• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

What Have We Learned From Highly Time-Resolved Measurements During EPA's Supersites Program and Related Studies?

Posted on: Saturday, 9 February 2008, 03:00 CST

By Wexler, Anthony S Johnston, Murray V

ABSTRACT A wide range of new and exciting highly time-resolved instruments were deployed during the U.S. Environmental Protection Agency (EPA) Supersite program and related studies that occurred during the same time period. These measurements elucidated the temporal variation of a suite of gas-phase species, particle physical properties, and size-resolved particulate chemical composition. Because the temporal resolution was so high, concentration and size distribution changes as short as 1 min or less were discerned. Often data from multiple instruments were correlated with each other and with meteorological measurements, and these correlations enabled conclusions to be drawn about the photochemical activity of the atmosphere, the location of point sources, and even the emissions characteristics of these sources. For instance, rapid changes in particulate matter (PM) concentration were due to meteorological conditions, emissions, and plume excursions that led to increases in nitrate, sulfate, and organic carbon concentrations. This paper summarizes the conclusions that have been reached, to date, using these new, highly time-resolved instruments, and demonstrates their promise for future studies.

INTRODUCTION

Numerous national and international government bodies regulate particulate air pollutants, often in standards related to PM^sub 10^ (particulate matter [PM] smaller than 10 [mu]m) and PM^sub 2.5^ (PM smaller than 2.5 [mu]m). In the United States, two temporal time scales are employed for these standards-1 day and 1 yr. That is, exceedances of the standards are based on daily or annually averaged concentrations exceeding a regulatory limit. Because these standards require local governments in the United States to monitor daily and annually averaged PM^sub 10^ and PM^sub 2.5^, monitors that record daily averages of these values are located throughout the United States and most of the developed world.

Over the last decade, many governmental agencies (most notably the U.S. Environmental Protection Agency [EPA]) and instrumentation companies have supported development and deployment of instruments that monitor the time-resolved concentration of PM^sub 10^ and PM^sub 2.5^, their chemical components, and their gas-phase precursors on time scales much shorter than the 1-day regulatory time scale. This research is motivated by several increasingly important factors. First, it has long been realized that sampling daily averages once every third or sixth day loses considerable information between sampling days, as well as loses any information on diurnal changes in PM pollutant concentrations. Highly time- resolved sampling would allow for both of these phenomena to be measured and, for example, help reduce uncertainty in determining compliance with the PM National Ambient Air Quality Standards.1 Second, a sampling of recent studies suggests that health effects due to inhalation of particulate matter may be at least partially the result of short-term exposures. For instance, Zeka et al.2 and Dockery et al.3 suggest a 2- to 3-day lag in health effects, whereas Urch et al.4 observe changes in cardiac endpoints with exposures as short as 2 hr. Similarly, Ruidavets et al.5 demonstrate a correlation of 1-2 days of ozone exposure to myocardial infarction. The lack of highly time-resolved instrumentation for measuring size and chemical composition is one limitation to uncovering epidemiological associations between short-term pollutant concentrations and health effects.

Highly time-resolved measurements may also assist in identifying emissions sources. Source apportionment is often performed by analysis of elements, water-soluble ions, and carbonaceous material in particulate matter-combinations of certain components are often associated with certain emissions sources or source categories. High time-resolution sampling enables measurement of shortterm duration variations in pollutant concentration due to emissions (e.g., exhaust from a single diesel bus) as well as meteorological variations influencing plume transport and magnitude. Because there are fewer opportunities for particles from one source to be mixed with those from another, source apportionment is much more straightforward. These high-frequency modulations (a few hours or less) are lost in longer-term (daily average) measurements. Thus, high time-resolution data provide an additional dimension in source apportionment, where source variations and meteorological variables can be included in the analysis. This concept has been exploited in the past.89 The recent development of other complementary and more sensitive methods has allowed sample sizes to be reduced, facilitating a wider range of applications.

Atmospheric dynamics is fundamentally different over temporal scales shorter than about 1 hr. Wind spectrum time scales in the atmosphere range from seconds to days but a natural break in the power spectrum occurs around 1 hr-shorter time scales are attributable to turbulence and longer ones are due to oscillations. 6 Below this temporal scale, turbulence causes the constituents to become well mixed. Measuring changes on this time scale elucidates photochemistry or sharp spatial gradients due to plumes and fronts in combination with high wind speeds or rapid changes in wind direction. As a result of this 1-hr spectral gap, photochemical aerosol chemical transport models (CTMs) time-split their various operators (e.g., diffusion, advection, photochemistry, aerosol dynamics) typically one to a few times per hour and report atmospheric pollutant concentrations every hour. Thus, reporting atmospheric constituent measurements hourly is ideal for validating photochemical aerosol CTMs.7

EPA has developed three levels of national PM measurement networks.8 The base network primarily serves for regulatory compliance so it monitors daily average PM^sub 10^ and PM^sub 2.5^ mass every third day at over 1000 sites. The Speciation Trends Network9 (STN) along with Interagency Monitoring of Protected Visual Environments10 (IMPROVE) network provides more detailed chemical speciation of PM^sub 2.5^ at over 260 urban and 175 rural monitoring sites where daily samples are collected on either a once every 3- (STN) or 6- (IMPROVE, and state and local networks) day schedule. The Supersites program (www.epa.gov/ttn/amtic/supersites.html) provides the greatest level of detail and, in conjunction with other on-going programs, is one of the most robust U.S. national air quality measurement programs to date.11 The Supersite locations were strategically placed in urban-focused regional areas in the United States, each with differing atmospheric pollution conditions resulting from variations in source emissions and meteorology. The locations included Atlanta, GA; Fresno and Los Angeles, CA; Pittsburgh, PA; New York, NY; Baltimore, MD; Houston, TX; and St. Louis, MO. Field measurements and laboratory studies were conducted for 1-4 yr at most sites; some are still operational. Although each site had its own specific objectives, a common objective for most sites was to develop, evaluate, and advance measurement technologies for PM mass, its chemical and physical properties, and related species, with a focus on highly time-resolved methods.

