2003 CMAS Workshop Community Scale Air Toxics Modeling with CMAQ by Jason Ching ARL,NOAA

1
2003 CMAS WorkshopCommunity Scale Air Toxics
Modeling with CMAQ by Jason ChingARL,NOAA
USEPA, RTP, NC, USAOctober 27-29, 2003
2
Contributors

• A. Lacser (Visiting Scientist from IIBR)
• T. Otte (ARL,NOAA)
• S. Dupont (UCAR Postdoc)
• J. Herwehe (ATDD/ARL, NOAA)
• R. Tang (CSC)

3
PROJECT CONTEXTAir Quality, Exposure Modeling

• NAAQS Traditional threshold goals
• Toxics Risk Based Strategy
• Community assessments
• RISK ASSESSMENT PARADIGM
• Source-Concentration-Exposure-Dose-Effects

4
General Steps in Performing a Risk Assessment
Emissions
Obtain concentrations of chemical in the medium
at distance of interest
Determine exposure of the population of interest
Calculate the risk of injury associated
with that exposure
5
Exposure Assessment Can Be Done At Different
Scales
Organ Level
Personal Level
6
Study Objectives, Approaches

• Objective Develop State-of-Science
  emissions-based, air quality grid modeling
  capability as tools to support (CAA-A90) NATA,
  air toxics community modeling exposure and risk
  assessments, and NAAQS implementations .
• Approach I Resolve air quality concentration
  distribution in urban areas at a horizontal scale
  resolution needed for (population) exposure
  models and for hot spots assessments (Census
  tract scale or finer).
• Approach II Model subgrid scale pollutant
  concentration distributions to complement the
  resolved scale fields as inputs to population and
  other exposure assessments.

7
Neighborhood Scale (N-S) Prototype Paradigm

• Air Quality modeling at N-S provides value when
  significant variability is present at that
  scale.
• Both Resolved and Subgrid concentration
  distributions are provided and needed for human
  exposure assessments
• CMAQ provides grid resolved concentrations
• Subgrid from local sources and photochemistry in
  turbulent flows to be modeled as PDFs of the
  concentration distribution histograms
• Urban focus in N-S for population exposure needs

8
Urban canopy parameterization (UCP) modeling
methodology

• Introduce UCPs into MM5 (and increase number of
  vertical layers inside the urban canopy)
• Lacser and Otte UCP methodology in MM5
• Modified DA-SM2U (with gridded UCP) in MM5
• Prepare gridded UCP fields for MM5, CMAQ based on
  high resolution raster and vector database of
  building and vegetation and urban features

9
Philadelphia Case Study

• 14 July 1995 (sunny day).
• MM5 has been run in a one-way nested
  configuration 108, 36, 12, 4 and 1.33 km
  horizontal grid spacing.
• UCPs used only for the 1.33 km domain.
• Turbulent scheme model Gayno -Seaman PBL with
  the turbulent length scale of Bougeault and
  Lacarrere (1989).

10
Results and findings

• I MM5 CMAQ sensitivity to UCP (See Dupont
  et al., details in session elsewhere in CMAS
  Workshop)
• II. Concentration fields at different grid
  resolutions
• III. Sub-grid spatial variability in coarse grid
  simulation using N-S results

11
II. Modeled Concentration Sensitivity to grid
resolution

• Top left 36 km (no UCP)
• Top Right 12 km (no UCP)
• Bottom left 4 km (no UCP)
• Bottom right 1.3km (UCP applied)
• July 14, 1995 _at_1800 EDT

12
CO Jul 14, 95, 6pm (local)
13
NOx (Jul 14, 95) 6 pm local
14
Ozone (Jul 14,95) 6 pm local
15
Aldehydes (can) 6pm localHCHO
CH3CHO
16
III. Results of multi-scale analyses

• Statistics on sub-grid variability at 12 and 4
  km grid cell resolution derived using outputs
  from 1.3 km grid (N-S) simulations
• Provides initial guidance on goal to develop
  generalized formulations for the gridded PDFs to
  represent subgrid variabilities, SGVs.

17
Ozone _at_ 4 PM EDT (12 Km)Top Left (Mean from
1.3) Bottom Left (Parent _at_ 12 km) RHS Mean
-Parent
18
NOx _at_7 EDT,(4 km grid)Top Left Mean from 1.3km,
Bottom Left Parent _at_ 4kmRHS Difference (Mean
from Parent)
19
CO _at_ 07 EDT Top 12 Km Grid Bottom 4 Km Grid
Grid means Std Dev/ Mean
20
Formaldehyde_at_ 15 EDTTop (12 km grid), Bottom 4
km gridGrid means (from 1.33) Range-to-means
21
Formaldehyde_at_ 07 EDTTop (12 km grid), Bottom (TS
for Central Philadelphia)Grid means (from 1.33)
Range-to-means
22
Skewness (Formaldehyde) _at_ 07 EDT 12 km
grid)Left 1 grid west of CP Center Central
Philadelphia (CP) Right 1 grid east of CP
23
Skewness (12 km) Top LHS CO
RHS OzoneBottom LHS
Acetaldehyde RHS NOx07 EDT
16 EDT
24
Kurtosis (12 km) Top LHS CO
RHS OzoneBottom LHS
Acetaldehyde RHS NOx07 EDT
16 EDT
25
Acetaldehyde (07 EDT)(4km grid from 1.3 km
simulations)Top LHS Mean
RHS Std DevBottom LHS Skewness RHS
Kurtosis
26
Concentration Distribution CO Ozone
NOx Acetaldehyde Formaldehyde
27
Generating gridded PDFs
28
FINDINGS Fine scale concentration distributions

• Characteristics of spatial concentration
  distribution patterns is dependent on grid
  resolution details of spatial features differs
  for different pollutant species.
• Compositing N-S simulations to coarser scales
  yield different results when compared to coarse
  grid native simulations
• N-S modeling (1.3 km) provides initial insights
  on sub-grid spatial concentration distributions
  at coarser grid sizes. (Preparatory to full
  method development for PDFs )
• Fine scale grid simulations provide indications
  of variability in coarser grid solutions
  Variability dependent on the scale of the coarse
  grid mesh.
• Initial survey of results Distribution functions
  appear to be highly variable in space and time,
  pollutant and grid resolution .
• .

29
Studies in progress

• Texas 2000 (Houston) air toxics neighborhood
  scale CMAQ study using DA SM2-U in MM5
• Develop methodology for generating concentration
  distribution functions
• Develop and apply modeling approaches to
  determine the contribution of sub-grid
  variability in CMAQ at neighborhood scale grid
  resolution.
• Linkage of AQ models with exposure models

30
The End (Figure curtesy of Alan Huber)