytime (6:00 a.m. 3:00 p.m.) declines of NO, NO x, and CO from Wednesday to Sunday were 62%, 57%, and 31%, respectively (Figure 2). The median decreases at 6:00 a.m. and 12:00 p.m. were similar to the decreases for the 3-hr averaging times of 6:00 a.m. 9:00 a.m. and 12:00 p.m. 3:00 p.m., respectively, and bracketed the median daytime (6:00 a.m. 3:00 p.m.) average decreases. Declines exceeding 30 40% were statistically significant, and all of the urban monitors exhibited statistically significant weekend decreases of NO, NO x, and CO during one or more time periods. Declining ambient concentrations of NO and NO x also occurred on high-o 3 days. Weekend declines of ambient NO, NO x, and CO concentrations could result from decreased emissions from gasoline vehicles, on-road and off-road diesel vehicles, stationary sources, or a combination of source types. The emission contributions from different source types in the Atlanta area constrain the possible emission reductions from the major source categories. According to the EPA 2001 national emission inventory, CO emissions derive predominantly from gasoline vehicles (79%), whereas gasoline and diesel motor vehicles contributed 18% and 25% of U.S. NO x emissions, respectively, or 43% combined (including both on-road and nonroad vehicles). 22 Within Fulton County (metropolitan Atlanta), on-road and nonroad motor vehicles accounted for 79% of NO x emissions (split as 38% diesel, 35% gasoline, 6% aircraft, railroads, and other) and 93% of CO emissions in 1999 (90% from gasoline vehicles). 23 Stationary sources in Fulton County accounted for the remaining 21% of countywide NO x emissions, so the 60% weekend reduction of ambient NO x levels could not have derived solely from changes in stationary-source emissions. However, SO 2 measurements made at the SEARCH site in Atlanta showed lower mean levels during morning and early afternoon hours on Sundays compared with Wednesdays (Figure 3), consistent with the occurrence of weekend decreases in operations of some commercial or industrial facilities. The 31% median reduction in ambient CO mixing ratios on Sundays could not have occurred unless gasoline vehicle CO emissions (90% of total CO emissions) were lower on Sundays than on Wednesdays, which also implies that NO x emissions from gasoline vehicles were lower on Sundays than on Wednesdays. However, gasoline vehicles were not the only contributor to weekend NO x reductions, because the ambient NO x reductions Figure 2. Median percent decreases in NO, NO x, and CO from Wednesdays to Sundays for three 3-hr periods (6:00 a.m. 9:00 a.m., 9:00 a.m. 12:00 p.m. and 12:00 p.m. 3:00 p.m.), hourly (at 6:00 a.m., 9:00 a.m., and 12:00 p.m.), and averaged over the period 6:00 a.m. 3:00 p.m. The data are from March to October, The medians were determined from all Atlanta-area AIRS monitoring sites with available data (Confederate Avenue, Conyers, South DeKalb, and Tucker for NO and NO x ; DeKalb Technical and Roswell for CO). The SEARCH site at Jefferson Street, Atlanta, was not included in the computation of the medians, but during the period it exhibited Wednesday to Sunday declines comparable to those shown here (NO, NO y, and CO declines at 12:00 p.m. of 64%, 51%, and 35%, respectively). ( 60%) on Sundays exceeded the gasoline-vehicle portion (35%) of NO x emissions in Fulton County. Because mobile sources accounted for 79% of NO x emissions in Fulton County, changes in the ambient NO x levels are expected to correlate with changes in CO and black carbon (BC), both of which derive primarily from mobile sources. 11 As noted, 90% of the Fulton County CO emissions are attributable to gasoline vehicles. In contrast, BC is found in both diesel and gasoline exhaust, although the BC emission rates are greater for diesel than for gasoline engines: depending on temperature, age, and engine condition, diesel vehicles may emit BC at 3 to 30 times the rate (grams per mile basis) as gasoline vehicles. 24 In the Denver area, mean BC emission rates Volume 56 March 2006 Journal of the Air & Waste Management Association 273

4 Figure 3. Diurnally averaged mixing ratios of O 3, CO, SO 2, and NO at Jefferson Street (Atlanta). All hourly measurements from the years during the months March to October were used in determining the averages. Each point is the average of 130 samples. from diesel vehicles averaged 12 times the BC emission rates from high-emitting gasoline vehicles and 490 times the BC emission rates from low-emitting gasoline vehicles. 11,25 The overall contributions of diesel and gasoline vehicles to ambient BC concentrations depend not only on emissions rates, but also on vehicle miles traveled. In the Denver area, enhanced chemical mass balance modeling using organic compounds as marker species yielded estimates of diesel to gasoline vehicle contributions to BC concentrations in the approximate ratios of 1:1 at one location and 3:1 at a second site. 25 In Los Angeles, BC was found to be a good indicator of diesel exhaust (albeit not a unique source marker) through comparison with vehicle activity data and highway sampling. 