Identification of a Molecular Activator for Insulin Receptor with Potent
ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR...
Transcript of ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR...
![Page 1: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/1.jpg)
ADVANCED RECEPTOR MODELINGFOR SOURCE IDENTIFICATION AND
APPORTIONMENT
Philip K. Hopke
Dept. of Chemical EngineeringClarkson University
![Page 2: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/2.jpg)
OUTLINE
l Receptor Modelsl Factor Analysisl Positive Matrix Factorization
Theory and Application
l Potential Source Contribution Function– Theory and Application
l Advanced Factor Analysis Modelsl Summary
![Page 3: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/3.jpg)
BUT First!
l We need to answer your first burningquestion:
l Where is Potsdam, NY?
![Page 4: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/4.jpg)
BUT First!
![Page 5: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/5.jpg)
Receptor Modeling
Receptor models are focused on the behavior ofthe ambient environment at the point of impactas opposed to the source-oriented models thatfocus on the transport, dilution, andtransformations that begin at the source andfollow the pollutants to the sampling or receptorsite.
![Page 6: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/6.jpg)
![Page 7: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/7.jpg)
Receptor Modeling
PRINCIPLE OF AEROSOL MASS BALANCE
The fundamental principle of receptor modelingis that mass conservation can be assumed anda mass balance analysis can be used toidentify and apportion sources of airborneparticulate matter in the atmosphere.
![Page 8: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/8.jpg)
Mass Balance
A mass balance equation can be written toaccount for all m chemical species in the nsamples as contributions from p independentsources
p
ij ik kjk 1
x g f=
= ∑Where i = 1,…, n samples, j = 1,…, m species
and k = 1,…, p sources
![Page 9: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/9.jpg)
MASS BALANCE
l SOURCES KNOWN• Chemical Mass Balance
• Multivariate Calibration Methodsl Partial-Least Squares
l Artificial Neural Networks
l Simulated Annealing
l Genetic Algorithm
![Page 10: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/10.jpg)
Mass Balance
A mass balance equation can be written toaccount for all m chemical species in the nsamples as contributions from p independentsources
p
ij ik kjk 1
x g f=
= ∑Where i = 1,…, n samples, j = 1,…, m species
and k = 1,…, p sources
![Page 11: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/11.jpg)
MASS BALANCE
• SOURCES UNKNOWN• Factor Analysis
• Principal Components Analysis
• Absolute Principal Components Analysis
• SAFER/UNMIX
• Positive Matrix Factorization
![Page 12: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/12.jpg)
Mass Balance
p
ij ik kjk 1
x g f=
= ∑
Where i = 1,…, n samples, j = 1,…, m speciesand k = 1,…, p sources
In a factor analysis, all we know are the ambientconcentrations for a series of samples.
![Page 13: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/13.jpg)
Mass Balance
The equation can be rewritten in matrix form as
X GF=The question is then what information is available
a priori in order to solve these equations. Ingeneral, good source profiles (F matrix) are notavailable and need to be derived from the data.
![Page 14: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/14.jpg)
![Page 15: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/15.jpg)
![Page 16: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/16.jpg)
Factor Analysis
Most factor analysis has been based on aneigenvector analysis. In an eigenvectoranalysis, it can be shown [Lawson and Hanson,1974; Malinowski, 1991] that the equationestimates X in the least-squares sense that itgives the lowest possible value for
( )pn m n m2 2
ij ij ik kji 1 j 1 i 1 j 1 k 1
e (x g f )= = = = =
= −∑∑ ∑∑ ∑
![Page 17: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/17.jpg)
FACTOR ANALYSIS
![Page 18: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/18.jpg)
FACTOR ANALYSIS
![Page 19: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/19.jpg)
Factor Analysis
The problem can be solved, but it does notproduce a unique solution. It is possible toinclude a transformation into the equation.
X=GTT-1F
where T is one of the potential infinity oftransformation matrices. This transformation iscalled a rotation and is generally included inorder to produce factors that appear to becloser to physically real source profiles.