Several new and exciting instruments were developed, tested, and deployed at one or more of the Supersites-these instruments are summarized elsewhere in this issue.12,13 In this work, we summarize many of the unique findings made possible by the subset of these instruments that make highly time-resolved measurements, defined here as those that report parameters on time scales of about 1 hr or faster. Although some of these instruments measure the concentration of gas-phase precursors, most measure the chemical and/or physical properties of PM^sub 2.5^ or other particulate size fractions. Some of the advantages of these highly time-resolved data accrue when they are combined with other data, both highly time resolved and time averaged. Additional advantages are obtained when data from suites of instruments are further combined with atmospheric models. This progression, from using data measured by one or two instruments, to using multiple datasets and models, leads to increasingly complex and intriguing deductions with reduced uncertainty concerning pollutant emissions, transport, transformation, and source-receptor relationships that would not be possible without the contribution of highly time-resolved measurements. Thus we have organized this review roughly along a scale of increasing use of multiple measurements or measurements and models together to reach conclusions about atmospheric transport and transformation of pollutants that would not be possible without at least some of the measurements being highly time-resolved. We begin in Section 1 with measurements of the gasphase precursors to PM. Then in Section 2 we discuss highly time-resolved measures of the chemical properties of PM, continuing in Section 3 with the use of multiple chemical and meteorological measurements to characterize emissions from ambient sources. Finally, in Section 4 we discuss physical measures of PM. GAS-PHASE CHEMICAL MEASUREMENTS

Instruments that perform highly time-resolved measurements are common for the "traditional" gas-phase constituents such as ozone, sulfur dioxide (SO^sub 2^), nitrogen oxides (NO^sub x^), and carbon monoxide (CO). These measurements have helped shape our understanding of now commonly accepted diurnal patterns for these compounds. For instance, in urban areas, CO concentrations peak during rush hours because of vehicular emissions sources and these concentrations are usually higher during the morning commute than in the afternoon because of typical low, morning inversion heights. Similarly, ozone concentrations below the inversion layer are usually low at night because of low inversion heights combined with titration of ozone to molecular oxygen by nitric oxide (NO) emission from vehicles-the NO is not available to the ozone aloft so it remains as ozone overnight. In the morning as the sun warms the surface, the inversion layer rises, mixing ozone aloft down to the surface and producing the oft-observed morning ozone peak.14

But the atmosphere contains a rich array of gas-phase species that directly influence health or indirectly form gaseous or particulate compounds that do. The temporal patterns of their concentrations, available with highly time-resolved instruments, can elucidate atmospheric transport and transformation processes that are key to their dynamics. Boring et al.15 simultaneously measured nitric acid (HNO^sub 3^), nitrous acid (HONO), and SO^sub 2^ at the Atlanta and Houston Supersites observing HONO peaks at night 180[degrees] out of phase with HNO^sub 3^ peaks in the daytime. Genfa et al.16 further analyzed these data, observing many weekdays with double HONO peaks, but less so on weekends; the morning peak is thought to be due to direct emission from vehicles during rush hour whereas the evening peak is thought to arise from the afternoon rush plus atmospheric reaction pathways. This secondary pathway is important, because it allows HONO to act as a reservoir for NO^sub x^ at night as well; HONO has been shown to trigger hydroxide (OH) radical reactions just after sunrise in Los Angeles, CA.14 HONO formation and destruction appear to be close to photochemical equilibrium-rapid decreases in ultraviolet (UV) actinic flux, e.g., because of cloud passage, result in concomitant rapid increases in HONO concentrations. Both characteristics, the double peak pattern and apparent photochemical equilibrium, are readily observed with 20- min averaged measurements (Figure 1).

Li et al.17 measured hydrogen peroxide and methyl hydroperoxide (MHP) in Philadelphia observing characteristic diurnal cycles for each but during one week the MHP was substantially higher than others, suggesting a unique but unidentified emission of precursor during this time. Collocated gas phase instruments can be used together to infer emissions. For instance, as part of the PMTACS-NY Supersite, measurements of gas-phase ammonia (NH^sub 3^) were performed using a tunable diode laser absorption spectrometer (TDLAS), whereas a LI-7000 was used to measure carbon dioxide (CO2). Both measurements were very fast. Correlations between CO2 and NH^sub 3^ were frequently observed, indicating vehicular emissions and enabling the investigators to deduce NH^sub 3^ emission per vehicle mile traveled.18

Formaldehyde (HCHO) has been measured at several EPA Supersites and related studies, as summarized by Dasgupta and coworkers.19 One strength of highly time-resolved measurements is the ability to correlate concentration with wind direction, thereby identifying potential emissions sources. For instance, Dasgupta summarizes HCHO measurements in Houston at the HRM3 site (near the petrochemical facilities) that clearly point to large sudden releases of HCHO or its precursor in the vicinity of the ship channel. Plume passages were also observed-because of sea breezes, the wind often rotates around Houston, bringing different plumes to fixed sites and indicating emissions sources. Correlating wind direction with highly time resolved particle measurements is a powerful technique discussed later in this paper.

Many of the same physical and chemical processes that lead to gas- phase pollutants also lead to particulatephase ones, such as changes in wind direction, emissions of gaseous and particulate pollutants from the same source, and fumigation down of pollutants stored aloft from the previous day. Temporal patterns can only be elucidated with highly time-resolved instrumentation. Some gaseous pollutants also indicate the overall photochemical activity of the atmosphere, its oxidizing capacity, and their diurnal patterns, again contributing to our overall understanding of the mechanisms that lead to PM. In this section, we reviewed some of the dynamics in these gas-phase precursors, whereas in the next one we discuss the corresponding highly time-resolved measurements of PM chemical composition.

PM^sub 2.5^ CHEMICAL COMPOSITION MEASUREMENTS

It has been known for some time that PM^sub 2.5^ concentrations can rise and fall substantially over the time period of a few hours. For example, Chuersuwan et al.20 evaluated 24- and 1-hr time- averaged PM^sub 2.5^ data from several locations in New Jersey. The 1-hr data revealed short excursions to high mass concentrations, even on days when the 24-hr averaged concentration was low. During the summer months, 1-hr PM^sub 2.5^ and ozone measurements were highly correlated and there was little variability among measurement sites, suggesting that regional transport and secondary formation processes contributed substantially to the PM^sub 2.5^ levels. In winter, mass concentrations varied considerably with time and location, consistent with a lower mixing height and a concomitantly greater influence of local sources.

With the availability of highly time-resolved chemical composition measurements, insight into the chemical processes that drive rapid changes in PM^sub 2.5^ levels can be gained and the associated health effects can potentially be better understood. For example, Weber21 showed that some excursions to high PM^sub 2.5^ concentrations in Atlanta during the summer were driven by sulfate, others by carbon (organic and elemental), and still others by both. Similar measurements by Ondov in Baltimore22 confirmed that short- term excursions to high PM^sub 2.5^ were driven by sulfate and/or carbon. The chemical complexity that underlies rapid variations in PM^sub 2.5^ is summarized below for several important chemical species. Short-term spikes in sulfate and/or carbon do not always correlate with high PM^sub 2.5^ excursions, and spikes in the concentrations of other components (nitrate, water, metals) are also routinely observed. Examples of highly time-resolved chemical composition data are shown in Figures 2-5.

Sulfate

A large fraction of the fine particulate sulfate is produced by the oxidation of SO^sub 2(g)^ that is released during fossil fuel combustion. Sulfate concentrations show little diurnal variation except for an increase during the summer afternoon hours that is observed primarily during episodes of high sulfate concentration.23 More often, sulfate concentrations build more slowly on a time scale of several hours to several days, indicating a regional origin.22- 26 These episodes are observed in the summer and are generally characterized by an extended period of high daytime temperature, low wind velocity and mixing height. Figures 2 and 3 (episode A) show examples from Atlanta and Baltimore, respectively. These figures show multiday trends of increasing sulfate (Figure 2, August 26, 1999 to August 30, 1999), spikes due to plume passage (e.g., Figure 2; August 17, 1999), and unlike the general trends in most measurements, exhibit diurnal patterns (e.g., Figure 2; August 3, 1999 to August 7, 1999). High sulfate excursions on a regional scale usually have little dependence on short-term variations of wind direction. Short-term spikes in sulfate concentration, also illustrated in Figures 2 and 3, occur both during and between longer term sulfate elevations as plumes from individual sources pass by the measurement site21,22,25 and are often strongly correlated with wind direction.