11 The median Wednesday to Sunday decline in 24-hr BC concentrations at urban sites in Atlanta and Birmingham was 31%, with statistically significant differences at three of the five urban sites (Table 1). The 24-hr BC declines on Sundays were smaller at the two nonurban sites of Centreville (16%) and Yorkville (13%). Hourly measurements of BC at SEARCH sites indicate that the reductions of BC on Sundays were larger during some time periods than others at both urban and rural sites. The median reductions of BC concentrations at the four SEARCH sites were 44% during the midday hours (6:00 a.m. to 3:00 p.m.) and 47% during the period 9:00 a.m. to 12:00 p.m. (Table 1). Again noting that gasoline vehicles generated 25 50% of ambient BC in Denver, whereas the observed median weekend reduction of CO levels was 31%, one would expect that weekend reductions of gasoline vehicle traffic in Atlanta might lower ambient BC concentrations by 5 15% on weekends. The observed 31% reductions of ambient concentrations of BC at urban sites indicate that diesel exhaust emissions were also lower on weekends than on weekdays. For comparison, weight-in-motion measurements and survey results in Los Angeles show that counts of heavy-duty truck traffic on freeways declined by 60 80% on weekends compared with midweek, whereas surface street traffic volumes (primarily light-duty vehicles) decreased by 15 30% on weekends. 9,13 Mean CO and NO x levels declined by 26% and 44%, respectively, on Sundays compared with midweek. 12 The authors have not attempted a precise quantification of the relative source-type reductions, but the observed weekend decreases in the ambient concentrations of BC and CO imply that both diesel and gasoline vehicle emission reductions contributed to the observed Sunday Table 1. Differences between mean BC concentrations on Wednesdays and Sundays computed for various times of day and temporal resolutions. No. Days Wednesday to Sunday Decrease (%) Site Location 24-hr Data 1-hr Data 6:00 a.m. 12:00 p.m. 6:00 a.m. 9:00 a.m. 9:00 a.m. 12:00 p.m. 12:00 p.m. 3:00 p.m. 6:00 a.m. 3:00 p.m. 12:00 a.m. 12:00 a.m N. Birmingham 32 0 NA NA NA NA NA NA Wylam 11 0 NA NA NA NA NA NA S. DeKalb 30 0 NA NA NA NA NA NA 37 BHM N. Birmingham CTR Centreville JST Jefferson St YRK Yorkville Notes: Positive numbers represent higher Wednesday concentrations. Statistically significant (P 0.01) differences are shown in italic typeface; The averages were determined from all available monitoring days during ; Sampling periods were: for AIRS/PM station sites (S. DeKalb/Atlanta GA, N. Birmingham, and Wylam/Birmingham AL), for 24-hr resolution SEARCH data (Birmingham AL, Centreville AL, Jefferson St./Atlanta GA, and Yorkville GA), and for 1-hr resolution carbon measurements at the four SEARCH sites. 274 Journal of the Air & Waste Management Association Volume 56 March 2006

5 Table 2. Differences between the mean Wednesday and mean Sunday ambient concentrations (6:00 a.m. 3:00 p.m.) of alkanes, alkenes, aromatics, benzene, isoprene, and the sum of PAMS target species. Wednesday to Sunday Decrease (%) Site AIRS No. Wednesdays Alkanes Alkenes Aromatics Benzene Isoprene Sum of Species South DeKalb Tucker Yorkville Conyers Mean Median Notes: Positive differences represent higher Wednesday concentrations; Statistically significant differences (P 0.01) are shown in italics; Data are from 1996 to reductions of ambient NO and NO x levels. More specifically, comparison of the ambient CO reductions on Sundays (median: 31%) with the gasoline-vehicle portion (35%) of NO x emissions in Fulton County suggests that the larger portion of the 60% Sunday reduction of NO x levels was because of weekend declines in diesel exhaust emissions. This conclusion is consistent with the results from Southern California, as well as with Atlanta-area declines in BC levels and the Denver results showing that diesel exhaust contributes as much or more to ambient BC levels as does gasoline engine exhaust. As a caveat, it is noted that both CO and BC may be transported over substantial distances, so urban CO and BC concentrations likely include regional background levels. The reductions of CO and BC emissions within Fulton County may have been greater than the corresponding declines in ambient concentrations. The ambient hydrocarbon measurements were more limited than the NO x and CO data but indicated that VOC declines occurred on Sundays, especially at urban locations. For the four Atlanta-area PAMS sites, the median midday (6:00 a.m. 3:00 p.m.) Wednesday to Sunday decreases in ambient hydrocarbon concentrations computed from all of the O 3 -season days were 16% for alkanes, 13% for alkenes, 28% for aromatics, 21% for benzene, 5% for isoprene, and 19% for the sum of PAMS target compounds (Table 2). The largest declines were observed for aromatics, including benzene (Figure 4), for which gasoline and gasoline vehicles are a known major source. The aromatics declines are of comparable magnitude to the previously discussed changes in CO levels. VOC/NO x Ratios. An important, although coarse, indicator of O 3 sensitivity to precursors is the ratio of VOC/NO x, with higher ratios indicating greater sensitivity to changes in NO x. 26,27 The differential changes in ambient NO x and VOC levels in Atlanta imply that weekend ratios of VOC/NO x differ from weekday ratios and suggest that local O 3 formation on weekends may be more NO x -sensitive than on weekdays. Considering the two urban locations, the mean VOC/NO x ratios increased from 1.3:1 to 2.3:1 at South DeKalb and from 2.7:1 to 4.4:1 at Tucker when computed from all of the O 3 -season days (Table 3), all of which are within the ranges historically considered to be VOC limited. 26,27 The later discussion of local O 3 formation will show that ratios of O 3 /HNO 3 and HNO 3 / NO y also increased on weekends. HNO 3, Nitrate, and Sulfate. The median Wednesday-to- Sunday daytime (6:00 a.m. 3:00 p.m.) decline in HNO 3 mixing ratios at the four SEARCH sites was 21% (Table 4), whereas the median 24-hr nitrate decrease at the same four SEARCH sites plus two Speciated Trends Network Figure 4. Diurnal profiles of mean ambient benzene levels at Yorkville, South DeKalb, Tucker, and Conyers during the period , shown for each day of the week. Volume 56 March 2006 Journal of the Air & Waste Management Association 275

