Modeling Regional Air Quality and Climate: Improving Organic Aerosol and Aerosol Activation Processes in WRF/Chem version 3.7.1
2016
Air quality and climate influence each other through the uncertain
processes of aerosol formation and cloud droplet activation. In this study,
both processes are improved in the Weather, Research and Forecasting model
with Chemistry (WRF/Chem) version 3.7.1. The existing Volatility Basis Set (VBS)
treatments for organic aerosol (OA) formation in WRF/Chem are improved
by considering the following: the secondary OA (SOA) formation from semi-volatile primary
organic aerosol (POA), a semi-empirical formulation for the enthalpy of
vaporization of SOA, and functionalization and fragmentation reactions
for multiple generations of products from the oxidation of VOCs. Over the continental US, 2-month-long simulations (May to June 2010) are conducted and
results are evaluated against surface and aircraft observations during the
Nexus of Air Quality and Climate Change (CalNex) campaign. Among all the
configurations considered, the best performance is found for the simulation
with the 2005 Carbon Bond mechanism (CB05) and the VBS SOA module with
semivolatile POA treatment, 25 % fragmentation, and the emissions of
semi-volatile and intermediate volatile organic compounds being 3 times
the original POA emissions. Among the three gas-phase mechanisms (CB05, CB6,
and SAPRC07) used, CB05 gives the best performance for surface ozone and
PM2. 5 concentrations. Differences in SOA predictions are larger for
the simulations with different VBS treatments (e.g., nonvolatile POA versus
semivolatile POA) compared to the simulations with different gas-phase
mechanisms. Compared to the simulation with CB05 and the default SOA module,
the simulations with the VBS treatment improve cloud droplet number
concentration (CDNC) predictions (normalized mean biases from −40.8 % to a range of
−34.6 to −27.7 %), with large differences between CB05–CB6 and SAPRC07 due to
large differences in their OH and HO2 predictions. An advanced aerosol
activation parameterization based on the Fountoukis and Nenes (2005) series reduces the large
negative CDNC bias associated with the default Abdul Razzak and Ghan (2000) parameterization from
−35.4 % to a range of −0.8 to 7.1 %. However, it increases the
errors due to overpredictions of CDNC, mainly over the northeastern US. This
work indicates a need to improve other aerosol–cloud–radiation processes in
the model, such as the spatial distribution of aerosol optical depth and cloud
condensation nuclei, in order to further improve CDNC predictions.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
79
References
8
Citations
NaN
KQI