Highly time-resolved measurements of sulfate and ammonium (plus nitrate and chloride if their concentrations are significant) allow for an estimation of particle acidity on short time scales. The molar concentration ratio of ammonium and sulfate concentration ratios are generally in the range of 1-2, suggesting a mixture of NH^sub 4^HSO^sub 4^ and (NH^sub 4^)^sub 2^SO^sub 4^, i.e., partial or complete neutralization in the urban environment.24,26,27 The conclusion that sulfate particles are generally neutralized is consistent with single particle chemical composition measurements28 that show many particles containing both sulfate and nitrate-a situation that is very unlikely if the particles are acidic. However, Weber21,24 has reported that ammonium-to-sulfate mole ratios less than 1 may be observed during high sulfate concentration episodes in Atlanta, where presumably the local H^sub 2^SO^sub 4^ concentration becomes large enough that neutralization is no longer possible. Zhang et al.26 observed similar episodes in Pittsburgh, where the ammonium-to-sulfate mole ratio was less than 1.5. During these time periods, the PM concentration was approximately 30% higher than average, almost exclusively due to an increase in sulfate mass. In addition, the nitrate and chloride concentrations were about a factor of 2 lower, consistent with the displacement of HNO^sub 3^ and hydrochloric acid (HCl) by sulfuric acid (H^sub 2^SO^sub 4^).

Sulfate appears to be the primary precursor to homogeneous nucleation in urban areas, as identified by highly time-resolved measurements during the Supersite studies. In Pittsburgh, as in other urban locations such as Mexico City and Atlanta,29-32 nucleation was observed when the condensational sink was small-that is, after an event that cleans the air of particles so that supersaturations lead to nucleation instead of condensation. Early in the nucleation process, sulfate dominates the particle composition.33,34 As the particles continue to grow, secondary organic components increase in concentration. 33 The section titled Particle Number Concentration Measurements below contains further discussion of nucleation observations and their interpretation. Carbon

Short-term variations in primary and secondary organic carbon (OC) (POA, SOA) have been examined in a couple of ways. One approach is to compare the concentration ratio of OC to elemental carbon (EC) [OC]/[EC] to the ratio that would be obtained if all of the OC was POA.35 When the SOA concentration becomes significant, the [OC]/ [EC] ratio increases and the SOA concentration is determined from the magnitude of the increase. A more detailed discussion of [OC]/ [EC] measurements is given within this special issue by Fine et al.36

In urban areas dominated by motor vehicle emissions, both EC and POA concentrations exhibit a similar diurnal variation, reaching a maximum in the early morning rush hour when vehicular emissions are high and the mixing height is generally low.37-39 During the day, the EC and POA concentrations decrease as the mixing height increases and the aerosol becomes more widely dispersed. In contrast, SOA steadily increases during the daytime as photochemical activity increases. The morning buildup and subsequent decrease of POA and the afternoon buildup of SOA are illustrated in Figure 4 for two time periods in Baltimore.37 SOA concentrations can remain high at night, apparently because of build up during the day and/or transport of air mass from upwind sources. Time-resolved measurements of particle-phase EC and gas-phase CO are highly correlated, supporting the use of CO as a surrogate for EC,37 whereas timeresolved SOA and ozone concentrations do not exhibit a significant correlation. CO and EC are primary, nonreactive on urban time scales, and emitted by incomplete combustion. Furthermore, time- resolved measurements in Pittsburgh show that SOA concentrations do not correlate strongly with particle acidity.33,40 In contrast, an apparent correlation between SOA concentration and particle acidity is observed when 24-hr averaged data from Atlanta and Fresno are evaluated.41 The different conclusions reached in these two studies illustrate how time resolution is key to the interpretation of measurements.

An alternative way to assess OC sources is with an aerosol mass spectrometer (AMS). In this approach, the mass spectra are partitioned into hydrocarbon-like organic aerosol (HOA) and oxygenated organic aerosol (OOA).42,43 The HOA is similar (though not equivalent) to POA, whereas OOA is similar to SOA. Like EC and POA, HOA exhibits a prominent diurnal profile that reaches a maximum in the morning rush hour then decreases for the remainder of the day.43 HOA peaks during rush hour traffic are not a coincidence. Car chase studies have demonstrated that HOA is predominantly emitted from motor vehicles; a result that would not have been observable without highly timeresolved instruments such as the AMS.44

Time-resolved measurements of HOA, EC, and gasphase CO are strongly correlated. OOA exhibits a weak diurnal pattern similar to sulfate and SOA in which the concentration increases during the day, remains high during the night and during episodes of high temperature, and stagnation builds up over a several day period. In highly polluted areas (e.g., Mexico City), OOA exhibits a stronger pattern of build up during the course of a day.29

In Pittsburgh, OOA and sulfate concentrations were observed to be relatively steady compared with HOA; secondary compounds are formed regionally and therefore should not have the same temporal variations as primary compounds emitted locally.43 This difference in the temporal nature of the primary and secondary compounds is also evident in the size distributions-ultrafine HOA decreased after the early morning during one event examined in detail, whereas the OOA increased, presumably due to condensation of these secondary compounds. But there was also evidence during this event of down mixing of material formed during the previous day and preserved in the residual layer overnight. Measurements at the New York Supersite also found diurnal patterns. A mass spectral signature analogous to HOA shows a diurnal variation associated with rush hour traffic, whereas another mass spectral signature analogous to OOA is derived more from regional influences.23

Nitrate

Ammonium nitrate is a semi-volatile material that cycles on and off pre-existing particles. The equilibrium state is strongly dependent on temperature and, if the particle is an aqueous solution, indirectly dependent on relative humidity (RH).45 This cycling not only occurs in the atmosphere but on the surface of filters and impactors that integrate their concentration of hours or more. Thus, highly time-resolved measurements are thought to have fewer evaporation/condensation artifacts when analyzing such volatile compounds.

Measurements in several urban locations in the Eastern United States (Atlanta, Pittsburgh, Baltimore, New York) with various instruments all show a strong diurnal dependence.22-24,46-49 The particulate nitrate concentration generally increases during the nighttime, reaches a peak in the early morning, then falls back to a low level during the afternoon and early evening. This behavior is illustrated in Figure 5 for time-resolved nitrate measurements in Atlanta. The morning increase is expected because increasing NO^sub x^ levels during the rush hour period provide a source of atmospheric nitrate that migrates to the particle phase, whereas the temperature and mixing height are still relatively low. However, particulate nitrate and gas-phase NO^sub x^ concentrations do not exhibit a high correlation, for it is the temperature rather than the total nitrate loading in the atmosphere that is the main driver for ammonium nitrate condensation. Although NO^sub x^ and gas phase HNO^sub 3^ concentrations increase during the day, particulate nitrate concentrations do not.