6 Table 3. Mean daytime (6:00 a.m. 3:00 p.m. EST) concentrations of the sum of PAMS target compounds, NO x or NO y, and the ratios of VOC/NO x or VOC/ NO y at four locations, determined from all O 3 -season days having both VOC and NO x Data. Wednesday Sunday Site No. Days a VOC (ppbc) NO x (ppbv) VOC/NO x No. Days a VOC (ppbc) NO x (ppbv) VOC/NO x S. DeKalb Tucker Conyers Yorkville b Notes: The uncertainty limits are 1 SEM; a No. days with NO x or NO y data; See Table 2 for No. days with VOC measurements; b NO y VOC/NO y. sites was 19% (Table 5). Although the 6:00 a.m., 12:00 p.m., 6:00 a.m. 9:00 a.m., 9:00 a.m. 12:00 p.m., 12:00 p.m. 3:00 p.m., and 6:00 a.m. 3:00 p.m. HNO 3 levels declined on Sundays at all four of the SEARCH sites, only the 9:00 a.m. 12:00 p.m. decrease at Yorkville was statistically significant (P 0.01). None of the 24-hr Wednesday/Sunday PM nitrate concentration differences was statistically significant, although the differences were consistent, ranging from 15% to 25% at five of the six sites. Mean Wednesday/Sunday sulfate concentration differences were not statistically significant; three sites showed Sunday increases, and three showed decreases, with the median change being a 1.4% decline on Sunday. The results indicate that HNO 3 and PM nitrate levels decreased on Sundays but not in proportion to the magnitude of the NO x decline. O 3. Both peak 1-hr and peak 8-hr O 3 concentrations were lower on high-o 3 Sundays than on high-o 3 Wednesdays, with one exception; these decreases were statistically significant at 6 of the 14 sites that were studied (Table 6). For the high-o 3 days, the median change among the sites was a 9.5% decrease from the mean Wednesday peak 8-hr concentrations to the mean Sunday peak 8-hr concentrations; the mean Sunday 1-hr concentrations were 11% lower than the mean Wednesday 1-hr concentrations (Table 6). In contrast, the differences between average Wednesday and Sunday O 3 levels nearly disappeared when all of the days (March October) were included in the averages. For the all-days averages, the median change among sites was a decrease of 2.3% from the mean Wednesday peak 8-hr levels to the mean Sunday peak 8-hr values; the mean Sunday 1-hr mixing ratios were 4.5% lower than the mean Wednesday 1-hr mixing ratios (Table 6). None of these differences was statistically significant. The modest changes in peak O 3 levels on Sundays compared with Wednesdays contrast with the much larger reductions in O 3 precursor concentrations occurring on Sundays. O 3 Formation, Loss, and Accumulation Ambient O 3 levels are the net result of numerous atmospheric processes, any or all of which could exhibit changes on weekends compared with weekdays. Different processes may predominate during different periods of the day, and all must be considered to understand the O 3 response to precursor changes on weekends. Overnight Surface Carry-Over. Because the nocturnal inversion confines pollutants within a shallow surface layer, and fresh emissions tend to be lower at night, mixing ratios from midnight to 4:00 a.m. typically reflect surface carry-over of previous-day pollutants more than at other times of day. All of the sites showed lower mixing ratios of CO and NO and higher O 3 levels on Sundays than on Wednesdays from midnight to 4:00 a.m., as exemplified by the Jefferson Street site (Figure 3). Thus, during this time period, surface carry-over of O 3 precursors was lower on Sundays than on Wednesdays, whereas surface carryover of O 3 was greater. The higher levels of surface O 3 on Sundays were associated with lower levels of NO during the same hours (Figures 3 and 5), implying that reduced titration of O 3 by NO caused the higher Sunday O 3 surface carry-over. Ozone carry-over aloft need not have been affected in the same way because of nocturnal isolation of air aloft from surface-level NO x emissions. Table 4. Differences between mean Wednesday and mean Sunday HNO 3 concentrations at 6:00 a.m., 12:00 p.m., 6:00 a.m. 9:00 a.m., 9:00 a.m. 12:00 p.m., 12:00 p.m. 3:00 p.m., and 6:00 a.m. 3:00 p.m. HNO 3 Wednesday Sunday Decrease (%) Site City 6:00 a.m. 12:00 p.m. 6:00 a.m. 9:00 a.m. 9:00 a.m. 12:00 p.m. 12:00 p.m. 3:00 p.m. 6:00 a.m. 3:00 p.m. BHM Birmingham, AL a CTR Centreville, AL JST Jefferson St/Atlanta, GA YRK Yorkville, GA Notes: Positive numbers represent higher Wednesday concentrations; Statistically significant (P 0.01) differences are shown in bold typeface; The averages were determined from days during March to October of ; All locations are SEARCH sites; a Data available from October 2000 to December Journal of the Air & Waste Management Association Volume 56 March 2006