![Page 20: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/20.jpg)
Factor Analysis
![Page 21: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/21.jpg)
Positive Matrix Factorization
l Least squares approach to solving the factoranalysis problem
l Individual data point weights
l Imposition of natural and other constraints, and
l Flexibility to build more complicated models
![Page 22: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/22.jpg)
Positive Matrix Factorization
l The Objective Function, Q, is defined by2p
ij ik kjn mk 1
i 1 j 1 ij
x g fQ
σ=
= =
− =
∑∑∑
where s ij is an estimate of the uncertaintyin xij
![Page 23: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/23.jpg)
Positive Matrix Factorization
l The problem can be solved by several differentprograms developed by Pentti Paateroincluding PMF2 and ME. In PMF2, theconstraints are imposed as soft constraintsthrough a penalty function. In ME, hardconstraints are imposed and the model can bemuch more complex.
![Page 24: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/24.jpg)
Positive Matrix Factorization
l As an illustration of PMF2, the analysis of datafrom Underhill, VT will be presented.
l Aerosol sampling in Underhill, Vermont (44.53N, -72.86 W, 400 m elevation), was conductedbetween September 1988 and June 1995
![Page 25: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/25.jpg)
Positive Matrix Factorization
Filters were analyzed for mass (gravimetric), lightabsorption (Babs - by laser integrating plate -LIPM), elemental hydrogen (by proton elasticscattering analysis - PESA), and elements fromNa through Pb (initially by proton induced x-rayemission - PIXE, and starting June, 1992, by acombination of PIXE and x-ray fluorescence -XRF).
![Page 26: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/26.jpg)
IMPROVE Data Analysis
l Sources of Fine Particle Composition in theNortheastern US, X.H. Song, A.V. Polissar,and P.K. Hopke, Atmospheric Environ.35:5277-5286 (2001).
l Atmospheric Aerosol over Vermont: ChemicalComposition and Sources, A.V. Polissar, P.K.Hopke, and R.L. Poirot, Environ. Sci. Technol.35: 4604-4621 (2001).
![Page 27: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/27.jpg)
[S] (µg m -3)0 1 2 3 4 5 6
[BC
] (
µg m
-3)
0
1
2 r2 =0.5
4
r2 =0.75
[BC] (µg m-3)0 1 2
[FM
] (µ
g m
-3)
0
10
20
30
40
50
r2 =0.7
6
r2 =0
.81
[S] (µg m -3)0 1 2 3 4 5 6
[FM
] (µ
g m
-3)
0
10
20
30
40
50
[S] (µg m-3)0 1 2 3 4 5 6
[Se]
(ng
m-3
)0
1
2
3r2 =0.6
8
r2 =0.60
r2 =0.3
5
A B
C D
1 2
3
4
2
4
r2 =0.8
2
CoalOilWoodIncineration
Gasoline "Winter""Winter" Regr."Summer""Summer" Regr.PMF factors
![Page 28: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/28.jpg)
103
105 Woodsmoke
103
105 MW W-coal
103
105 East coast oil
103
105 MW S-coal
103
105 Zn-Pb
103
105 Na-S
103
105 Can. Mn
103
105 Soil
Con
cent
ratio
n (p
pm)
103
105 Cu smelting
103
105 Can. smelting
As Br Cl Cu H Mg
Mo
Ni Pb S Si Ti ZnAl BC Ca Cr Fe K Mn
Na
P Rb Se Sr V Zr
103
105 Salt11
10
9
8
7
6
5
4
3
2
1
Al BC Ca Cr Fe K Mn Na P Rb Se Sr V ZrAs Br Cl Cu H Mg Mo Ni Pb S Si Ti Zn
![Page 29: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/29.jpg)
![Page 30: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/30.jpg)
![Page 31: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/31.jpg)
![Page 32: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/32.jpg)
![Page 33: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/33.jpg)
![Page 34: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/34.jpg)
Date
Coa
l Con
trib
utio
n (n
g m
-3)
10
100
1000
10000
100000 WinterSummerTotal Coal
1988 1989 1990 1991 1992 1993 1994 1995
![Page 35: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/35.jpg)
![Page 36: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/36.jpg)
![Page 37: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/37.jpg)
![Page 38: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/38.jpg)
![Page 39: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/39.jpg)
![Page 40: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/40.jpg)
![Page 41: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/41.jpg)
![Page 42: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/42.jpg)
![Page 43: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/43.jpg)
Potential Source Contribution Function
The potential source contribution function can bedescribed as follows: if a trajectory endpoint lies at acell of address (i, j), the trajectory is assumed tocollect material emitted in the cell. Once aerosol isincorporated into the air parcel, it can be transportedalong the trajectory to the receptor site. The objectiveis to develop a probability field suggesting likelysource locations of the material that results in highmeasured values at the receptor site.