In the western United States, highly time-resolved measurements show that the patterns differ from those in the East. Under stable atmospheric conditions common in the San Joaquin Valley of California, nitric pentoxide forms aloft and is then converted to HNO^sub 3^. As the mixed layer rises during the daytime, EC and other tracers decrease as fresh air aloft is mixed down, but surface nitrate levels may rise as the elevated HNO^sub 3^ concentrations aloft are fumigated down. The changes are quite abrupt in that particulate nitrate increases of 20 [mu]g/m^sup 3^ in 30 min are common and would be missed without the use of highly time-resolved measurements.87,88

Time-resolved measurements of particulate nitrate provide a relatively stringent test of thermodynamic modeling of the partitioning of inorganic species between the gas and particle phases.50,51 For example, simulations of time-resolved measurements in Pittsburgh show that the physical state of the aerosol does not significantly influence model error. However, particle-mixing state does appear to be important-an assumption of external mixing during low RH periods appears to improve the agreement between the two. Robust models of nitrate partitioning are needed to evaluate PM^sub 2.5^ control strategies, for example the increase in particulate nitrate that would be expected to accompany a decrease in sulfate.52

Ammonium nitrate partitioning to the particle phase during periods of low temperature and high RH also influences the particle number concentration. Measurements by Tolocka et al.48 using a single PM spectrometer show that in Baltimore the nitrate condensation is often accompanied by a rapid increase of particles in the 60-100-nm size range. These particles are composed primarily of ammonium nitrate and are presumably formed by rapid growth of much smaller particles whose initial composition is now masked by the large amount of ammonium nitrate that has been added. At the same time, virtually all particles at other sizes gain at least some ammonium nitrate by condensation.

Particulate Water

Time-resolved measurements of particulate water in comparison with predictions from thermodynamic modeling allow the contribution of organics to water absorption to be assessed.27 For summer data, the model consistently underpredicts the water content during the afternoon hours. The excess water correlates strongly with shortterm variations of sulfate but not OC, suggesting that the additional water is not due to absorption by organic compounds. Instead, the results are consistent with the hypothesis that water absorption by sulfate is influenced by the presence of organic compounds.

PM SOURCE CHARACTERIZATION

In the previous section, we discussed many of the highly time- resolved measurements that occurred during the Supersites program and what we have learned to date from these measurements; however, we confined our attention to lessons learned when the data from these instruments were used alone. In the next section, we focus our attention on the use of multiple measurements, occurring simultaneously, to deduce point source characteristics; deductions that would not be possible without multiple highly time-resolved instruments working in concert.

Source Identification

The chemical composition of an ambient particle is determined by its source and atmospheric processing. The latter may lead to condensation of secondary components including ammonium sulfate, ammonium nitrate, and secondary OC. Internally mixed particles, those whose chemical composition is dominated by secondary components, account for approximately 50% of the PM and 40% of the particle number in Baltimore aerosol, averaged over an 8-month period from April through November 2002.53 Similar loadings have been found in Pittsburgh, Houston, and Atlanta,54-56 although the loading varies seasonally and diurnally. In Atlanta, where single particle measurements were made during an intensive study in August, the particles were dominated by organics, which, as a result, showed no appreciable diurnal variation, whereas the nitrate loading was greatest at night and decreased throughout the day as temperature increased and humidity decreased. 56 Time-resolved measurement of nitrate during the same campaign also showed substantial diurnal variations in particle nitrate loadings.25 The use of multiple highly time-resolved instruments together to draw more detailed conclusions about sources and atmospheric processes will be discussed more below. Brook and coworkers57 used an AMS and Light Detection and Ranging (LIDAR) system, along with other measurements, to investigate PM in the Lower Fraser River Valley in British Columbia, Canada. The temporal variations in PM composition observed by the AMS were generally due to either (1) changes in wind direction that brought plumes to the measurement sites or (2) changes in mixing height that promoted down mixing of pollutants from aloft. The LIDAR can provide an invaluable addition to the highly time-resolved physical and chemical measurements and standard meteorological measurements common in most studies.

Metals and other trace elements are minor but important components of particulate matter. Typically fewer than 10% of particles in urban air contain these species.53-56 Although many of these species are toxic (e.g., As, Cd, Cr, Pb, Se), their atmospheric concentrations are often too low to constitute a significant toxic health threat, although metals with multiple oxidation states may elicit health effects by production of oxidative stress through Fenton reactions. More importantly, these elements serve as tracers for specific combustion sources and can be used for source attribution by receptor modeling.58 There is no intrinsic diurnal pattern associated with these species other than that due to the diurnal mixing height variation. A stronger temporal influence is the operating schedule of the source, and the wind speed and direction bringing plumes to the measurement site. Until the advent of highly timeresolved measurements demonstrated during the Supersite program, sampling times of hours or even days were needed to obtain sufficient signal-to-noise ratio, so temporal resolution was poor. Highly time-resolved metal measurements provide a wealth of information by correlating trace element composition changes with shortterm changes in source variations, wind direction, and other meteorological variables to identify and track plumes from specific point sources. This is illustrated in Figure 6, in which a trace metal signature of a coke facility is obtained from fence line measurements when the fence line location is downwind of the facility.59

The power of time-resolved elemental analysis has been demonstrated by Ondov and colleagues59,60 who compared 2.5-hr and 30- min averages of PM^sub 2.5^ samples collected outside of Washington, DC, with the University of Maryland Semi-Continuous Elements in Aerosol Sampler (SEAS). Principal component analysis was applied to the 30-min averages to identify several sources, including urban dust (Cr, Cu, Fe, Mn, Pb), an incinerator (Cd, Zn), a coal-fired power plant (Pb, Se), a tour bus (Al, Fe), an unknown arsenic source (As, Cr), and an oil-fired power plant (Ni). In contrast, when 2.5- hr averages were used the oil-fired power plant could not be resolved. A second generation SEAS was used in the Bay Regional Atmospheric Chemistry Experiment (BRACE)60 and at the Pittsburgh, St. Louis, and Baltimore Supersites. In BRACE, previously unreported As sources were detected as was the plume from an animal feed supplements plant, which had not been detected previously with standard 24-hr sampling methods.

In Tampa, Pittsburgh, Baltimore, and St. Louis, plumes of individual coal- and oil-fired power plants were readily detected by observing short-lived increases in ambient Se and Ni concentrations, respectively. The duration of these excursions contains a measure of the standard deviation of the plume widths, a parameter that depends on source distance as well as meteorological conditions. Park and coworkers61 were able to exploit this information using an approach suggested by Yamartino;62 Gaussian plume equations were used to back calculate emission rates of SO^sub 2^ and metals from various point sources in Tampa and Pittsburgh. Park and colleagues were able to discern sources separated by as little as 3 [degrees] of arc. Their Pseudo-Deterministic Multivariate Receptor Model (PDMR) was shown to accurately identify SO^sub 2^ emissions. Thus, meteorological dispersion factors predicted by the model were also accurate, providing a relatively convenient method for validating deterministic plume models that are needed to predict human exposures to toxic substances.

