Hypotheses and Their Testing

Ambient Aerosol Characterization
Measurement Methods
Atmospheric Processes
Table 3. Some Tracer compounds for Biogenic Organic Aerosols
Source-Receptor Relationships
Aerosol Properties
Health Effects
Indoor Exposure

Ambient Aerosol Characterization

Hypothesis 1.1 The measured aerosol mass can be fully explained if one accounts for the water retained by organics and inorganics, the full organic aerosol contribution, and the full crustal contribution.

The available measurements will allow us to quantify all the known contributions to the aerosol mass. The TDMA measurements by the Pandis group will provide the residual water for the RH (30-40%) corresponding to the measurement of the aerosol mass. The FTIR measurements by the Turpin group together with the organic chemical speciation by the Rogge team will allow the estimation of the organic aerosol H, O, C, and N content and the accurate quantification of the total organic aerosol mass. The crustal component contribution will be quantified using the approach described by Andrews et al. (1999). The sum of these individual contributions will then be compared with the PM concentration.

Hypothesis 1.2 The ambient aerosol surface area can be calculated with an error less than 20% using aerodynamic size measurements and assuming that all particles are spherical.

The Epiphaniometer measurements will be used for the direct calculation of the surface area concentration of the particles as a function of time. These results will be compared with the surface area concentration estimated from the number distribution measurements of the two Differential Mobility Analyzers and the Aerodynamic Particle Sizer assuming spherical particles. The results of the two approaches will be compared to quantify the magnitude of the error introduced by assuming spherical particles. The samples will be analyzed by SEM during periods where the two measurements are in agreement or when large discrepancies appear to independently check the particle shape.

Measurement Methods

Hypothesis 2.1 Single Particle Mass Spectrometers can be used to obtain the full number and mass composition distributions of ambient aerosols.

The Wexler team is currently developing methods for the quantitative interpretation of the single particle measurements. The millions of available measurements (a typical rate is one particle per second) will be combined to produce a multi-dimensional aerosol size/composition distribution. These distributions will be compared with the ones measured by the traditional instruments (impactors, filter-based measurements, differential mobility analyzers, aerosol particle sizers, and others) and the method will be further developed if necessary. The results of the single particle instrument will also be compared against the suite of continuous measurements in the site (continuous sulfate, nitrate, metals, and size distributions). Our hypothesis is that by the end of the proposed program, an approach will be available both for the operation and the analysis of the single particle mass spectrometer data that will allow it to produce quantitative results in agreement with the traditional instruments.

Hypothesis 2.2 Semi-continuous nitrate, sulfate, carbon and elements measurement technique can quantify their concentrations under conditions prevalent at the site.

The semi-continuous measurements of nitrate, sulfate, and carbon by ADI, and the elemental measurements by Ondov and Buckley will be compared with the more traditional filter and impactor based measurements to evaluate their performance.

Hypothesis 2.3 There is a negative artifact from sampling nitrate over several hours and it can be avoided by employing semi-continuous techniques.

Size-resolved measurements of the PM in the Eastern US have indicated that the available nitrate often exists in particles larger than 1 micrometer or so, while the smaller aerosol particles are often acidic. Interaction among the two groups of particles can result in volatilization of the available nitrate, even for sampling systems using denuders and after-filters. This hypothesis will be tested by the comparison of the daily and 4-6 hr measurements with the continuous measurements of the Hering team. The sodium carbonate backup filters will also be analyzed to quantify this effect.

Atmospheric Processes

Hypothesis 3.1 Aerosol nucleation (biogenic precursors or SO₂) can be a major source of aerosol number in both urban and rural areas in the study region.

The measurements of the ultrafine aerosol number concentration both in the central site and in the rural Holbrook site will allow the Pandis group to quantify the contribution of nucleation as a source of particle number. The primary contributions to ultrafine particle concentration (e.g., from transportation) will be quantified. These contributions are usually plumes that influence the site for a few minutes at random periods. Nucleation events in urban areas like Atlanta (McMurry personal communication) and Leipzig, Germany (Weidensholer et al., 1998) last for an hour or more and sometimes are associated with significant photochemical activity but sometimes they occur early in the day. Comparing the measurements in both sites will allow us to investigate if these nucleation events are a regional phenomenon or a local phenomenon. The contribution of these events to particle number concentrations will be quantified. Correlations of these events with the measurements of SO₂, RH, Temperature and biogenic concentrations will provide insights about the species involved in the nucleation events. The single particle measurements of the University of Delaware team will be used to quantify for the first time the composition of these nuclei. Understanding the source of these particles will be a major advancement for the atmospheric chemistry research area.