![Page 44: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/44.jpg)
Potential Source Contribution Function
The number of endpoints falling into a single gridcell is ni,j. Some of these trajectory endpointsare associated with sampling intervals whenthe greater contribution of a particular source isequal to that of the 60th percentile value in thedistribution of that species. The number ofendpoints is mi,j.
![Page 45: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/45.jpg)
Potential Source Contribution Function
The potential source contribution function (PSCF)is then defined as
ijij
ij
mPSCF
n=
![Page 46: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/46.jpg)
Potential Source Contribution Function
The potential source contribution function can beinterpreted as a conditional probabilitydescribing the spatial distribution of probablegeographical source locations inferred by usingtrajectories arriving at the sampling site. Cellsrelated to the high values of potential sourcecontribution function are the potential sourceareas.
![Page 47: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/47.jpg)
Potential Source Contribution Function
The trajectory data consist of 12 trajectories forthe 10 random air parcels calculations for eachday of the 853 days of sampling. There are 37endpoints in each trajectory. Therefore, thetotal number of the endpoints was12×10×853×37 or about 3.8 x 106. The gridcovers most of North America with 2,966 cellsof 1 degree by 1 degree latitude and longitudeso that in average there are 3.8 x 106/2966 orabout 1280 endpoints per cell.
![Page 48: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/48.jpg)
Woodsmoke
![Page 49: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/49.jpg)
Winter Midwestern Coal
![Page 50: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/50.jpg)
Summer Midwestern Coal
![Page 51: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/51.jpg)
East Coast Oil
![Page 52: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/52.jpg)
Canadian Manganese
![Page 53: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/53.jpg)
Copper Smelter
![Page 54: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/54.jpg)
Canadian Ni Smelters
![Page 55: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/55.jpg)
NEW DIRECTIONS
l The least-squares factor analysis approachpermits more complex models to bedeveloped.
l Such models permit greater resolution ofsources and provide indication of sourcelocation relative to receptor site.
l Can be modified to include reactivity.
![Page 56: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/56.jpg)
Advanced Models
l The concentration of particles from a sourcedepends on the transport to the receptor siteincluding wind speed, direction, mixing height,and atmospheric chemistry.
l Strength of the emissions may depend onweekday/weekend, time of year, etc.
![Page 57: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/57.jpg)
Advanced Models
l How to incorporate additional information likewind speed, direction, etc. into the analysis.This is what we are terming an advanced factoranalysis model.
![Page 58: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/58.jpg)
Advanced Models
l To present the expanded factor analysisapproach, the model is described from theviewpoint of one source, denoted by p.
l In reality, there are several sources and theobserved concentrations are sums ofcontributions due to all sources, p=1,..., P.
![Page 59: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/59.jpg)
Advanced Models
l In the customary bilinear analysis, thecontribution rijp of source p on day i toconcentration of chemical species j isrepresented by the product gipfjp, where gip
corresponds to the strength of source p on dayi, and fjp corresponds to the concentration ofcompound j in the emission signature of sourcep.
![Page 60: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/60.jpg)
Advanced Models
l In the expanded PMF analysis, the bilinearequation is augmented by another morecomplicated set of equations that containmodeling information.
![Page 61: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/61.jpg)
Advanced Models
l The contribution rijp of source p isrepresented by the following expression:
l The known values äi and vi indicate winddirection and wind speed on day i. Thesymbols D and V represent matrices,consisting of unknown values to be estimatedduring the fitting process. Their columnsnumbered p correspond to source number p.
( , ) ( , )= = D Vijp ip jp i i jpr m f p v p fδ
![Page 62: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/62.jpg)
Advanced Models
l The index value äi for day i is typically obtainedby dividing the average wind direction of day i(in degrees) by ten and rounding to nearestinteger.
l The values vi are obtained from a chosenclassification of wind speeds.
