Development of an Optimum Tracer Set for Apportioning Emissions of Individual Power Plants Using Highly Time-Resolved Measurements and Advanced Receptor Modeling

Abstract

In previous studies, 11 elements (Al, As, Cd, Cr, Cu, Fe, Mn, Ni, Pb, Se, and Zn) were determined in 30-minute aerosol samples collected with the University of Maryland Semicontinuous Elements in Aerosol Sampler (SEAS; Kidwell and Ondov, 2001, 2004; SEAS-II) in several locations in which air quality is influenced by emissions from coal- or oil-fired power plants. At this time resolution, plumes from stationary high temperature combustion sources are readily detected as large excursions in ambient concentrations of elements emitted by these sources (Pancras et al. ). Moreover, the time-series data contain intrinsic information on the lateral diffusion of the plume (e.g., {sigma}{sub y}), which Park et al. (2005 and 2006) have exploited in their Pseudo-Deterministic Receptor Model (PDRM), to calculate emission rates of SO{sub 2} and 11 elements (mentioned above) from four individual coal- and oil-fired power plants in the Tampa Bay area. In the current project, we proposed that the resolving power of source apportionment methods might be improved by expanding the set of maker species and that there exist some optimum set of marker species that could be used. The ultimate goal was to determine the utility of using additional elements to better identify and isolate contributions of individual power plants to ambient levels of PM and its constituents. And, having achieved better resolution, achieve, also, better emission rate estimates. In this study, we optimized sample preparation and instrumental protocols for simultaneous analysis of 28 elements in dilute slurry samples collected with the SEAS with a new state-of-the-art Thermo-Systems, Inc., X-series II, Inductively Coupled Plasma Mass Spectroscopy (ICP-MS), and reanalyzed the samples previously collected in Tampa during the modeling period studied by Park et al. (2005) in which emission rates from four coal- and oil-fired power plants affected air quality at the sampling site. In the original model, Park et al. (2005), included 6 sources. Herein, we reassessed the number of contributing sources in light of the new data. A comprehensive list of sources was prepared and both our Gaussian Plume model and PMF were used to identify and predict the relative strengths of source contributions at the receptor sites. Additionally, PDRM was modified to apply National Inventory Emissions, Toxic Release Inventory, and Chemical Mass Balance source profile data to further constrain solutions. Both the original Tampa data set (SO{sub 2} plus 11 elements) and the new expanded data set (SO{sub 2} plus 23 elements) were used to resolve the contributions of particle constituents and PM to sources using Positive Matrix Factorization (PMF) and PDRM.