Additional information can be obtained from realtime single PM spectrometry where the size, chemical composition and time (and consequently temporally resolved meteorological parameters such as wind speed and direction) can be recorded on a particle-by-particle basis. For example, Tolocka et al.63 were able to resolve two sources from the same wind direction on the basis of chemical composition: one particle type contained K and Pb, whereas the other contained K and As. Essentially no particles were observed containing both As and Pb, confirming the existence of two separate sources. In another example, Bein et al.54 were able to distinguish several sources that emitted particles containing iron (see Figure 7). Several coal combustion facilities were found to emit particles containing Si, K, Fe, Ga, and sometimes smaller amounts of Al, Na, Li, and/or Pb. The particle size range was narrowly centered around 200 nm, consistent with a high-temperature combustion facility.64 Highly time-resolved data permitted emissions from three such sources, two large coal-fired power plants located far from the site and one coal-fired boiler located close to the site, to be distinguished even though their directions relative to the measurement site differed by less than 10[degrees].65 Another source type emitted particles containing almost exclusively Fe, and the wind direction (almost 180 [degrees] from the coal combustion facilities) suggested a steel processing facility. Yet another source type emitted particles containing Fe and Ce, the wind direction was substantially different from the previously discussed sources, and whereas the other particle classes had mean particles sizes near 200 nm, indicating high-temperature combustion sources, for this class the size distribution was shifted to large size bins centered close to 600 nm, which is near the upper size limit of this instrument suggesting that the tail of a coarse mode was being detected.

Likewise, Tolocka et al.53 using a single particle mass spectrometer observed the impact at the Baltimore site of emissions from a nearby metropolitan bus depot. Vehicle activity at this site was the greatest in the early morning hours. When the wind was from the depot, short-term peaks in the concentration of carbonaceous particles were detected at the site, whereas when the wind was from other directions, no peak was observed. Particles containing transition metals and/or trace elements are almost always detected in short-term peaks, which can last as short as several minutes53- 56,66 facilitating correlation with wind direction. Invariably in these single particle measurements, there are strong correlations between particle composition and wind direction so that wind roses of many particle composition classes show strong directional dependencies-that is, they indicate the direction of point sources.

Source Strength Estimation

Correlating highly time-resolved measurements to local wind speed and direction can also be used to estimate emissions from the source. Two approaches have been used to estimate source strengths that do not employ highly time-resolved instrumentation. The first approach involves direct source measurements, which are usually limited to special studies designed to characterize the source or source type. The exception is continuous monitoring systems in use on large emissions stacks by major industry, but these provide concentrations of gas-phase species and typically not primary PM emissions. Second, receptor-based modeling with 2- to 24-hr time- averaged samples can be used to estimate source strength. However, this poor time resolution limits the use of ambient meteorological data in the estimation, which degrades accuracy and makes it difficult to separate the contributions of some sources.67

Park et al.61,68 have applied the PDMR described in Section 3a to exploit highly time-resolved measurements for strength estimation. This model explicitly incorporates knowledge of the number and locations of major stationary sources, source and transport wind directions, stack emission parameters, and meteorological plume dispersion parameters. The model is able to determine average emission rates and time-resolved ambient concentrations for each measured species. When applied to time resolved SO^sub 2^ and metals data in Tampa,68 agreement between predicted and measured ambient SO^sub 2^ levels was high (ratio of measured/expected concentration was 1.00 +- 0.18 with R^sup 2^ = 0.97) and the predicted SO^sub 2^ emission rates for each of four utility sources were within 6% of the measured values. For elemental markers of coal (As, Se), the agreement between predicted and measured ambient levels was similarly high. The PDMR approach was subsequently applied to SO^sub 2^ and metal concentration measurements in Pittsburgh,61 a more stringent test because of complex local topography. When applied to a 12.5-hr period when winds blew from a narrow range of directions that included four coal boilers, agreement between predicted and measured SO^sub 2^ and metal (As, Cr, Cu, Ni, Pb, Se, Zn) levels was comparable to the Tampa study.60 In principle, Gaussian plume modeling can be used in an inverse mode to predict emissions from any source, as long as the ambient measurement data are sufficiently time resolved compared with changes in the wind speed and direction and mixing height. That is, if the distance to the source is known along with estimates of atmospheric turbulence characteristics and local pollutant concentrations, the remaining unknown is the source strength. This technique has been employed to estimate the emission rates from several point sources during the Pittsburgh Supersite study.69 Receptor Relationships

As exhaust from high temperature combustion sources is emitted into the atmosphere, the emissions dilute, which (1) reduces the partial pressure of the emitted gases and particles and (2) cools the emissions, reducing the vapor pressure of the emitted gases. Elevated stacks are difficult to probe but vehicle emissions from freeways are at ground level, and in certain locations the wind blows perpendicular to the freeway at a relatively constant velocity. This situation prevails in Los Angeles and was exploited by Zhu et al.70,71 to deduce the temporal dynamics of the aerosol evolution by measuring the particle size distributions at different distances from the freeway. These data were analyzed by Zhang et al. and Jacobson and Seinfeld to more fully ascertain the underlying dynamics. 72-75 Jacobson and Seinfeld showed that coagulation cannot explain the observed dynamics. Zhang et al. showed that condensation and evaporation of vapors, probably organic ones, explains the dynamics-no evidence for nucleation was observed. Initially, when the exhaust mixes with the atmosphere, rapid cooling dramatically decreases the vapor pressure of the gases causing them to condense on the available particles, leading to rapid growth in the ultrafine particles. However, after continued mixing the temperature no longer decreases substantially but the partial pressure of the organics continues to decrease and eventually drops below that of their vapor pressures in the particle phase. Thus, evaporation commences, causing ultrafine particles to then shrink. Although this story gives the big picture, different particle sizes have different dynamics, growing and shrinking at different times partially because of their initial composition differences and partially because of differences in their size. Zhang and co-workers72,73 estimate that approximately 50,000 particles/cm^sup 3^ grow into the observable size range (greater than ~10 nm) then shrink back again, so that only people within approximately 100 m of the freeway are exposed to them. In a subsequent work, Zhang and coworkers74 show that the number-based emissions factors for light- and heavy-duty vehicles depend on the receptor; receptors close to the roadway experience much different particle size distributions and compositions than those far afield, so that the number-based emissions factors really depend on the receptor that is being considered.