![Page 63: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/63.jpg)
Advanced Modelsl In component form, the equations of the
model are
1
1
1
'
( , ) ( , ) '
=
=
=
= +
= +
= +
∑
∑
∑D V
P
ij ip jp ijp
P
ij ip jp ijp
P
i i jp ijp
x g f e
x m f e
p v p f eδ
![Page 64: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/64.jpg)
Advanced Models
l The notation mip does not indicate a factorelement to be determined, such as gip, but theexpression defined by the physical model inquestion.
l In different physical models, mip willcorrespond to different expressions.Because the variability of mip is restricted bythe model, the second set of equations willproduce a significantly poorer fit to the datathan the first set of equations.
![Page 65: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/65.jpg)
Advanced Models
l Thus, the error estimates connected with thesecond set of equations must be (much)larger than the error estimates connectedwith the first set of equations.
l The task of solving this expanded PMF modelmeans that values of the unknown factormatrices G, F, D, and V are to be determinedso that the model fits the data as well aspossible.
![Page 66: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/66.jpg)
Advanced Models
l In other words, the sum-of-squares value, Q,defined by
is minimized with respect to the matrices G, F,D, and V, while the residuals eij and e’ij aredetermined by the model equations.
2 2
1 1 1 1
( / ) ( ' / ' )= = = =
= +∑∑ ∑∑I J I J
ij ij ij iji j i j
Q e eσ σ
![Page 67: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/67.jpg)
Advanced Models
l Since there are other sources of variation suchas weekend/weekday source activity patternsor seasonal differences in emission rates or inatmospheric chemistry, additional factors areincluded in the model.
![Page 68: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/68.jpg)
Advanced Models
l Wind direction, wind speed, time of year, andweekend/weekday were used. In this case,24 one-hour average values are available forwind speed and direction.
l Time of year will be aggregated into six two-month periods or seasons, indicated for eachday i by the index variables ói.
![Page 69: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/69.jpg)
Advanced Modelsl The non-linear dependencies are now defined by:
l where D(äih, p) is an element with the index for thewind direction during hour h of day i for the pthsource, V(í ih, p) is the element with the index for thewind speed during hour h of day i for the pth source,W(ù i, p) is the element with the index correspondingto day i for the weekday/weekend factor for the pthsource, and S(ó i, p) is the element with the indexcorresponding to the time-of-year classification of dayi for the pth source.
24
1
( , ) ( , ) ( , ) ( , )=
= ∑W S D Vip i i ih ihh
m p p p v pω σ δ
![Page 70: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/70.jpg)
Advanced Models
l The expanded model consists of the basicbilinear equations plus the set of physicalmodel equations:
1=
= +∑P
ij ip jp ijp
x g f e
1
24
1 1
'
( , ) ( , ) ( , ) ( , )
'
=
= =
= +
=
+
∑
∑ ∑W S D V
P
ij ip jp ijp
P
i i ih ih jpp h
ij
x m f e
p p p v p f
e
ω σ δ
![Page 71: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/71.jpg)
Palookaville
l Simulated data were produced by EPA for aFebruary 2000 workshop.
l 16 Sources distributed around a singlereceptor site.
l Data prepared through the use of the ISC-STmodel.
![Page 72: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/72.jpg)
Palookaville
l Fifteen of the 16 sources are foundl Good reproduction of the source profiles is
obtainedl Details are in
Utilizing Wind Direction and Wind Speed asIndependent Variables in Multilinear ReceptorModeling Studies, P. Paatero and P.K. Hopke,Chemom. Intell. Lab. Syst. 60: 25-41 (2002).
![Page 73: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/73.jpg)
Palookaville
![Page 74: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/74.jpg)
Palookaville
![Page 75: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/75.jpg)
Palookaville
![Page 76: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/76.jpg)
Palookaville
![Page 77: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/77.jpg)
Palookaville
![Page 78: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/78.jpg)
Palookaville
![Page 79: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/79.jpg)
Phoenix, AR
The location of sampling site is Phoenix, AZ(3847 West Earl Drive, Latitude: N33E48'46'',Longitude: W112E14'17'', Elevation: 1007 ft).Two different samplers were operated at thissite: a dual fine particle sequential sampler(DFPSS) (URG Corporation, Cary, NC) and adichotomous sampler (Andersen Instruments,Inc., Smyrna, GA). The DFPSS collected onlyfine particle samples while the dichotomoussampler provided samples of fine and coarseparticles.