Hypothesis 3.2 Biogenic primary and secondary aerosols are a major component of the organic aerosol in the Pittsburgh region.

Recent laboratory, source, and field studies (Table 3) have identified a number of tracer compounds for primary and secondary biogenic aerosols. The Rogge group will measure their concentrations and the contributions of the corresponding sources or precursors will be estimated using them as tracers (Rogge et al., 1993a).

The contribution of these biogenic sources to PM will also be estimated theoretically during the EPA-STAR project by the Carnegie Mellon team. As a third way of quantification of the biogenic contribution selected samples will be composited and sent for 14C analysis to provide the non-fossil component.

Table 3. Some Tracer compounds for Biogenic Organic Aerosols

Organic Aerosol Component	Tracer Compounds	Reference
Primary Biogenic	C27, C29, C31, C33 n-alkanes	Mazurek and Simoneit (1984) Simoneit (1984) Rogge et al. (1993c)
Secondary - a-pinene	Pinonaldehyde Pinonic acid Pinic acid	Kamens et al. (1998) Kavouras et al. (1999)
Secondary - b-pinene	Nopinone	Paulson et al. (1989)

Hypothesis 3.3 Fogs and low clouds are responsible for extreme acid sulfate conditions in the Pittsburgh region since (a) there is substantial SO₂ imported from the west, and (b) aqueous phase oxidizers, such as hydrogen peroxide, are present in significant concentrations.

Published rate laws will be used by the Collett group along with measured cloud/fog composition and gaseous H₂O₂, O₃, and SO₂ concentrations to determine rates of aqueous phase sulfate production in sampled clouds/fogs. Rates will be compared for several possible S(IV) oxidation pathways, including oxidation by H₂O₂, oxidation by O₃, and trace metal catalyzed auto-oxidation, in order to test our hypothesis that aqueous sulfate production is limited by the availability of H₂O₂. Concentrations of major ions in the cloud/fog samples (µg/l) will be multiplied by fog/cloud liquid water content (lH₂O/m³air) to determine the total concentrations of major ionic species contained in fog/cloud drops per unit volume of air (µg/m³). These values will be compared to measurements of total airborne concentrations of these species made at the site in order to determine the fraction of each species undergoing cloud/fog processing. In several cases, cloud/fog concentrations must be compared with the sum of aerosol and gas concentrations (e.g., aqueous nitrate is derived from scavenging of both aerosol nitrate and gaseous nitric acid). The contribution of the fogs in the area will be estimated combining these measurements with the frequency of these events in the study area. The possibility of these events leading to high concentrations of sulfate the day after the fog episode will be investigated.

Hypothesis 3.4 The response of PM_2.5 to changes in sulfate is highly non-linear during the winter and linear during the summer and is controlled by the ammonia availability.

Ansari and Pandis (1999) suggested that the response of the fine PM to sulfate concentration changes can be estimated by describing the atmospheric condition space by three parameters, the total (gas and particulate) nitric acid, the temperature, and a parameter:

GR=(total ammonia moles- 2*sulfate moles)/(total nitrate moles)

West et al. (1998; 1999) showed that daily aerosol samples are sufficient for the calculation of the system response to sulfate. All of the above quantities will be measured by the Davidson group, allowing the parametric investigation of the changes of the observed sulfate concentrations with respect to these three parameters. The linearity of the relationship will be investigated based only on the measurements during the 18 months of sampling (more than 1000 samples should be available). The response of the PM to total ammonia availability for constant temperature and GR values will be quantified. This analysis will provide the sensitivity of the PM concentrations to ammonia availability. The sensitivity of the PM concentrations to sulfate and nitric acid will be quantified by a similar analysis. The sensitivities of the system to these components will be quantified based on the measurements and will be compared with each other.

These results will be compared with the graphical approach of Ansari and Pandis (1999). The modeling work of the CMU group (funded by the EPA-STAR program) will be used to further investigate this relationship and also to elucidate the links between NO_x, and SO₂ emissions and nitric acid and sulfate concentrations in Pittsburgh.