7 Table 5. Differences between mean Wednesday and Sunday nitrate and sulfate concentrations. AIRS or SEARCH City No. Days Wednesday to Sunday Decrease (%) Nitrate Sulfate N. Birmingham, AL S. DeKalb/Atlanta, GA BHM Birmingham, AL CTR Centreville,AL JST Jefferson St/Atlanta, GA YRK Yorkville, GA Notes: Positive numbers represent higher Wednesday concentrations. None of the differences are statistically significant (P 0.01). The averages were determined from all days during (AIRS/PM station: S. DeKalb/ Atlanta GA and N. Birmingham) and (SEARCH: Birmingham AL, Centreville AL, Jefferson St./Atlanta GA, and Yorkville GA). Fresh Emissions. Diurnal profiles indicate that fresh emissions of CO and NO were greater from 6:00 a.m. to 8:00 a.m. on Wednesdays than on Sundays (Figure 3). Because O 3 reacts with NO, O 3 cannot accumulate until NO levels fall below O 3 mixing ratios. 10 As a result of lower levels of NO on Sunday mornings, less O 3 was lost to NO titration on Sundays. O 3 accumulation began earlier on Sundays than on Wednesdays, as indicated by the times when morning O 3 concentrations first exceeded the ambient NO levels (Figure 5). At two sites, these times were 1 hr earlier on Sundays compared with Wednesdays; at two other sites, NO levels were always lower than O 3 concentrations on Sundays but not on Wednesdays. O 3 concentrations exceeded NO levels on all of the days at Yorkville, but even there, NO levels during the hours of 6:00 a.m. to 8:00 a.m. were lower on Sundays than on Wednesdays. Carry-Over Aloft. Ozone mixing ratios increased most rapidly during the hours from 9:00 a.m. to 12:00 p.m. (Figures 3 and 5). On Sundays, the mean temporal rate of O 3 mixing ratio increase from 9:00 a.m. to 12:00 p.m. was lower than on Wednesdays, so that by 12:00 p.m. the mean Sunday and Wednesday O 3 levels were nearly identical at most sites, although O 3 levels at 9:00 a.m. on Sundays exceeded those at 9:00 a.m. on Wednesdays (Figures 3 and 5). Concentration increases during morning hours could reflect either or both O 3 formation or increasing mixing depth and incorporation of higher O 3 levels from aloft (as noted previously, because aloft air masses are isolated from surface emissions of NO overnight, higher O 3 levels could have persisted above the nocturnal inversion). Separation of the effects of mixing from O 3 formation is difficult. The authors did not have vertical soundings, but they observed that average surface wind speeds increased most rapidly between the hours of 7:00 a.m. and 10:00 a.m., which suggests that vertical mixing effects need to be considered especially during this time period. Additional analyses discussed next indicate that O 3 transport from upwind was greater on Wednesdays than on Sundays. Regional Transport. The authors evaluated weekend changes in O 3 transport into the Atlanta area by comparing weekend to weekday peak 8-hr O 3 levels at upwind and downwind sites. The specific question to be addressed was: are day-of-week differences in O 3 mixing ratios within and downwind of Atlanta attributable to day-ofweek differences in O 3 production within the metropolitan area or to day-of-week differences in O 3 transport from upwind areas? Peak 8-hr O 3 levels typically occurred between the hours of 10:00 a.m. and 7:00 p.m., an interval during most of which substantial vertical mixing is typical. They used surface wind directions to select subsets Table 6. Differences between mean Wednesday and Sunday peak 1-hr and 8-hr concentrations. No. Days Wednesday to Sunday Decrease (%) Top 3 per Day of Week All Days AIRS/ SEARCH/ CASTnet City Sunday Wednesday 8-hr O 3 1-hr O 3 8-hr O 3 1-hr O S. DeKalb/Decatur Tucker Douglasville Fayetteville Confederate/Atlanta Brunswick Lawrenceville Conyers CTR Centreville a JST Jefferson St/Atlanta a YRK Yorkville a COW137 Coweeta GAS153 Georgia Station SND152 Sand Mountain Mean Median Notes: CASTnet indicates Clean Air Status and Trends Network; Positive numbers represent higher Wednesday concentrations; Statistically significant (P 0.01) differences are shown in bold typeface; The averages were determined from both all days and high-o 3 days during ; All sites listed had 3 years of data within the study period; a Volume 56 March 2006 Journal of the Air & Waste Management Association 277

8 Figure 5. Diurnally averaged mixing ratios of O 3 and NO on Wednesdays and Sundays. of days when it was reasonable to characterize one group of sites as upwind from Atlanta and another as downwind and confirmed the characterizations using HYSPLIT back trajectories. 