PARTICLE NUMBER CONCENTRATION MEASUREMENTS

Highly time-resolved chemical measurements of particulate matter are all relative newcomers to the field of atmospheric science, whereas highly time-resolved measurements of the physical properties have been available for decades. In the previous sections we used some of these physical measurements along with chemical measurements to constrain the conclusions and strengthen the deductions reached by the chemical measurements. In this section, we discuss some more recent developments in physical measurements, some of which, such as shape and density, are especially important for converting the particle number distribution to other moments such as mass or area.

Nucleation and Ultrafine Number Concentrations

Ultrafine particles may have substantial health effects both because they penetrate deep into the lungs and because they are more readily transported from the lungs to the blood stream, where they may be distributed to other organ systems.76 Ultrafine particles are almost always formed by a homogeneous nucleation process, whereby supersaturated vapors spontaneously form a new solid or liquid particle and subsequent condensation and/or coagulation causes the nuclei to grow to sizes at which they are observable (i.e., greater than ~10 nm for a standard condensation particle counter (CPC); greater than approximately 3 nmfor a nanoCPC). The nucleation process may occur in a combustion flame where soot is often the nucleus formed, in the plume from a combustion source where rapid cooling dramatically decreases the vapor pressure, or in the atmosphere where gas-phase chemical reactions form low volatility precursors such as sulfuric acid.

A number of Supersite-related studies employed scanning mobility particle sizers (SMPSs) using both long columns for particles greater than approximately 10 nm and short columns for particles greater than approximately 3 nm. Woo et al.77 measured size distributions of particles in Atlanta for 13 months. On average, the number of particles greater than 10 nm peaked in the early morning, presumably as a result of rush hour traffic emissions and low inversion heights limiting dilution. Consistent diurnal patterns were observed over a span of weeks, indicating nucleation associated with rapid cooling of exhaust as it mixes with surrounding air.78 In contrast, the concentration of particles smaller than 10 nm peaked during midday, indicative of photochemical production of a nucleation precursor, such as sulfuric acid vapor. They note that these midday nucleation observations have been observed elsewhere, for instance by Birmili.79 Figure 8 illustrates the diurnal patterns observed in Atlanta during the measurement period for several ultrafine particle size ranges.

Woo et al.77 also observed bursts of particles in the 3- to 10- nm size range in Atlanta, most likely due to homogeneous nucleation events. During these events, the particle concentrations increased by a factor of up to approximately 50, and often the concentration distribution increased in the 3- to 10-nm range for the smaller particles, indicating that the concentrations of particles were even larger for particles smaller than the minimum detection limit of 3 nm.31 Note that these events can be very short, even as short as one or two scans of the SMPS, although sometimes they last for a few hours and are scattered throughout the year with larger occurrences reported in April and August. These particles do not coagulate in the atmosphere but instead grow by condensation to sizes around 100- 300 nm-sizes that penetrate deep into the lungs.

Nucleation of sulfuric acid vapor is related to the concentration of SO^sub 2^ precursor and of water, because the sulfuric acid nucleation process is certainly bimolecular (H^sub 2^SO^sub 4^ and water) but may be tertiary (H^sub 2^SO^sub 4^, water, and NH^sub 3^).80 In this way, Fresno provides an instructive contrast to Atlanta. In both locations, the NH^sub 3^ concentration is sufficiently high that it does not limit nucleation but in Fresno the SO^sub 2^ concentrations are much lower than in Atlanta because of generally more stringent fuel sulfur regulations in California and Fresno is usually drier than Atlanta. As in Atlanta, particles were observed in the 3- to 10-nm size range in Fresno. In contrast, the Fresno size distributions usually showed (1) increasing concentration with particle size in this 3- to 10-nm range, and (2) concentrations typically an order of magnitude lower during the nucleation events.81 These diurnal patterns are illustrated in Figure 9. The largest competition for nucleation is condensation, because the compounds that can nucleate may also condense, so a larger condensational sink means that the nucleating gases may be more likely scavenged by pre-existing particles. In Fresno, there is a clear correlation between a small condensational sink and high probability of observing a nucleation event, as has been found in other locations around the world and in Pittsburgh,80 counter to what was observed in Atlanta where the correlation was weak. A morning peak in nanoparticle concentrations was observed in both Fresno and Atlanta and the peak was smaller during the weekend, presumably because of lower traffic levels. The weekend evening peak in the winter was not lower in Fresno, presumably due to wood smoke, known to be a substantial fraction of PM there. It is posited that the Fresno nucleation events occur aloft and are observed at the surface after subsidence or down mixing, whereas those occurring in Atlanta occur in situ.

Nucleation was also observed at the Pittsburgh Supersite and more frequently than Atlanta or Fresno.72-74,82 Occurrence of the nucleation events was correlated to the product of UV radiation intensity (peaking midday) and SO^sub 2^ concentration, indicating a photochemical pathway for H^sub 2^SO^sub 4^ gas production and subsequent nucleation, similar to that observed in Atlanta and Fresno. Stanier et al.82 also report the temporal development of the size distribution, showing nucleation events that generally start mid-morning. The nucleated particles are first rapidly neutralized by NH^sub 3^, and then grow by condensation of organics, NO^sub 3^, and additional sulfate over the next few hours until their growth slows substantially near the 30- to 100-nm size range. The temporal resolution of this growth profile illustrates two important points. First is the dynamics of the aerosol-the nucleation is followed by rapid growth that slows as the particles age, in agreement with many other observations and theoretical predictions. 83 Second, the nucleation events are usually regional-the clean and repeatable growth of the particles while the wind brings new air masses to the site indicates that the ultrafine size distribution is spatially uniform at least over the few hours of the nucleation dynamics. During a couple of weeks, an Aerodyne AMS was deployed in Pittsburgh and several nucleation events were observed.33 Whereas SMPS data provide highly time-resolved size distributions by number (that can be inverted to mass), AMS data provide highly time-resolved mass distributions of the chemical components of the particles in the measured size range (>/=30 nm), in this case reported for sulfate, NO^sub 3^, ammonium, and organics. For one nucleation event examined in detail (Figure 8), the volume distribution over the day inverted from the SMPS data was very similar to the mass distribution deduced from these four components from the AMS data. The power of the AMS is the ability to follow the size distribution of these components as the particles grow. As shown in Figure 10, the early growth was due generally to condensation of sulfate (also supporting that sulfate is the compound that is nucleating) and its associated ammonium, although during the initial phases the sulfate did not appear to be completely neutralized by ammonia. This may be due to the lack of NH^sub 3^ early on, that then became available because of increased vehicular emissions, or it may be due to the Kelvin effect (the effect of a curved surface on vapor pressure over that surface that limits condensation onto the smallest particles of volatile compounds). As the particles grew, the growth was increasingly due to ammonium nitrate and organics; as the particles grow their Kelvin effect decreases allowing higher volatility components to condense.

The time-resolved chemical composition available with the AMS also enabled Zhang and co-workers33,43 to separate the portions of the size distribution during the nucleation that were due to fresh vehicular emissions from those portions due to the nucleation event because the vehicle-derived particles were dominated by organics- this was clear from the composition size distribution before the nucleation event began-whereas the fresh nuclei were dominated by sulfates. The chemical resolution also enabled these researchers to compare the chemical factors underlying the three events that they observed, showing how the differences in the number of particles produced during the event and the growth of these particles are a result of the chemical components analyzed.