Hypothesis 3.5 The secondary aerosol contribution to OC exceeds 50% during the peak PM days, but is around 20% on a yearly average basis (based on similar contributions estimated for the Western US).

The calculation of the primary and secondary organic aerosol contributions to the organic aerosol concentrations in the area will be one of the major tasks of the Supersite Project. Three different approaches (Strader et al., 1999) will be used:

(a) The OC/EC ratio method will be used by the Turpin group especially for the intensive sampling periods where 1-2 hr measurements will be available. Time-resolved elemental carbon concentrations will be used as a tracer for primary combustion-derived aerosol. The relationship between elemental carbon and organic carbon on days when secondary formation is unlikely and the approach of Strader et al. (1999) will be used to develop an expression that describes the primary organic carbon as a function of the elemental carbon concentration. Secondary organic carbon will be considered to be the difference between total and primary organic carbon. Thus, one-hour average organic and elemental carbon will be reported, as well as primary and secondary organic carbon.

(b) The organic tracer method will be used by the Rogge group to quantify the contributions of the various primary sources and to estimate the secondary organic aerosol fraction.

(c) The direct modeling method for the quantification of the secondary contribution will be used by the parallel modeling effort of the CMU/Georgia Tech. groups.

Hypothesis 3.6 The regional contributions to the PM_2.5 levels in the Pittsburgh region for some compounds exceed the local contribution, whereas for others the local exceeds the regional.

The parallel measurements in the rural satellite site in Holbrook will provide values of the regional PM concentration and composition. These results will be compared with the values measured in the central Supersite to distinguish the regional and local contributions. The PM components that are enriched in the urban area will be quantified. Air trajectories by the Kahl group for the periods of interest will be used to inform this comparison. Measurements from different sites along the same trajectory will be compared for the appropriate time periods.

The University of Delaware team will also quantify the regional or local sources for the various particles. Once the single particle spectra are classified, their spectra will be examined to determine their source and atmospheric processing they have undergone. Typical sources of the condensation nucleus that can be identified include homogeneous sulfuric acid - water and organic nucleation, soot, biogenic materials, metals, and crustal material. The nuclei then undergo atmospheric processing that includes condensation of vapors and cloud processing that increase the particulate mass of secondary compounds. Typical secondary components that can be identified include sulfates, nitrates, and aromatic condensible organics. Both composition and back trajectories will be used to identify the likely source of the nucleus and the secondary species contributions on a particle-by-particle basis.

The estimates based on measurements will be compared with the theoretical predictions of the 3D modeling effort to provide further insights into the two contributions.

Source-Receptor Relationships

Hypothesis 4.1 A complementary suite of instruments (single particle instruments, continuous composition monitors) and techniques (enhanced organic tracers, inorganic tracers) can directly determine the local air quality contributions from a broad range of sources including a) primary emissions from power plants fired by coal, oil, or gas, diesel- or gasoline-powered transportation, meat cooking, coke plants, biogenics, biomass burning, incineration, and crustal sources and b) secondary compounds emitted from power plants, and transportation systems.

Methods with single particle resolution (RSMS-II, LIBS, SEM), continuous operation (metals) and the availability of a wide range of organic and inorganic tracers provide a unique opportunity to develop the next generation of source-receptor methods that combine all of this information to apportion all the PM components to their sources. For example, it could become possible to identify a specific particle as the product of coal combustion, on which ammonium nitrate and biogenic secondary organic material has condensed. A DOE funded study by the Robinson/Pandis groups on source characterization in the area will provide source-fingerprints for the major local sources (coke ovens, power plants, transportation, and others) using the same instrumentation. The EPA/STAR modeling project will investigate theoretical methods for combining these approaches to provide daily source apportionment for the study period. The result will be daily source-resolved PM, which can be used both by local authorities for SIP development, and by the epidemiological study (see hypothesis 6).

Hypothesis 4.2 An increase in temporal resolution of elemental constituents of atmospheric aerosol coupled with sulfate and carbon analyses of comparable frequencies will permit (a) unprecedented resolution of sources by receptor modeling techniques, e.g., Factor Analysis (FA), Multi-Linear Regression (MLA), and Chemical Mass Balance (CMB) analyses, (b) resolution of plumes from individual stationary sources impacting the site, and (c) resolution of local and regional sources.

