Prakash V. Bhave, Ph.D.

1
Postprocessing Model Output for Comparison to
Ambient Data

• Prakash V. Bhave, Ph.D.
• Physical Scientist
• PM Model Performance Workshop
• February 10, 2004

2
Traditional PM Model Evaluations
Model Evaluation
3

• Model Tuning

• Model Postprocessing

Modification of model inputs or atmospheric
chemical physical formulae to improve agreement
between model results and observations. e.g.,
multiply NH3 emissions by 2 and rerun the
model e.g., raise or lower the minimum eddy
diffusivity and rerun the model e.g., modify a
rate constant and/or reaction probability and
rerun the model
4
Enhancing PM Model Evaluations
5
Ambient Equilibration

• Models predict both gas and particle-phase
  concentrations
• Gas/particle partitioning of semi-volatile
  species is a function of T, RH
• What if the met model provides a gridded
  temperature field that is warmer than at the
  site?i.e., Tmodel gt Tambient

6
Ambient Equilibration
7
PM2.5 Inlet Simulation
PM2.5 16.7 mg/m3
8
PM2.5 Inlet Simulation
PM2.5 15.5 mg/m3
9
PM2.5 Inlet Simulation
10
PM2.5 Inlet Simulation
PM2.5 14.9 mg/m3
11
Other Postprocessing Steps

• Substrate Artifacts
• Adsorption/desorption from substrate during
  sampling
• Have been characterized experimentally
• Can be estimated with moderate difficulty
• Chemical Analyses
• Lab conditions 20-23 ºC, 30-40 RH
• Exceeds CRH of NH4HSO4, NH4NO3, NaHSO4,
• Aerosol water content can be estimated
• Organic Mass/Carbon Ratio
• Measurements reported as Organic Carbon
• Models assume some ratio (e.g., POA 1.2 in NEI,
  SOAb 1.475 in CMAQ, SOAa 1.67 in CMAQ)

12
Acknowledgements

• Pete Finkelstein, ORD
• John Irwin, OAQPS
• Brian Timin, OAQPS
• Rob Gilliam, ORD
• Brian Eder, ORD
• Ken Schere, ORD
• S.T. Rao, ORD
• Disclaimer Notice
• This work has been funded wholly by the United
  States Environmental Protection Agency. It has
  been subjected to Agency review and approved for
  presentation.

13
Atmospheric Processes Modeling