Evaluation of the CMAQ Model for Size-Resolved PM Composition

1
Evaluation of the CMAQ Model for Size-Resolved PM
Composition
Prakash V. Bhave, K. Wyat Appel U.S. EPA, Office
of Research Development, National Exposure
Research Laboratory, Atmospheric Modeling
Analysis Division
INTRODUCTION Evaluations of CMAQ model results
for particulate matter (PM) have focused largely
on mass concentrations (e.g., PM2.5 or PM10) and
bulk chemical composition (e.g., total SO4, OC,
etc. in the PM2.5 size range). To accurately
assess the effects of PM on human health,
ecosystem exposure, radiative forcing, and
visibility, a detailed understanding of the
size-resolved chemical composition is needed. To
date, evaluations of CMAQ output at that level of
detail have been conducted only in coastal
locations (e.g., Tampa, Vancouver, and Shenzhen).
Here, we inspect CMAQ results at 6 IMPROVE sites
across the U.S. where data were collected during
different seasons using cascade impactors.
Results are contrasted against our previous study
at 3 coastal urban sites in the Tampa region.

• RESULTS
• Sulfate and Ammonium MOI data reinforce the CMAQ
  result that these 2 species are almost
  identically distributed by size. At all sites,
  CMAQ correctly simulates their peak in the lt1 µm
  range. But, CMAQs mean diameter is consistently
  biased high (especially at Smoky Mountains in
  summer) and the measured size distributions are
  consistently taller and frequently narrower than
  CMAQ.
• Nitrate At all IMPROVE sites except Bondville,
  CMAQ fails to match the observed size
  distribution. MOIs show some indication of coarse
  NO3 on crustal/soil particles, which CMAQv4.7 is
  not designed to capture.
• Sodium and Chloride CMAQ correctly simulates
  that Na peaks gt 2 µm diameter, but CMAQs
  diameter is biased high at all IMPROVE sites.
  CMAQ is missing a minor Na source at several
  IMPROVE sites and a minor Cl- source at Grand
  Canyon.
• In general, CMAQ performance at the IMPROVE sites
  is poorer than at the Tampa sites.

• METHODOLOGY
• Modeling see Wyat Appels presentation Wed. at
  830am
• CMAQ v4.7, CB05 chemistry
• Time period 2002 2006
• Horizontal grid 36 km continental 12 km
  eastern U.S.
• Vertical grid 24 layers up to 100 mb
• Emissions 2002 NEI substituted with
  year-specific fire, mobile, biogenic, point EGU
  data
• Meteorology MM5 with 34 vertical layers
• BCs month-specific from a 2002 GEOS-Chem
  simulation
• Output mode-specific variables (DG_WET, STDEV)
  from aerosol diagnostic file
• Post-processing output converted to aerodynamic
  diameter using the mode-specific particle density
• Measurements
• Micro-orifice impactors (MOI) segregate ambient
  particles by size on impaction substrates, which
  are subsequently analyzed by ion chromatography
  for SO4, NH4, NO3-, Na, and Cl-.
• MOIs were operated with 8 or 10 stages (dlnDp
  0.6), that spanned the 0.056 18 µm particle
  size range.
• Data at 6 IMPROVE sites were collected during
  2003 2004 in 48h sampling intervals by
  Taehyoung Lee et al. (Atmos. Environ.,
  422720-2732, 2008).
• Data at 3 Tampa sites were collected during
  May/June 2002 in 23h sampling intervals by
  Melissa Evans et al. (Atmos. Environ.,
  384847-4858, 2004).

Tampa Bay area

• Figure Caption Legend
• CMAQ results were paired in space and time with
  MOI observations. Although sample-specific
  comparisons are possible (e.g., 48h or 23h),
  results are averaged across all sampling
  intervals (n 12 to 15) for this display.
• MOI data are shown in black CMAQ output in red.
• Plots are color-coded to emphasize seasonal
  differences winter, spring, summer, fall.
• Vertical axes are scaled to accentuate the shapes
  of each size distribution

• FUTURE WORK IMPLICATIONS
• Gather any other ambient MOI data from 2002
  2006 to augment this evaluation of CMAQ.
• Invert the MOI data to account for imperfect size
  cuts. This should dampen some of the tallest
  peaks.
• Reconsider the application of sharp 2.5 µm size
  cuts for regulatory-grade model evaluations.
• Assess the impact of the particle-size biases
  identified here on CMAQ calculations of dry
  deposition, visibility degradation, and radiative
  forcing.
• Improve CMAQ model algorithms to reduce biases in
  the PM size distributions.

• ACKNOWLEDGEMENTS
• Chris Nolte, Jim Kelly, Shawn Roselle (U.S. EPA)
• Taehyoung Lee, Jeffrey Collett (Colorado State
  Univ.)
• Melissa Evans, Noreen Poor (Univ. of South
  Florida)
• now at California Air Resources Board

U.S. Environmental Protection Agency Office of
Research and Development