Smith et al.34 have directly measured the composition of particles smaller than 20 nm during nucleation events in Atlanta. They found that all of the aerosol mass consisted of ammonium and sulfate, which is consistent with the AMS measurements in Pittsburgh suggesting that nucleation and early growth are due to sulfate.

Density and Shape Factors

Aerosol instruments measure size on the basis of one or more surrogates: mobility instruments, such as the SMPS, report mobility diameter; continuum aerodynamic sizing instruments, such as the aerodynamic particle sizer (APS), report the aerodynamic diameter; most mass spectrometers, such as the AMS, rapid single particle mass spectrometer (RSMS), or aerosol time-of-flight mass spectrometer (ATOFMS), report the vacuum aerodynamic diameter. These various measures of size, as well as the number and mass distributions, are all related to each other via the particle density and shape factor.84 Fresh emissions from combustion sources are highly carbonaceous soot, often in the form of fuzzy agglomerates of coagulated nanoparticles, whereas aged particles are often spherical if they have a substantial liquid water or liquid organic fraction or faceted if they are dried crystals. These shapes also affect observed density because of their fractal dimension but density is also a function of the chemical composition, with organics having densities substantially less than 1 g/cm^sup 3^ and metals having densities of 10 g/cm^sup 3^ or higher. Current thinking is that fuzzy soot particles have a low fractal dimension (that is, their dimension is characteristic more of a line, dimension 1, or a plate, dimension 2, than a compact particle, dimension 3) on emission but that this dimension increases during atmospheric processing. As various liquids condense on and evaporate from the particle during its lifetime, the surface tension tends to drag the chain agglomerates together into a more compact shape and the liquids also fill the interstices.

Highly time resolved measures of shape and density often involve two complementary measures of size that when combined provide an estimate of these parameters. McMurry and co-workers85 measured the mobility diameter with a tandem differential mobility analyzer (DMA), which classifies size independent of density, and an aerosol particle mass analyzer (APM),86 now available commercially as the Kanomax APM-10, which balances centrifugal and electrostatic forces on a particle to produce a stream of particles with a given mass. The density of particles was measured in the 100- to 300-nm size range by combining the DMA and APM measurements. These measurements showed that particles of a given mobility often have several distinct masses, which indicates particles of different densities. Some of the densities were substantially less than unity, suggesting that density is not really the proper measure because the particles were chain agglomerates such as soot with fractal dimension substantially less than 3.

Density variations attributed to chain agglomerates were also observed to have rapid time variation during measurements in Pittsburgh-measurements were performed with a SMPS and AMS. Zhang et al.26 compared the PM1 (PM < 1 [mu]m) mass diurnal variation estimated with the SMPS and AMS. During some periods, both PM1 masses varied in a similar fashion but at other times they diverged severely, demonstrating variations in the density during these periods. Similar results were observed in Mexico City29 using an AMS outfitted with a beam width probe to infer the particle shape.

CONCLUSIONS

Increasing evidence points to the association between health effects and short-term exposures to particulate matter concentration elevations. Air quality is strongly affected by anthropogenic emissions into the atmosphere and the subsequent chemical and physical processes they undergo. Because numerous sources may exist and atmospheric processes are complex, human exposure to particulate matter varies widely with time and location. Understanding this variability necessarily requires complementary and highly time- resolved chemical and physical measurements.

Before the Supersite initiative, the total mass and size distribution of particulate matter was known to change quite rapidly. Supersite and related measurements have contributed substantially to our understanding by providing highly time- resolved chemical composition measurements for precursor gases and fine particles, along with more complete measurements of the particle size distribution, particularly for ultrafine particles. Excursions to high particulate matter levels are usually the result of rapidly increasing nitrate, sulfate, and/or OC concentrations that are driven by favorable atmospheric conditions such as low mixing height and wind velocity, high photochemical activity, favorable temperature and RH conditions, and wind directions that efficiently transport point source emissions to the measurement site. Concentrations of trace components (e.g., metals) are also found to change rapidly. Temporal changes in particulate mass levels typically evolve over a time scale of 1 hr to 1 day, whereas the smallest particles (3-50 nm in diameter) may show substantial variations on time scales as short as 1 min.

By combining multiple highly time-resolved measurements with atmospheric models, sources have been identified and their source strengths estimated; such conclusions could not have been reached without the advent of this new family of instruments.

As highly time-resolved instruments transition from research tools to products manufactured and sold at reasonable prices, these instruments will appear at greater numbers of monitoring sites. The resulting dataset will facilitate epidemiological and other studies that associate health with particulate matter composition, concentration, and temporal behavior. Thus, by providing a more detailed picture of human exposure, highly time-resolved measurements further elucidate the link between exposure and health and will likely lead to promulgation of improved air quality standards, especially as these tools are increasingly employed by health scientists. By more accurately determining the identity and strength of particulate matter sources, highly time-resolved measurements will provide the opportunity to develop more effective control strategies to meet these standards. Applications to both of these areas will most certainly continue in the future.

ACKNOWLEDGMENTS

The authors thank Paul Solomon and John Ondov for helpful suggestions and critical reading of this manuscript. Although the research described in the article has been funded wholly or in part by EPA through grant RD-83241401-0 to the University of California- Davis, it has not been subject to the agency's required peer and policy review and therefore does not necessarily reflect the views of the agency and no official endorsement should be inferred.

IMPLICATIONS

Several new instruments were deployed during the EPA Supersites program that measure physical and chemical properties of PM^sub 10^ and PM^sub 2.5^ with temporal resolution better than 1 hr. These instruments offer a wide range of new insights into the dynamics of PM in the atmosphere.

REFERENCES

1. National Ambient Air Quality Standards for Particulate Matter Revisions to Ambient Air Monitoring Regulations Proposed Rule; CFR 40, Parts 50, 53, and 58, 1972. 2. Zeka, A.; Zanobetti, A.; Schwartz, J. Short Term Effects of Particulate Matter on Cause Specific Mortality: Effects of Lags and Modification by City Characteristics; Occup. Environ. Med. 2005, 62, 718-725.

3. Dockery, D.; Luttmann-Gibson, H.; Rich, D.; Link, M.; Mittleman, M.; Gold, D.; Koutrakis, P.; Schwartz, J.; Verrier, R. Association of Air Pollution with Increased Incidence of Ventricular Tachyarrhythmias Recorded by Implanted Cardioverter Defibrillators; Environ. Health Perspect. 2005, 113 1516-1521.

4. Urch, B.; Silverman, F.; Corey, P.; Brook, J.; Lukic, K.; Rajagopalan, S.; Brook, R. Acute Blood Pressure Responses in Healthy Adults during Controlled Air Pollution Exposures; Environ. Health Perspect. 2005, 113, 1052-1055.