28 The transport analyses were carried out for high-o 3 days (top three peak 8-hr daily maximum days per day of week per year for each site). For each high-o 3 day, hourly surface wind speed and direction were obtained for the three SEARCH sites (Jefferson Street, Yorkville, and Centreville) and three Clean Air Status and Trends Network sites (Georgia Station, Sand Mountain, and Coweeta). The surface meteorological data from these six sites were also used for O 3 sites lacking meteorological measurements (Yorkville for Douglasville, Georgia Station for Fayetteville, and Jefferson Street for Confederate Avenue, Tucker, DeKalb, Conyers, and Gwinnett; Figure 1). The authors computed 24-hr vector average wind speed and direction for the high-o 3 days for each site (from the hourly averages with start times of 7:00 p.m. on a previous day to 6:00 p.m. on each high-o 3 day). For each monitoring site, they then classified the 24-hr surface wind direction according to quadrant of origin (northeast, southeast, southwest, and northwest). The HYSPLIT model was run using 3 trajectory levels (100, 500, and 1000 m) for 10 days with vector-average surface winds from the northwest quadrant and 10 with surface winds from the southwest quadrant, randomly chosen from high-o 3 days (top three 8-hr days per day of week) during June to August, Trajectories were carried back from DeKalb County airport for 48 hr at 6-hr time intervals. The three trajectory heights remained reasonably coherent (within the classification of transport quadrants) and supported the classifications into quadrants using surface winds. 278 Journal of the Air & Waste Management Association Volume 56 March 2006

9 Figure 6. Mean daily 8-hr O 3 maxima by day of week for high-o 3 days at downwind sites (Conyers and Gwinnett), The data have been averaged across monitoring locations representing far upwind, near upwind, and urban/downwind locations. The error bars are 1 SEM. No error bars are shown for averages based on one sampling day. (a) The days with winds from the northwest had 24-hr vector-average surface wind directions of The far upwind site was Sand Mountain (Clean Air Status and Trends Network). The near upwind sites were Douglasville ( ) and Yorkville ( ; PAMS Type 1). The urban sites were Jefferson Street (SEARCH), Georgia Tech, and Confederate Avenue ( ). The downwind sites were DeKalb ( ; PAMS Type 2) and Conyers ( ; PAMS type 3). (b) The days with winds from the southwest had 24-hr vector-average surface wind directions of The far and near upwind sites were Centreville (SEARCH) and Douglasville ( ), respectively. The urban/downwind sites were Jefferson Street (SEARCH), Tucker ( ; PAMS type 2A), and Gwinnett ( ). The urban/downwind sites exhibited mean mixing ratios of ppbv, whereas the far-upwind sites showed mean mixing ratios of ppbv, thus indicating that transported O 3 made a large ( 80%) contribution to peak O 3 levels in and near Atlanta (Figure 6). Note that the Atlanta urban plume could itself have contributed to the regional background depending on circulation patterns, as is suggested by the partially elevated mean O 3 levels at the nominally upwind sites of Yorkville, Douglasville, and Fayetteville relative to Sand Mountain, Georgia Station, and Centreville. All of the back trajectories for the cases with winds from the northwest quadrant extended 48 hr back into areas to the northwest without recirculation; 30% of the back trajectories arriving from the southwest exhibited a trajectory rotation extending to the southeast of Atlanta and suggestive of recirculation around a high-pressure system. Other than Atlanta, nearby urban areas, including, for example, Chattanooga, could have contributed to the observed regional background when winds were from the northwest quadrant. Volume 56 March 2006 Journal of the Air & Waste Management Association 279

10 Previous studies have shown that transported O 3 contributes substantially to the magnitude of peak O 3 concentrations in the major metropolitan areas of the Southeastern United States. 29 The differences between the mean 8-hr peak O 3 levels at the urban/downwind sites and the far upwind sites are interpreted here as indicating the magnitude of the mean contribution of the Atlanta urban plume on high-o 3 days. These differences were ppbv (Figure 6). The differences between the far upwind and urban/downwind O 3 mixing ratios varied somewhat among the days of the week, but none of the differences was statistically significant (Figure 6). These analyses, therefore, show no statistically significant variation in the amounts of local O 3 production in Atlanta by day of the week. However, other comparisons of day-of-week differences in ratios of reactive species, discussed in the next section, indicate that day-of-week differences occurred in the rates of local O 3 formation at sites in Atlanta. The mean far upwind concentrations were lower on Sundays than on Wednesdays, Thursdays, or Fridays, although the differences were not statistically significant (Figure 6), implying that regional carry-over of O 3 was lower on Sundays than on Wednesdays through Fridays. The mean O 3 mixing ratios at the far upwind sites were marginally lower ( 1 3 ppbv) on Mondays and Tuesdays than on Sundays, suggesting that some of the benefits of weekend emission reductions in upwind areas appeared on Mondays or Tuesdays, rather than on Sundays, that is, some time lag occurred in the response of regional-scale O 3 levels to regional-scale weekend reductions of O 3 precursor emissions. However, the differences between the mean Sunday and the mean Monday or Tuesday O 3 levels were small ( 1 3 ppbv) and were not statistically significant. The differences between the minimum and maximum day-of-week means at the upwind sites indicate that the regional contribution varied by ppbv among days of the week (Figure 6). Under both northwesterly and southwesterly transport patterns, the mean day-of-week concentrations at far upwind, upwind, and urban/downwind sites followed approximately parallel lines (Figure 6). Although we grouped sites into far