Evaluation of MANEVU CMAQ Annual Modeling on PM2.5 Species

1
Evaluation of MANE-VU CMAQ Annual Modeling on
PM2.5 Species

• Shan He, John Graham, Jung-Hun Woo, Emily
  Savelli, and Gary Kleiman
• NESCAUM
• Review of Application and Assessment of CMAQ in
  OTC
• Albany, NY
• November 16, 2005

2
Acknowledgement

• Cooperative CMAQ modeling team members
• NYDEC !
• NJDEP Rutgers !
• UMD !
• VADEQ !
• and NESCAUM !

3
Outline

• Background
• - CMAQ modeling overview (meteorology, emission,
  CMAQ configuration)
• - Observations used in evaluation (STN, IMPROVE,
  CASTNET)
• Evaluation and Analysis
• - Performance on PM2.5 species (SO4, NO3, NH4,
  OC, EC,
• Fine Soil, PM2.5)
• - CMAQ performance from other RPOs
• - Visibility parameters (Extinction Coefficient,
  and Haze Index)
• - PM2.5 composition on 20 best visibility days
  and 20 worst visibility
• days at Class I areas in Northeastern U.S.
• Summary

4
CMAQ Modeling

• CMAQ v4.4 (2004 Release) running on Linux
  Clusters
• CB-IV gas-phase chemistry, EBI solver, AERO3 and
  AERO_DEPV2 module for aerosol dynamics and
  aerosol deposition, RADM for cloud chemistry
• Eastern U.S. domain from 66oW94oW in longitude
  and 29oN50oN in latitude with 172X172 grid at
  12km resolution and 22 vertical layers with the
  first level at 10m and a radiative upper-boundary
  condition at 50hPa
• Dynamic boundary condition generated from annual
  CMAQ modeling on 36km US domain which using
  boundary condition derived from global chemistry
  model GEOS-CHEM
• Meteorology generated by UMD running NCAR/PSU MM5
  V3.6.1 using the Blackadar high-resolution
  planetary boundary layer parameterization scheme,
  Lambert Conformal map projection, and 29
  vertical layers. MM5 meteorology processed with
  MCIP for CMAQ input
• Emission processed by SMOKE with UMD MM5 12km
  meteorological field using RPO 2002 NEI, and CEM
  data (except MANE-VU) using Mobile 6.2 for
  on-road EI processing and BEIS 3.12 for biogenic
  emissions estimate

5
CMAQ Modeling Domain
6
Observation Networks

• STN (urban) Daily SO4, NO3, NH4, EC, OC and
  PM2.5
• IMPROVE (rural) Daily SO4, NO3, EC, OC, Fine
  Soil, PM2.5, and Bext
• CASTNet (sub-urban and rural) Weekly SO4, NO3,
  and NH4

7
Comparison of Sulfate
8
Comparison of Sulfate (II)
9
Comparison of Nitrate
10
Comparison of Nitrate (II)
11
Comparison of Ammonium
12
Comparison of Ammonium (II)
13
Comparison of Organic Carbon
14
Comparison of Elemental Carbon
15
Comparison of Fine Soil
16
Comparison of PM2.5
17
Model Performance from other RPOs
Model Performance Evaluation Michael Ku Denver,
June 2005
Sulfate
Nitrate
OC
EC
Soil
18
Comparison of Extinction Coefficient
19
Visibility (deciview) in Acadia, ME
IMPROVE
CMAQ
20
Visibility (deciview) in Brigantine, NJ
IMPROVE
CMAQ
21
Visibility (deciview) in Lye Brook, VT
IMPROVE
CMAQ
22
Visibility (deciview) in Shenandoah, VA
IMPROVE
CMAQ
23
Comparison of Haze Index at 4 Class I areas
24
PM2.5 and Visibility
25
PM2.5 Composition and Visibility
26
Summary

• MANE-VU CMAQ modeling for 2002 base year shows
  reasonably well performance on PM2.5 species, and
  visibility. Its consistent with other RPOs
  modeling performance
• Model captures general trend of PM2.5 species
  throughout year with bias. PM2.5 underestimated
  in Summer due to OC and Soil peak, overestimated
  in Fall due to Sulfate, and overestimated at
  beginning and end of the year due to Nitrate
• HI of 20 (best or worst visibility) modeling
  days agrees with HI of 20 (best or worst
  visibility) measured days
• Linear relationship observed between average
  PM2.5 concentration at 20 worst visibility days
  and average PM2.5 concentration at 20 best
  visibility days
• Sulfate is the dominant species making observed
  clean days to dirty days while Sulfate and
  Nitrate both are major contributors making
  modeling clean days to dirty days