Aerosols in the atmosphere - Air Pollution Modeling and Its Application - page 382

Figure 1a shows that in the summertime POA concentrations are only significant
near heavily urbanized areas. In the summer, photochemical aging of these evaporated
emissions creates large amounts of regional OPOA ( Fig. 1b) . Wintertime simulations
show a somewhat larger fraction of the primary organics partitioning into the
particle phase ( Fig. 1d) and less production of OPOA (Fig. 1e) due to the combination
of lower temperatures and oxidant levels.
The substantial evaporation of existing POA is due to multiple factors. Although
dilution samplers are used to measure POA emission factors, a large fraction of
the low-volatility organics has been misclassified as POA because these samplers
are often operated at unrealistically high concentrations that are orders of magnitude
higher than typical ambient levels. This biases gas-particle partitioning towards
the particle phase relative to atmospheric conditions (Lipsky and Robinson, 2006;
Shrivastava et al., 2006). Interestingly, the same problem exists in many of the
chamber SOA data sets, which were collected at unrealistically high OA concent-
rations. This underscores the need for future source tests (and chamber experiments)
to be conducted at atmospherically relevant temperatures and concentrations. In
addition, quartz filters used to measure POA concentrations collect a substantial
amount of organic vapors (positive artifact) during source tests (Lipsky and
Robinson, 2006). The net effect of these two problems is that POA emission factors
greatly overestimate the amount of POA that exists at typical atmospheric conditions.
The aging of evaporated POA reduces its volatility, and shifts its partitioning into
the condensed phase. The net result is the production of significant amount of
oxidized OA.
Explicit accounting of partitioning and aging of primary emissions only has a
modest effect on the total amount of OA (much less than a factor of 2 throughout
the domain) relative to a traditional model that assumes POA is non-volatile
(Robinson et al., 2007). The change is modest because CTMs based on traditional
emission inventories already contain substantial amounts of low-volatility organics,
albeit misclassified as POA. For misclassified emissions, the effects of partitioning
and aging partly offset each other, resulting in modest changes to the total OA
concentrations but substantial increases in the fractional contribution of oxidized
OA. If one only accounts for partitioning and aging of the existing primary emissions,
the predicted OA levels are lower than the traditional model. In order to create
additional OA, one must add emissions to the inventory above and beyond the
existing POA emissions. The limited available data suggests that the traditional
inventories underestimate the emissions of low volatility organic vapors by a factor
of 2-3. Accounting for these emissions has the potential to increase predicted OA
concentrations by 10-50%. Therefore this mechanism has the potential to help
close the gap between model predictions and ambient observations.
A more significant change associated with the revised framework for primary
emission is the primary-oxidized OA split. Accounting for partitioning aging shifts
the split towards oxidized OA throughout the domain; this brings model predictions
into much better agreement with the ambient AMS data. Sensitivity analysis reveals
that the PMCAMx cannot predict the high levels of observed oxidized OA unless
a substantial fraction of the existing non-volatile primary emissions evaporate