upwind, near upwind, and urban/ downwind locations for clarity, the results for individual sites were also approximately parallel. The parallelism (within the standard errors of the means) implies that the day-of-week differences in mean O 3 concentrations at Atlanta sites were partially or wholly driven by day-ofweek differences in upwind transport concentrations. That is, the magnitudes of the day-of-week variations at downwind sites were approximately the same as those occurring at upwind sites, because regional O 3 levels substantially exceeded the amounts of local O 3 formation and because the amount of local O 3 formation did not show statistically significant day-of-week variations. Local O 3 Formation. The SEARCH sites provide measurements of one of the most important NO x reaction products, namely, HNO 3, as well as NO y. For these sites, it is, thus, possible to compare O 3 levels to concentrations of NO x reaction products and to compare product levels to total oxidized nitrogen species concentrations. A key reaction product is HNO 3, which is produced by reaction of the HO radical with nitrogen dioxide (NO 2 ). This reaction removes a radical and an NO 2 molecule. Because each conversion of NO 2 to NO produces an O 3 molecule, with some loss processes, removal of an NO 2 molecule ends the contribution of that molecule to O 3 production (some reaction products, e.g., peroxyacetyl nitrate, can serve as a reservoir of NO 2, however, because a reverse reaction may occur). The ratio of O 3 to HNO 3 concentrations provides a relative measure of the mean number of O 3 molecules produced per NO x molecule. The ratio of HNO 3 to NO y provides another relative measure of the amount of NO x reaction product (the ratio of NO x to NO y is more indicative of the amount of reactant to the sum of reactants and products, but the available NO 2 record was much shorter than the HNO 3 record). The mean Sunday HNO 3 levels were lower than Wednesday levels (Table 4). However, both central city and rural SEARCH sites exhibited higher ratios of HNO 3 to NO y on Sundays compared with Wednesdays. Because the atmospheric lifetime of HNO 3 is of the order of hours to a day, these day-of-week differences imply that a greater fraction of the NO and NO 2 was converted to HNO 3 by a specified hour on Sundays compared with the same hour on Wednesdays. This result is consistent with the observed increase in ambient VOC/NO x on Sundays (Table 3), which suggests the occurrence of a more reactive mix on weekends. For brevity, we refer to the mean number of O 3 molecules produced per NO x molecule as O 3 production efficiency. 30 An increase in the ratio of O 3 to HNO 3 concentrations indicates that the efficiency of O 3 production was greater on Sundays than on Wednesdays (Figure 7). At Jefferson Street, the O 3 /HNO 3 ratios increased from 18:1 on Wednesdays to 24:1 on Sundays during the afternoon (sample start hours 12:00 p.m. 4:00 p.m.), indicating an 35% increase in the number of O 3 molecules produced per NO x. The increase in O 3 production efficiency acted in opposition to the decreases of NO emissions on Sundays. Modeling studies confirm the importance of the increase in O 3 production efficiency as NO x emissions decline. 31 The ratios of O 3 /NO y or O 3 /NO x were also higher on weekends than on Wednesdays at other monitoring sites. Ratios of O 3 /CO were greater on Sundays compared with Wednesdays at the three locations having CO data (Jefferson Street, Yorkville, and DeKalb), and on Saturdays relative to Wednesdays at Jefferson Street and DeKalb. Potential O 3 Responses to Emission Controls The analyses discussed previously indicate that local NO x emission reductions on the order of 60% did not change the amount of O 3 formed locally, suggesting that local O 3 formation was not typically limited by the availability of NO x. Two other indicators support this conclusion. First, the mean 6:00 a.m. 3:00 p.m. VOC/NO x ratios were low (1.3:1 to 5.8:1) at all of the PAMS monitoring sites on both Wednesdays and Sundays (Table 3; the 6:00 a.m. 9:00 a.m. ratios ranged from 1:1 to 5.1:1). These ratios underrepresent VOC, because they are based on the sum of PAMS target compounds; the mean ratio of the 280 Journal of the Air & Waste Management Association Volume 56 March 2006

11 Figure 7. Hourly ratios of the mean mixing ratios of O 3 and HNO 3 on Wednesdays and Sundays. Hourly averages were determined from measurements made at the SEARCH sites at Jefferson Street (Atlanta; ), Yorkville (GA; ), Birmingham (AL; ), and Centreville (AL; ) during the months of March through October, with each point computed as the average of 30 hourly measurements per year. target compounds to the measured total VOC in this dataset was Historic EPA guidance has indicated that 6:00 a.m. 9:00 a.m. ratios of ambient VOC/NO x less than 8:1 are indicative of VOC-limited conditions. 