5. Ruidavets, J.M.; Cournot, M.; Cassadou, S.; Giroux, M.; Meybeck, M. Ozone Air Pollution Is Associated with Acute Myocardial Infarction; Circulation 2005, 111, 563-569.

6. Stull, R.B. An Introduction to Boundary Layer Meteorology; Kluwer: Boston, 1991.

7. Seigneur, C.; Pun, B.; Pai, P.; Louis, J.-F.; Solomon, P.; Emery, C.E.; Morris, R.; Zahniser, M.; Worsnop, D.; Koutrakis, P.; White, W. Tombach, I. Guidance for the Performance Evaluation of Three-Dimensional Air Quality Modeling Systems for Particulate Matter and Visibility; J. Air & Waste Manage. Assoc. 1999, 50, 588- 599.

8. National Ambient Air Monitoring Strategy; Final Draft; U.S. Environmental Protection Agency: Research Triangle Park, NC, 2004; available at http://www.epa.gov/ttn/amtic/files/ambient/ monitorstrat/allstrat.pdf (accessed 2007).

9. Speciation Trends Network; available at http://www.epa.gov/ ttn/amtic/ speciepg.html (accessed 2007).

10. IMPROVE Interagency Monitoring of Protected Visual Environments available at http://vista.cira.colostate.edu/improve/ (accessed 2006).

11. EPA Research Highlights: PM Supersites Program Provides Data to Support PM^sub 2.5^ Reduction Strategies; EM 2006, January, 39.

12. Solomon, P.A.; Sioutas, C. Continuous and Semicontinuous Monitoring Techniques for Particulate Matter Mass and Chemical Components: a Synthesis of Findings from EPA's Particulate Matter Supersites Program and Related Studies; J. Air & Waste Manage. Assoc. 2008, 58, 164-195.

13. Watson, J.G.; Chen, L.-W.A.; Chow, J.C.; Doraiswamy, P.; Lowenthal, D.H. Source Apportionment: Findings from the U.S. Supersites Program; J. Air & Waste Manage. Assoc. 2008, 58, 265- 288.

14. Solomon, P.A.; Cowling, E.; Hidy, G.; Furiness, C. Comparison of Scientific Findings from Major Ozone Field Studies in North America and Europe; Atmos. Environ. 2000, 34, 1885-1920.

15. Boring, C.B.; Al-Horr, R.; Genfa, Z.; Dasgupta, P.K. Field Measurement of Acid Gases and Soluble Anions in Atmospheric Particulate Matter Using a Parallel Plate Wet Denuder and an Alternating Filter-Based Automated Analysis System; Anal. Chem. 2002, 74, 1256-1268.

16. Genfa, Z.; Slanina, S.; Boring, C.B.; Jongejan, P.A.C.; Dasgupta, P.K. Continuous Wet Denuder Measurements of Atmospheric Nitric and Nitrous Acids during the 1999 Atlanta Supersite; Atmos. Environ. 2003, 37, 1351-1364.

17. Li, J.; Dasgupta, P.K.; Tarver, G.A. Pulsed Excitation Source Multiplexed Fluorometry for the Simultaneous Measurement of Multiple Analytes: Continuous Measurement of Atmospheric Hydrogen Peroxide and Methyl Hydroperoxide; Anal. Chem. 2003, 75, 1203-1210.

18. Li, Y.; Schwab, J.J.; Demerjian, K.L. Measurements of Ambient Ammonia Using a Tunable Diode Laser Absorption Spectrometer: Characteristics of Ambient Ammonia Emissions in an Urban Area of New York City; J. Geophys. Res. Atmos. 2006, 111, (D10S02), doi: 10.1029/ 2005JD006275.

19. Dasgupta, P.K.; Li, J.; Zhang, G.; Luke, W.T.; McClenny, W.A.; Stutz, J.; Fried, A. Summertime Ambient Formaldehyde in Five U.S. Metropolitan Areas: Nashville, Atlanta, Houston, Philadelphia, and Tampa; Environ. Sci. Technol. 2005, 39, 4767-4783.

20. Chuersuwan, N.; Turpin, B.J.; Pietarinen, C. Evaluation of Time-Resolved PM^sub 2.5^ Data in Urban/Suburban Areas of New Jersey; J. Air & Waste Manage. Assoc. 2000, 50, 1780-1789.

21. Weber, R. Short-Term Temporal Variation in PM^sub 2.5^ Mass and Chemical Composition during the Atlanta Supersite Experiment 1999; J. Air & Waste Manage. Assoc. 2003, 53, 84-91.

22. Park, S.S.; Kleissl, J.; Harrison, D.; Nair, H.P.; Kumar, V.; Ondov, J. Investigation of PM^sub 2.5^ Episodes Using Semi- Continuous Instruments at the Baltimore Supersite at Ponca Street; Aerosol Sci. Technol. 2006, 40, 845-860.

23. Drewnick, F.; Schwab, J.J.; Jayne, J.T.; Canagaratna, M.; Worsnop, D.R.; Demerjian, K.L. Measurement of Ambient Aerosol Composition during the PMTACS-NY 2001 Using an Aerosol Mass Spectrometer. Part 1: Mass Concentrations; Aerosol Sci. Technol. 2004, 38 (Suppl. 1), 92-103.

24. Weber, R.J.; Orsini, D.; Duan, Y.; Lee, Y.-N.; Klotz, P.J.; Brechtel, F. A Particle-into-Liquid Collector for Rapid Measurement of Aerosol Bulk Chemical Composition; Aerosol Sci. Technol. 2001, 35, 718-727.

25. Weber, R.; Orsini, D.; Duan, Y.; Baumann, K.; Kiang, C.S.; Chameides, W.; Lee, Y.N.; Brechtel, F.; Klotz, P.; Jongejan, P.; ten Brink, H.; Slanina, J.; Dasgupta, P.; Hering, S.; Stolzenburg, M.; Edgerton, E.; Harstell, B.; Solomon, P.; Tanner, R. Intercomparison of Near Real-Time Monitors of PM^sub 2.5^ Nitrate and Sulfate at the U.S. Environmental Protection Agency Atlanta Supersite; J. Geophys. Res. 2003, 108, D78421, doi: 78410.71029/72001JD001220.

26. Zhang, Q.; Canagaratna, M.R.; Jayne, J.T.; Worsnop, D.R.; Jimenez, J.-L. Time- and Size-Resolved Chemical Composition of Submicron Particles in Pittsburgh: Implications for Aerosol Sources and Processes; J. Geophys. Res. 2005, 110, D07S09, doi: 10.1029/ 2004JD004649.

27. Khlystov, A.; Stanier, C.O.; Takahama, S.; Pandis, S.N. Water Content of Ambient Aerosol during the Pittsburgh Air Quality Study; J. Geophys. Res. 2005, 110, D07S10, doi: 10.1029/2004JD004651.

28. Lake, D.A.; Tolocka, M.P.; Johnston, M.V.; Wexler, A.S. The Character of

Source: Journal of the Air & Waste Management Association

More News in this Category