The hypothesis that higher frequency sampling rates will improve source resolution based on inorganic tracers will be tested by the Ondov group with limited applications of FA and MLR on the most highly resolved data and composited data. Evidence that plumes from stationary sources can be resolved will be tested by comparing time series plots of the elements with wind roses and results of FA and MLR for these periods. The hypothesis that local and regional sources can be resolved will be tested by comparing time series data collected simultaneously at the two sites.

Hypothesis 4.3 Specific aerosol signatures are associated with transport from specific source regions and along different altitudes.

The air trajectories calculated by the Kahl group will be statistically compared with the aerosol measurements. This statistical analysis will demonstrate the relationships between the origin and path of the air masses affecting the region and the measured aerosol concentration and chemical composition.

Aerosol Properties

Hypothesis 5.1 Visibility in the area can be predicted from RH and size/composition information obtained from aerosol sizing instruments and size-resolved bulk chemistry instruments (MOUDI).

The scattering measurements of the nephelometer will be compared with the theoretically estimated values using the visibility model of Pilinis (1989). The input to the model will be the measured size/composition of the aerosol (from aerosol number distribution measurements and impactor measurements).

Hypothesis 5.2 Most particles in the area are liquid throughout the day in both winter and summer.

The single particle measurements by the University of Delaware will provide qualitative evidence for the existence of water in the ambient particles as a function of RH and chemical composition. The TDMA evaporation experiments (gradually decreasing the RH) by the Pandis group will quantify this liquid water content for the submicrometer particles. A smooth change in size as the RH decreases will indicate the existence of metastable particles existing as supersaturated solutions. If different particles exhibit different size changes after the RH decrease the single particle instrument will be used to quantify their composition. The combination of the two techniques will provide the quantitative link between liquid water content of the particles, RH, and their chemical composition.

Hypothesis 5.3 Aerosol in the area consists of two groups of particles based on the hygroscopic properties: those consisting mainly of sulfates that grow rapidly with relative humidity and those consisting of mainly carbonaceous material that grow slowly.

The TDMA growth experiments coupled with the single particle measurements of the RSMS-II will be used to investigate the link between chemical composition of particles and their ability to grow with increasing RH. This unique experiment will measure for the first time the chemical composition of often observed "more" and "less hydrophilic" atmospheric particles.

Health Effects

Hypothesis 6 The health effects attributed to PM are due to one of the following characteristics shown in Table 1 or a combination of these.

All of these characteristics (with the exception of the individual source contribution) will be measured directly during the Supersite Program for 18 months with at least daily resolution by a variety of groups and often using multiple approaches. The measurement program will result in a comprehensive database of PM characteristics for use in epidemiological studies. The Samet group will use this PM characterization together with measurements of health outcomes (mortality, morbidity, and others) to investigate the links between the two.

Indoor Exposure

Hypothesis 7.1 There is significant seasonal variability of the contribution of outdoor PM_2.5 to indoor PM_2.5 levels.

The PM_2.5 measurements in the research house, which will be located near to the central Supersite location, will be performed for a few weeks every season. The measurements will be directly compared to the outdoor measurements to investigate the seasonal variability of indoor to outdoor fine PM concentrations.

Hypothesis 7.2 Regional housing stock characteristics influence PM_2.5 penetration.

The data collected in Pittsburgh houses will be compared to those currently being collected by the LBNL team in houses in the San Joaquin Valley. These results together with analysis of the corresponding house characteristics will be used to investigate the role of regional variations in housing stock on indoor PM exposure.

Hypothesis 7.3 Sulfate can be used as a tracer for the penetration of ambient PM indoors as a function of size.

The indoor and outdoor sulfate PM_x concentrations will be compared to establish the relationship between the corresponding concentrations as a function of PM size.

Hypothesis 7.4 Transport and fate of outdoor PM to and in the indoor environment is size and composition dependent.

For selected days the indoor and outdoor size-resolved PM concentrations of the major concentrations will be compared. The possibility of significant changes in nitrate and semivolatile organics concentrations as the particles enter the indoor environment will be investigated. We hope to use a second single particle mass spectrometer owned by the Wexler group to perform this comparison for the first time on a single particle basis.