CRC Report No. A-34 - Coordinating Research Council

April 2005
Comparing the performance of CMB in Rounds 2b and 4 (Figure ES-1) lead to several findings:
• In Round 2, the CMB analysis apportioned VOCs to categories called “CNG and Aged”
and “LPG” that were not actually present in the samples, illustrating that CMB category
names may describe chemical characteristics (fingerprints) rather than specific activities
tracked in emission inventories. CNG is compressed natural gas and LPG is liquid
petroleum gas.
• CMB tended to over-estimate the contribution of gasoline emissions except when
complete profile information was available in Round 4.
• CMB performance for diesel was poorer in Round 4 than Round 2 because of a decision
to exclude nonane, decane and undecane as CMB fitting species in Round 4 that resulted
in partial colinearity between the diesel and gasoline source profiles.
The results presented in Figure ES-1 show that CMB can achieve accurate results with adequate
supporting information but that even with detailed supporting information, results can be biased
by decisions made during the analysis. Consequently, it may be difficult to judge the accuracy of
CMB results in each application.
The comparisons presented in Figure ES-1 are based on the actual source contributions to VOCs
present in air samples. However, CMB results are frequently interpreted as measures of
emission inventories. There are at least two major reasons why actual contributions may differ
from emission contributions; namely chemical reaction and spatial heterogeneity in emissions.
The impacts of chemical reaction on source contributions are illustrated in Figure ES-2 for
several source categories at a downwind receptor (Crestline). Figure ES-2 compares source
category contributions from the photochemical model with and without chemical degradation
included. Low reactivity VOCs (CNG/aged and LPG) are only slightly depleted by chemical
degradation, whereas high reactivity VOCs (biogenics) are almost completely depleted during
the night and highly depleted at day. Differences in VOC degradation rates introduce a bias
toward low reactivity VOCs having higher contributions to actual concentrations than to
emission inventories.
H:\crca34-receptor\report\Final\ExecSum_r.doc
ES-3