26,27 However, photochemical aging is as important as VOC/NO x ratios in determining the sensitivity of O 3 to changes in VOC or NO x emissions, 32 so that variations in O 3 response to emission reductions should be expected depending on the hour of the day, downwind distance, meteorological conditions and their relation to photochemical aging, and the influence of highly reactive VOCs (e.g., isoprene). A second indicator of where and when VOC or NO x limits local O 3 formation is the extent of reaction estimated from a smog production algorithm The authors calculated this indicator on an hourly basis using all of the available data, and they observed differences among hours, days of the week, and types of days (i.e., high O 3 or not). On high-o 3 Wednesdays, the mean hourly extent of reaction remained 60% (VOC-limited conditions; refs ) before 10:00 a.m. at all of the sites (Figure 8). By 12:00 p.m. and until 6:00 p.m., mean extent exceeded 80% (NO x limited ) at the nonurban sites of Yorkville and Conyers. At the suburban sites of DeKalb and Tucker, extent varied from 60% to 80% (transitional; refs ) from 12:00 p.m. to 6:00 p.m., whereas at Jefferson Street the extent remained 60% during most hours (Figure 8). In contrast to these high-o 3 days, the mean extent of reaction averaged 40% (VOC limited) from 12:00 p.m. to 6:00 p.m. at all of the sites when the averages were determined from all of the O 3 -season days. According to this calculation, therefore, high-o 3 days exhibited a different sensitivity to O 3 precursors than did other days, and the sensitivity of the peak hour was different than the sensitivity of the 8 hr comprising the 8-hr Volume 56 March 2006 Journal of the Air & Waste Management Association 281

12 Figure 8. Hourly mean extent of reaction and O 3, NO, NO x, and NO y mixing ratios on high-o 3 Wednesdays (top three peak 8-hr O 3 -concentration Wednesdays per year), For Yorkville and Jefferson Street, two alternative formulations of the extent of reaction were calculated using the measurements of NO y and NO x ; the NO x data were available only for 2001 and 2002 at Jefferson Street and 2002 at Yorkville. For the remaining three sites, upper and lower bounds of the extent of reaction were calculated using the available NO x measurements (which overestimate true NO 37 x ), and a best estimate is presented as the mean of the bounds. 34,35 O 3 peaks. In addition, the mean extent of reaction estimates between 12:00 p.m. and 6:00 p.m. on high-o 3 Sundays were higher than during corresponding hours on high-o 3 Wednesdays by 10 20% at Tucker and DeKalb, a result that is directionally consistent with the observed increase in VOC/NO x ratios at those sites on Sundays compared with Wednesdays (Table 3). Increases in the mean extent of reaction from Wednesdays to Sundays were also observed for all of the sites when the means were determined from all O 3 season days. Although the Atlanta weekend effect indicates that local NO x emission reductions did not reduce the amount of O 3 formed locally, the response of locally formed O 3 to future local emission controls could differ from the present weekend response, because the emission changes occurring from weekdays to weekends might not exactly reflect the emission reductions characteristic of control plans. The authors observed median daytime (6:00 a.m. 3:00 p.m. EST) decreases of 62%, 57%, and 31%, respectively, in the ambient concentrations of NO, NO x, and CO from Wednesdays to Sundays during the O 3 season (March to October). They also observed decreases of 28% for the sum of aromatic compounds and 19% for the sum of PAMS target compounds. In comparison, the Georgia Department of Natural Resources projects mobile-source NO x emissions to decline from 304 tons per day in 2004 to 192 tons per day in 2010 (37%) in a 13-county area around Atlanta and mobile-source VOC emissions to decline from 164 tons per day in 2004 to 112 tons per day in 2010 (32%). 38 For the state of Georgia, EPA projects that the recently adopted Clean Air Interstate Rule will yield stationary-source NO x reductions of 23% by 2009 and 38% by 2015 relative to In the EPA air quality modeling analyses of the effects of the Clean Air Interstate Rule, Georgia VOC and NO x emissions from all sources were projected to decline by 17% and 36%, respectively, from 2001 to In overall magnitude, therefore, the expected reductions of mobile-source and stationarysource O 3 precursor emissions between 2001 and 2010 are somewhat less than the weekend reductions in ambient 282 Journal of the Air & Waste Management Association Volume 56 March 2006

13 concentrations of NO, NO x, CO, and VOCs that are reported here. However, the locations and times of day exhibiting future emission reductions might not correspond closely with the locations and times that now show weekend reductions of ambient precursor concentrations. For example, the emission projections include stationarysource NO x emission reductions from nonurban sources, whereas the current weekend reductions of ambient NO x may derive primarily from weekend decreases in motor vehicle traffic. In contrast to the small changes that were observed in weekend relative to weekday O 3 levels, EPA modeling studies predict that the fourth-highest annual maximum peak daily 8-hr O 3 levels in Fulton County will decline from 99 ppbv in to 85 ppbv in Ambient monitoring data now show that the corresponding peak 8-hr O 3 values in Fulton County dropped to 93 ppbv in and 91 ppbv in It is possible that the types of emission reductions that were modeled by EPA might lower O 3 levels more effectively than did weekend reductions of urban VOC and NO x levels that were observed; for example, geographically widespread emission reductions might reduce regional O 3 concentrations, whereas the weekend decreases