Project at a Glance
Project Status: complete
Title: Multivariate receptor modeling of SCAQS VOC and airborne particle composition data
Principal Investigator / Author(s): Henry, Ronald C
Contractor: University of Southern California
Contract Number: A832-131
Research Program Area: Atmospheric Processes
Innovative multivariate receptor modeling techniques were applied to several types of data collected during the Southern California Air Quality Study (SCAQS). The purpose of the modeling was to deduce as much information about the sources as possible from the data alone and with the aid of some basic physical constraints. The data examined were the Volatile Organic Compound (VOC) gas phase data, the airborne particle composition data, and routine gas monitoring data, The models applied were the Source Apportionment by Factors with Explicit Restrictions (SAFER) multivariate receptor model, and the Source Identification Through Empirical Orthogonal Functions (SITEOF) hybrid source -receptor model.
The VOC data from all sites for the fall intensive were combined into a singe data set that had a sufficient number of data points for multivariate analysis. The summer VOC data differed too much from site to site to be consolidated into a single set and the number of data points at each site was insufficient to support a multivariate analysis. Three factors were found to explain almost all the variability of the fall data. A graphical, three-dimensional version of the SAFER model determined the composition of these three sources, which could be identified as roadway (direct tailpipe plus running evaporative) emissions, whole gasoline, and gasoline vapor. On average, 60 % of the Total Non-Methane Hydrocarbons (TNMHC) were apportioned to roadway emissions, 19 % to whole gasoline, 15 % to gasoline vapor, and 6 % were unexplained or to background levels of TNMHC. There was a significant difference between the morning and afternoon source apportionment. In the morning, 71 % of the TNMHC were apportioned to roadway emission, 14 % to whole gasoline, 11 % to gasoline vapor, and 4 % were unexplained. In the afternoon, 44 % of the TNMHC was apportioned to roadway emission, 27 % to whole gasoline, 20 % to gasoline vapor, and 10 % were unexplained. This is consistent with increased evaporative emissions from parked vehicles at higher afternoon temperatures.
Multivariate receptor modeling of the particle composition data was restricted to summer PM-10 from five sites: Anaheim, Azusa, Burbank, Hawthorne, and Long Beach. The summer PM-2.5 and all the fall particle composition data sets had too little data for a valid multivariate analysis. Two source types, roadway and soil, account for most of the non-secondary PM-10. The composition and contributions of these sources were estimated by a two-dimensional version of the SAFER model. Roadway is defined as all PM-10 emissions from roadways, including direct tailpipe emissions and reentrained road dust. Soil is defined as all crustal material not directly associated with roadway emissions. The composition of roadway and soil sources varies little between the five sites. The average PM-10 source apportionment allocates 30 % of the total PM-10 to roadway, 18 % to soil, 35 % to inorganic secondary species, and the remaining 17 % represents, by process of elimination, an upper limit on the amount of secondary organic particulate matter.
The hybrid receptor modeling was unsuccessful because of the limitations in the wind fields.
For questions regarding this research project, including available data and progress status, contact: Heather Choi at (916) 322-3893
Stay involved, sign up with ARB's Research Email Listserver