![Page 80: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/80.jpg)
Phoenix, AR
Daily, integrated 24-hour samples werecollected on 37 millimeter (mm) diameterTeflon and quartz filter media for fine particlemass and species measurements using a dualfine particle sequential sampler (DFPSS). Thesamples were collected during the time periodfrom March 1995 through June 1998. A total of981 samples were finally obtained.
![Page 81: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/81.jpg)
Phoenix, AR
Energy dispersive X-ray spectrometers were used toproduce the chemical elemental concentration data.The quartz filters collected with the DFPSS wereanalyzed by Sunset Laboratory using the TOTtechnique. This technique measured both organiccarbon (OC) and elemental carbon (EC). Each samplewas characterized by the measured concentrations ofthe following 46 chemical elements: Na, Mg, Al, Si, P,S, Cl, K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga,Ge, As, Se, Br, Rb, Sr, Y, Zr, Mo, Rh, Pd, Ag, Cd, Sn,Sb,Te, I, Cs, Ba, La, W, Au, Hg, Pb, OC, and EC.
![Page 82: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/82.jpg)
Conventional PMF Results
We extracted 8 factors including two biomassburning factors, motor vehicles, diesel vehicles,sea salt, copper smelters, soil, and asecondary aerosol tentatively identified ascoal-fired power plants. These results werepublished in Ramadan et al., J. Air WasteManage. Assoc. 50:1308-1320 (2000).
![Page 83: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/83.jpg)
Expanded PMF Model
In this case, we have attempted to solve theproblem with only the model equations and notsimultaneously with the normal CMB modelequations.
![Page 84: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/84.jpg)
Expanded PMF Model
The model is then
1
'=
= +∑P
ij ip jp ijp
x m f e
![Page 85: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/85.jpg)
Expanded PMF Model
Wind direction, wind speed, time of year, time-of-day,inlet change, and weekend/weekday were used. Inthis case, 24 one-hour average values are available forwind speed and direction.
Time of year will be aggregated into six two-monthperiods or seasons, indicated for each day i by theindex variables Fi. (The Greek letter F is used for twopurposes: Fij indicates the error estimates of datavalues, while Fi indicates the season number for day i.).
For the values i=1 to i=60, Fi=1, meaning that Januaryand February belong to the first season. For thevalues i=61 to i=121, Fi=2, and so on.
![Page 86: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/86.jpg)
Expanded PMF Model
The non-linear dependencies are now definedby:
ip i i i
24
ih ih ihh 1
m I( ,p) ( ,p) ( ,p)
( ,p) ( ,p)T( ,p)
ι ω σ
δ υ τ=
=
∑
W S
D V
![Page 87: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/87.jpg)
Expanded PMF Model
where D(*ih, p) is an element with the index forthe wind direction during hour h of day i for thepth source, V(<ih, p) is the element with theindex for the wind speed during hour h of day ifor the pth source, T(JJih,p) is the element withthe index for time of day,
![Page 88: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/88.jpg)
Expanded PMF Model
W(Ti, p) is the element with the indexcorresponding to day i for theweekday/weekend factor for the pth source,I(4I,p) is the element with the indexcorresponding to before or after the change inthe inlet, and S(Fi, p) is the element with theindex corresponding to the time-of-yearclassification of day i for the pth source.