of O 3 precursors that were observed in Atlanta did not change the amount of O 3 transported into the Atlanta area from upwind regions. However, the observations could also imply that projected future O 3 decreases may be larger than what will actually occur. Potential PM Nitrate Responses to Emission Controls As demonstrated above, HNO 3 and PM nitrate levels decreased in response to lower NO x levels on Sundays but not in proportion to the magnitude of the NO x decline ( 20% compared with 60%). Future changes in PM nitrate levels will depend on changes in SO 2 emissions and sulfate concentrations, as well as on changes in NO x emissions and HNO 3 concentrations. In addition, changes in ammonia levels may affect PM nitrate concentrations. Present cool-season concentrations of PM nitrate in the Southeastern United States are low and tend to be limited by the availability of ammonia. 42,43 With declining sulfate concentrations, future PM nitrate levels could become more responsive to changes in ambient HNO 3 because PM nitrate formation would then be less ammonia limited. Model simulations show small predicted changes in PM nitrate in response to future changes in ambient levels of HNO 3, 43 and the actual changes in PM nitrate levels will depend also on the responses of HNO 3 to NO x reductions. CONCLUSIONS Future O 3 concentrations in Atlanta will depend on the responses of both transported and locally formed O 3 to emission controls. Regional O 3 levels were 80% of the ambient O 3 concentrations observed in and downwind of Atlanta, with the average O 3 mixing ratios at sites in Atlanta being ppbv higher than the levels measured at far upwind locations. Because of the substantial influence of regional, upwind O 3 on Atlanta s observed O 3 levels, the day-of-week differences observed in Atlanta were attributed in large measure to day-of-week differences in upwind transport conditions. Because transported O 3 may have formed over a period of many days, including both weekdays and weekends, the differences between weekday and weekend concentrations of O 3 do not necessarily represent the changes that might occur in response to regional-scale emission reductions. However, our analyses indicate that local emission reductions on the order of 30 60% did not change the amount of O 3 formed locally, and, on weekends, O 3 began forming 1 hr earlier, and more O 3 formed per unit of NO x. The peak O 3 hours did not shift to an earlier mean time on weekends as might be expected if lower weekend NO x levels were limiting O 3 production. Because NO levels were lower on Sundays, overnight surface carry-over of O 3 was greater, and morning titration of O 3 by NO was reduced so that local O 3 accumulation began earlier on Sunday mornings than on Wednesdays. During midday hours, ratios of O 3 /NO x,o 3 /NO y, and O 3 /HNO 3 were greater on Sundays than on Wednesdays, indicating that, in addition to the titration effect, the mean number of O 3 molecules formed per ambient NO x molecule increased on Sundays. The available data did not allow for the evaluation of the effects of changing levels of NO x on hydroxyl radical concentrations or whether O 3 formation rates changed with respect to specific VOC compounds. In overall magnitude, the projected reductions of mobile-source and stationary-source O 3 precursor emissions in Georgia between 2003 and 2010 are less than, or comparable to, the weekend reductions in ambient concentrations of NO, NO x, CO, and VOCs that are reported here. The locations and times of day exhibiting emission reductions in the future may differ from the locations and times that now show weekend reductions of ambient precursor concentrations. Nonetheless, these results have important implications for the potential efficacy of planned emission reductions. ACKNOWLEDGMENTS The authors thank the Coordinating Research Council for providing funding and the Council s reviewers who offered suggestions that significantly improved this study. REFERENCES 1. Lawson, D.R. Forum: The Weekend Ozone Effect-the Weekly Ambient Emissions Control Experiment; EM, July 2003, Cleveland, W.S.; Graedel, T.E.; Kleiner, B.; Warner, J.L. Sunday and Workday Variations in Photochemical Air Pollutants in New Jersey and New York; Science, 1974, 186, Elkus, B.; Wilson, K.R;. Photochemical Air Pollution: Weekend-Weekday Differences; Atmos. Environ. 1977, 11, Graedel, T.E.; Farrow, L.A.; Weber, T.A. Photochemistry of the Sunday Effect; Environ. Sci. Tech. 1977, 11, Hoggan, M.; Hsu, M.; Kahn, M.; Call, T. Weekday/weekend Differences in Diurnal Variation in Carbon Monoxide, Nitrogen Dioxide, and Ozone-Implications for Control Strategies. In Proceedings of the 82nd Annual Meeting & Exhibition, Air & Waste Management Association; A&WMA: Pittsburgh, PA, 1989; pp Lebron, F. A Comparison of Weekend-Weekday Ozone and Hydrocarbon Concentrations in the Baltimore-Washington Metropolitan Area; Atmos. Environ. 1975, 9, Altshuler, S.L.; Arcado, T.D.; Lawson, D.R. Weekday vs. Weekend Ambient Ozone Concentrations: Discussion and Hypotheses With Focus on Northern California; J. Air & Waste Manage. Assoc. 1995, 45, Blanchard, C.L.; Tanenbaum, S.J. Differences Between Weekday and Weekend Air Pollutant Levels in Southern California; J. Air Waste Manage. Assoc. 2003, 53, Volume 56 March 2006 Journal of the Air & Waste Management Association 283