The Pittsburgh Air Quality Study (PAQS) [PDF-744KB] is comprised of three inter-related components: 1) ambient PM measurements, 2) source characterization, and 3) deterministic and statistical air quality modeling. This effort will permit clarification of the contribution of coal-fired power plants to fine ambient PM2.5 (particulate matter with an aerodynamic diameter less than 2.5 µm). The resources from the Department of Energy (DOE) will be leveraged with resources from the Environmental Protection Agency (EPA) and other organizations.

Clarkson University (Hopke group) will apply advanced receptor models to identify the nature, location and contribution of the sources of particulate matter observed by the measurements made as part of the PAQS. Several forms of factor analysis including Positive Matrix Factorization (PMF) and UNMIX will be applied in order to identify the composition and contributions of the sources. Potential Source Contribution Function analysis as well as Residence Time Weighted Concentration analysis will be applied to the determination of the locations of the likely major contributing sources. The aforementioned factor analysis methods will also be applied to the spatially distributed data both on a single species and multiple species basis and to compare these results with those obtained utilizing the back-trajectory-based methods. The availability of highly time resolved data should permit greater source resolution and will be examined to determine how much increased source specificity can be obtained from the increased time resolution in the data. Assistance will be provided with the multivariate calibration that will permit the use of single-particle mass spectrometry data to estimate ambient concentrations of particulate species. These analyses should provide a better understanding of the source/receptor relationships that lead to the observed particle concentrations in the Pittsburgh area.

Single particle measurements provide an opportunity to improve upon the currently available source-apportionment capabilities. These measurements offer the possibility to create source fingerprints with multiple dimensions (chemical composition and size of individual particles). The addition of APS-IMS capabilities to RSMS-II will allow improved identification of the sources of both inorganic and organic, and fine and coarse particles. Combining these fingerprints with similar ambient single particle measurements could improve the identification of the origin of each individual particle measured in the ambient atmosphere. This work will be done in conjunction with the Carnegie Mellon University, University of California at Davis, Clarkson University, and University of Maryland groups.

The ART2a neural network based algorithm will be used to categorize the particles into like composition classes, as the particles are analyzed in the field. Later, comparisons of the source-oriented and the ambient-particle compositions will be made with ART2a to identify source categories associated with the ambient particles that are detected.

Advanced methods for source apportionment will be developed by combining the single particle source fingerprints with classified atmospheric particles (characterized with support from EPA). The approach will be based on existing Factor Analysis and Chemical Mass Balance approaches. The approach will likely combine both composition and back trajectories. The hope is that these will be able to identify the source on a particle-by-particle basis. It is anticipated that this approach will yield insight into both sources and atmospheric processing that particles have undergone. Typical sources of the condensation nuclei that can be identified include organic nucleation, soot, biogenic materials, metals, and crustal material. The nuclei then undergo atmospheric processes such as condensation of vapors and cloud processing to increase their secondary mass content. Typical secondary materials that can be identified include sulfates, nitrates, and aromatic condensable organics. Both composition and back trajectories will be used to identify the likely source of the nuclei and the secondary species contributions on a particle-by-particle basis.

Three-Dimensional Deterministic Modeling
The model developed by the EPA-STAR funded Research Consortium on Ozone and Fine Particle Formation in California and in the Northeastern United States will be used to simulate the air quality in the Pittsburgh region. The model is an extension of the URM model. It describes the evolution of the aerosol size-composition distribution (using 12 or more chemical components and 10 or more size sections) and for approximately 100 gas-phase species for every grid cell in the three-dimensional modeling domain. Aerosol chemical components simulated include: sulfate, nitrate, chloride, sodium, ammonium, elemental carbon, primary and secondary organic carbon, crustal elements, H⁺, water, and others. The model uses first principles to simulate spatially and temporally resolved emissions, gas-phase chemistry, advection, dispersion, aerosol dynamics and chemistry, cloud chemistry, and dry and wet removal processes. It is also coupled to a sensitivity analysis module so it can calculate directly the sensitivities of PM concentrations to source strength.

Simulations will be performed for selected periods during the intensive sampling periods. The region around Pittsburgh will be described with high spatial resolution (5x5 km). The focus of these simulations will be on:

• Model evaluation using the ambient PM measurements.
• Estimating the lifetime and transport distances of PM reaching Pittsburgh.
• Quantifying the relationships between total nitric acid and sulfate in Pittsburgh and the NOx and SO₂ emissions in the modeling domain.
• Investigating the sensitivity of the PM in the area to these NOx, SO₂, and VOC emissions
• Quantifying the contribution of primary organic aerosol sources to the Organic Carbon (OC) in Pittsburgh
• Estimating the contribution of primary and secondary biogenic aerosol to the organic PM in the study area.