![Page 89: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/89.jpg)
Expanded PMF Model
The expanded model permitted the extractions ofeleven factors including sources identified asmotor vehicles, diesel vehicles, wood smoke,soil, sulfur, 2 copper smelters, marine, metals,fireworks, and unmeasured mass
![Page 90: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/90.jpg)
NAMGALSI P S CLK CASCTI V CRM
NFECONI CUZNGAGEAS SEBRRBSRY ZRM
ORHPDAGCDSNSBTEI CSBALAW AUHGTL PBOCECTC
0.0010.010.1
1
0.0010.010.1
1
0.0010.010.1
1
0.0010.010.1
1
0.0010.010.1
1
0.0010.010.1
1
0.0010.010.1
1
0.0010.010.1
1
NAM
GAL SI P SCL KCASC TI VCRM
NFECONI
CUZNGAGEASSEBRRBSR YZR
MORHPD
AGCDSNSBTE ICSBALA WAUHGTLPBOCECTC
0.0010.01
0.11
Soil
Marine
Copper Smelter 1
Copper Smelter 2
Motor Vehicles
Wood
S, V
Diesel (Mn)
Fireworks
0.0010.010.1
1
0.0010.010.1
1 Unexplained Mass
Metals
![Page 91: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/91.jpg)
Soil
Mari
ne
Firewks
S+VOCEC
Cu1+Pb
Cu2+Cd
Mn
Woo
dM
ass
Meta
ls
Cor
rect
ion
Fact
or
0.0
0.5
1.0
1.5
2.0
2.5
3.0
inlet-c weekend
![Page 92: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/92.jpg)
Copper Smelters
0.00 0.04 0.08 0.120.00
0.04
0.08
0.12
0.000.040.080.120.00
0.04
0.08
0.12
0
330
300
270
240
210
180
150
120
90
60
30
Cu Smelter 1Cu Smelter 2
![Page 93: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/93.jpg)
Fireworks
0.00 0.04 0.08 0.120.00
0.04
0.08
0.12
0.000.040.080.120.00
0.04
0.08
0.12
0
330
300
270
240
210
180
150
120
90
60
30
![Page 94: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/94.jpg)
Motor Vehicles
0.00 0.05 0.100.00
0.05
0.10
0.000.050.100.00
0.05
0.10
0
330
300
270
240
210
180
150
120
90
60
30
![Page 95: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/95.jpg)
Diesel (Mn)
0.00 0.05 0.100.00
0.05
0.10
0.000.050.100.00
0.05
0.10
0
330
300
270
240
210
180
150
120
90
60
30
![Page 96: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/96.jpg)
![Page 97: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/97.jpg)
S-V
0.00 0.04 0.080.00
0.04
0.08
0.000.040.080.00
0.04
0.08
0
330
300
270
240
210
180
150
120
90
60
30
![Page 98: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/98.jpg)
.05-.20 .20-0.6 0.6-1.0 1.0-1.5 1.5-2.52.5-***
Win
d Sp
eed
Fact
or
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Motor VehiclesDiesel (Mn)
.05-.20 .20-0.6 0.6-1.0 1.0-1.5 1.5-2.52.5-***
Win
d Sp
eed
Fact
or
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Fireworks
.05-.20 .20-0.6 0.6-1.0 1.0-1.5 1.5-2.52.5-***
Win
d Sp
eed
Fact
or
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Copper Smelter 1Copper Smelter 2
.05-.20 .20-0.6 0.6-1.01.0-1.5 1.5-2.52.5-***
Win
d Sp
eed
Fact
or
0.0
0.5
1.0
1.5
2.0
Wood
![Page 99: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/99.jpg)
8-11 12-16 17-21 22-2 3-7
Tim
e-of
-Day
Fac
tor
0.0
0.5
1.0
1.5
2.0Motor VehiclesDiesel (Mn)
8-11 12-16 17-21 22-2 3-7
Tim
e-of
-Day
Fac
tor
0.0
0.5
1.0
1.5
8-11 12-16 17-21 22-2 3-7
Tim
e-of
-Day
Fac
tor
0.0
0.5
1.0
1.5
2.0
Copper Smelter 1Copper Smelter 2
Wood
8-11 12-16 17-21 22-2 3-7
Tim
e-of
-Day
Fac
tor
0.0
0.5
1.0
1.5
Unmeasured Mass
![Page 100: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/100.jpg)
Months-of-the-Year
Jan-
Feb
Mar
-Apr
May
-Jun
Jul-A
ug
Sep-O
ct
Nov-D
ec
Seas
onal
Cor
rect
ion
0.5
1.0
1.5
Soil Marine Firewks S+V OCEC Mass
Months-of-the-Year
Jan-
Feb
Mar
-Apr
May
-Jun
Jul-A
ug
Sep-O
ct
Nov-D
ec
Seas
onal
Cor
rect
ion
0.5
1.0
1.5
Cu1+Pb Cu2+Cd Mn Wood Metals
![Page 101: ADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION · PDF fileADVANCED RECEPTOR MODELING FOR SOURCE IDENTIFICATION AND APPORTIONMENT ... The trajectory data consist of 12 trajectories](https://reader033.fdocuments.in/reader033/viewer/2022052709/5a78a1c57f8b9a1f128ddc75/html5/thumbnails/101.jpg)
QUESTIONS?
And thanks for the invitation to visit!