CRC Report No. A-34 - Coordinating Research Council

April 2005
3.0 RECEPTOR MODELING METHODOLOGY
Receptor models infer contributions from different source types using multivariate measurements
taken at one or more receptor locations and the abundances of chemical components in source
emissions. Receptor models include the Enrichment Factors (EF), Chemical Mass Balance
(CMB), eigenvector analysis (also termed Principal Component Analysis [PCA], Factor Analysis
[FA], and Empirical Orthogonal Functions [EOF]), Multiple Linear Regression (MLR), neural
networks, cluster analysis, Fourier Transform time series, and a number of other multivariate
data analysis methods. CMB is well established for VOC apportionment and is the receptor
model that is evaluated in this study. The review by Watson et al. (2001) examines how the
CMB receptor model has been applied to quantify ambient volatile organic compound (VOC)
source contributions to ambient concentrations of organic gases.
DRI recently prepared a protocol for applying the CMB to Photochemical Assessment
Monitoring Station (PAMS) VOC data and for evaluating and interpreting model outputs (Fujita
and Campbell, 2003). The guidance includes a summary of the fundamentals of CMB,
descriptions of the features of CMB Version 8, and sample VOC source and ambient input data
files, default source and fitting species selection files, and a current library of available source
VOC composition profiles in CMB8-ready format. The applications and validation protocol
provides recommended procedures for validating ambient VOC data, assigning uncertainties to
ambient and source measurements, selecting and evaluating source composition profiles and
fitting species, evaluating and validating model outputs, and analyzing and interpreting the CMB
source contribution estimates and associated uncertainties. The document and supporting files
are intended to facilitate and encourage the application of the CMB receptor model to PAMS
VOC data by State and Local air pollution agencies as an evaluation of emissions inventories.
The actual profiles are available electronically. This library is a compilation of source profiles
that have been used by the Desert Research Institute in prior VOC source apportionment studies.
They include profiles that were newly developed for specific studies, the literature, and from the
California Air Resources Boards Modeling Emissions Data System (MEDS). Studies for which
profiles were newly developed include the 1993 Coast Oxidant Assessment for Southeast Texas
(Fujita et al., 1995b), 1995 Boston and Los Angeles VOC Source Apportionment Study (Fujita et
al. 1997a), 1995/96 Washington Ozone Transport Study (Fujita et al., 1997c), 1996 El
Paso/Juarez Ozone Study (Fujita, 1998; Seila et al., 2001), and 1998 Central Texas On-Road
Hydrocarbon Study (1999a), 1999 VOC Source Signatures in Houston, TX (Fujita et al., 1999b),
apportionment of 1994-97 South Coast Air Basin PAMS VOC data (Fujita and Campbell,
2003b), and the 2000 Weekend Ozone Observations in the South Coast Air Basin (Fujita et al.,
2003a).
3.1 FUNDAMENTALS
The CMB procedure requires: 1) identification of the contributing source types; 2) selection of
chemical species to be included; 3) estimation of the fractions of each chemical species
contained in each source type; 4) estimation of the uncertainties to both ambient concentrations
and source compositions; and 5) solution of the chemical mass balance equations. The CMB
model assumes that: 1) compositions of source emissions are constant over the period of ambient
and source sampling; 2) chemical species do not react with each other, i.e., they add linearly; 3)
H:\crca34-receptor\report\Final\sec3.doc 3-1