CRC Report No. A-34 - Coordinating Research Council

April 2005
28. Spatial heterogeneity in emissions inventories had a greater impact than atmospheric
chemical degradation on the differences between the source category contributions found in
ambient air and local emissions.
Measures of Total Organic Compounds
29. Comparing different measures of total organic compounds (e.g., PAMS, ROG) between
receptor modeling and emissions inventories will introduce biases (either positive or
negative) in comparisons of source category contributions.
30. Receptor modeling studies should state the measure of organic compounds that is
apportioned by the receptor model and define any conversion factors used to adjust this to a
different basis (e.g., PAMS/ROG ratios for each source category).
31. If adjustment factors are used (e.g., PAMS/ROG ratios) they must be consistent with the
source profiles used for the receptor modeling and ideally should be developed specifically
for the study along with source profiles; i.e., the adjustment factors and source profiles
should be developed from the same set of detailed VOC measurements.
5.2 EVALUATION OF STATED ASSUMPTIONS FOR CMB
The stated assumptions for CMB applied for VOCs (Watson, Chow and Fujita, 2001) are
evaluated based on the findings of this project.
1. The composition of source emissions is constant over the period of ambient and source
sampling.
The need for this assumption is obvious. In general, this assumption is likely to be satisfied
when VOC samples are collected so as to average many individual sources of the same type:
E.g., the source profiles for individual cars may vary, but averages over sub-populations of cars
in an urban area are similar. Because the modeling experiments in this study had 5-km grid cells
they tended to provide “regionally representative” ambient samples where many sources had
been averaged. Real-world VOC samples may be influenced by single sources, such as a nearby
stationary source, where averaging over multiple sources does not tend to eliminate temporal
variability in source profiles.
Experiment 7 varied the source profiles for all source categories randomly in time (hour to hour)
and space (grid cell to grid cell). CMB was robust against this random variation in source
profiles (findings 6 and 7) because the modeled atmosphere (diffusion and advection) and
averaging results over samples tended to average out the random perturbations.
2. Chemical species do not react with each other, i.e., they add linearly.
This assumption was met by the chemical reaction scheme for the 55 PAMS species employed in
this study. There are no known reactions among the 55 PAMS species (or other VOCs) that
violate this assumption for the dilute concentration levels found in ambient VOC samples.
H:\crca34-receptor\report\Final\sec5.doc 5-4