Receptor Modeling for Air Resources Management

1
Receptor Modeling for Air Resources Management

• Arpa Wangkiat
• Department of Environmental EngineeringRangsit
  University
• presented at the Workshop on Source
  Apportionment for Particulate MatterBangkok,
  ThailandMarch 4, 2008

2
Objectives

• Identify types of receptor models
• Summarize prior uses of receptor models in air
  quality management
• Describe some of the properties that distinguish
  source contributions

3
What is receptor model ?
The Receptor Sampling site
Receptor model is the method to identify the
sources and the contribution rate of SPM based on
the data on chemical composition, particle size,
concentration variation, particle form, etc.
which are obtained at a certain point and a
certain period
Estimate the contribution of each type of
sources but not of individual release source
4
Receptor Model
Microscopic methods
Chemical Methods
optical
Enrichment Factor
Qualitative analysis
Time series analysis
S.E.M.
Special series analysis
Chemical Mass Balance
Quantitative analysis
Multivariate data analysis methods
5
Type of Receptor Model Microscopic methods
Morphological observation

• Focus on particle size, color, form, surface
• characteristics and optical nature
• advantage most effective to identify the source
  from trees tissue and pollen
• disadvantage take time, a large number of
  particles must be observed, costly, not sensitive
  enough to identify the org. particles and
  non-crystalline particles.

6
(No Transcript)
7
Type of Receptor Model Chemical methods
Enrichment Factors

• Enrichment Factors Efi
• Efi (Ci/Cn)ambient
• (Ci/Cn) background
• Cn normalising element/unique element
• of the background
• Ci the element whose enrichment is to be
  determined

8
Type of Receptor Model Chemical methods Time
series and spatial series analysis

• This methods are to assume the emission source
  from the time series correlation of particulate
  weight and chemical components, and from space
  distribution of chemical components and those at
  the emission sources

Simple method but not suitable for quantitative
analysis
9
Quantitative receptor Model
Multivariate model
Chemical mass balance (CMB) Model
- Tracer element - Linear programming -
Ordinary linear least-square - Effective
variance least-squares -Ridge regression
-Factor analysis-Target transformation FA-
Multiple linear regression-Extended Q-mode FA-
Principal component analysis
Henry et. al,1984
10
Receptor Model Types(Brook et al., 2003, Watson
et al., 2002)

• Chemical Mass Balance
• Enrichment Factor
• Spatial or Temporal Eigenvectors
• Multiple Linear Regression
• Neural Networks
• Edge Detection
• Aerosol Equilibrium
• Aerosol/Gas Evolution
• Back Trajectory

11
The First Receptor ModelWhat you can see or smell
12
Black Carbon Remains at Mesa Verde National
Park, Colorado, USA
Not all black carbon is from diesel and other
vehicular emissions Marker is a better term
than tracer Theres something of everything in
everything
13
Material balance is starting point(Mexico City,
Feb/Mar 1997, Chow et al., 2002)
14
Source profiles are nextCommonly measured
elements, ions, and carbon (Zielinska et al.,
1998)
15
Worldwide PM Source Contribution Estimates by
Chemical Mass Balance (Chow and Watson, 2002)
16
Organic source profiles better distinguish among
sources(lactones, hopanes, guaiacols, syringols,
steranes, and sterols)(Zielinska et al., 1998)
17
One atmosphere (gases and particles) also works
for receptor models (Gertler et al., 1996)
Light Duty Emission Rates
Heavy Duty Emission Rates
18
We should get more out of existing samples
(Thermally evolved carbon fractions, Watson et
al., 1994)
Diesel-fueled vehicles
Gasoline-fueled vehicles
19
Time resolution is really nice(Spikes indicate
local sources, Watson and Chow, 2001)
20
Source Contributions to Hourly VOC(Houston, TX,
1993, Lu, 1996)
21
Receptor models can get at secondaries under
certain conditions

• OC/EC enrichment factors used to estimate
  secondary OC contributions (Turpin and
  Huntzicker, 1991, Gray et al., 1986)
• Secondary organic marker end-products (Pandis,
  2001)
• Aerosol evolution to represent changes in
  profiles (Lewis and Stevens, 1985)
• 34S or 35S isotopes to follow sulfate changes
  (Forrest and Newmann, 1973, Hidy, 1987)
• Regional source profiles (Rahn and Lowenthal,
  1984, Eatough et al., 1997)
• Eigenvector-derived profiles (Poirot et al., 2001)

22
Markers for Biogenic SOA(Pandis, 2001)

• Pinic acid, pinonic acid, norpinic acid, and
  norpinonic acid are products of the oxidation of
  most monoterpenes
• There are some (apparently) unique tracers
• Hydropinonaldehydes for a-pinene
• Nopinone for b-pinene
• 3-caric acid for carene
• Sabinic acid for sabenene
• Several of these compounds measured in field
  studies in forests (usually a few nanograms per
  cubic meter sometimes as much as 0.1 µg m-3)

23
PMF and UNMIX profiles from Vermont IMPROVE
data(Poirot et al., 2001)
24
Emission reduction effectivenessLong-term trends
in SO2 emissions and SO4 levels (Malm et al.,
2002)
25
Receptor models can estimate the future in some
circumstances(Denver, CO, 1997, Watson et al.,
1998)
Effect of ammonia reductions on ammonium nitrate
particles
Effect of nitric acid reductions on ammonium
nitrate particles
26
Murphys Law of ReproducibilityIf
reproducibility is a problem, just use one
modelMohave Generating Station contributions to
Meadview sulfate (Pitchford et al., 1999)
27
Model discrepancies help to improve the
processPM2.5 Inventory/Receptor Model
Comparison, Denver, CO 1997(Watson et al.,
2002)
28
Receptor Model Needs

• Source properties that identify and quantify
  source contributions at a receptor (Daisey et
  al., 1986, Gordon et al., 1984)
• Better designed networks (Chow et al., 2002,
  Demerjian, 2000) with respect to
• Sampling locations
• Sampling periods
• Sample durations
• Particle sizes
• Precursor gases
• Chemical and physical components

29
Receptor Model Needs

• Emissions profiles (with cooling and dilution
  including marker species and gases, England et
  al., 2000)
• More convenient availability and documentation of
  source profile and ambient data (U.S. EPA, 1999)
• More evaluation, validation, and reconciliation
  of receptor and source modeling results (Javitz
  et al., 1988)