1. Sulfate Attribution Methods
Gary Kleiman
NESCAUM/MANE-VU
MANE-VU SIP Planning Meeting
June 12, 2006

2. M-V Contribution Assessment
Empirical data analysis methods exploit observed relationships between emissions, meteorology and observations
Trajectory models/dispersion models use mimimal chemistry but link source and receptor locations via meteorology
Factor Analysis/Source Apportionment techniques determine optimal combinations of various sources which match observed pollution
Chemical Transport Models attempt to fully integrate knowledge about emissions, meteorology, atmospheric physics and chemistry to simulate ambient concentrations

3. Contributions to Regional Haze in the Northeast and Mid-Atlantic
United States
Available in draft form – comments through Friday
REMSAD with SO2 tagging
CALPUFF with 2 types of Meteorology
Trajectory methods (Incremental Probability and Cluster Weighted Probability)
Residence time probability x emissions
Emissions/distance
Plus conceptual model, monitoring strategy,
baseline conditions, emission inventory, and more…

4. Contribution Assessment Analysis Domain10 9 11 8
Black : Run 1
Red : Run 2
Blue : Run 3

5. Contributions by State
Brigantine

6. CALPUFF x 2

7. Q/d Methodology Q d
Impact = C x
Assumes the SO4 deposited at receptor is proportional to the ratio Q/d:
Q is the source emission rate,
d is the distance from source to receptor, and
C is a constant determined by wind sector d
Mass
Unit Time
Q =

8. Example: Lye Brook Receptor
Lye Brook Wilderness
o C=1.46E-4
o C=1.03E-4
o C=5.81E-5
o C=2.70E-5
90o
CALPUFF data used for linear regression
4 Wind Sectors fitted to R>0.90
Relative strength of impact varies by factor of 5 between sectors 0o
180o
Regression Analysis: x-axis is Q/d where Q is given, d is calculated
y-axis is Impact, given by CALPUFF model from Paul Washinsky
270o

9. Residence Time
Sulfur Emissions S
State
Source
Contribution = X

10. Methodology Residence Time grid and SO2 emissions grid do not overlap
Residence Times of Endpoints1 in Chesapeake Bay Region
Residence Time grid and SO2 emissions grid do not overlap
ArcView calculated product of an RT pt. and closest SO2 data pt.
1/4deg as cutoff distance
1Darker dots indicate larger residence times
= Residence Time
= SO2 Emission

11. Sulfate Contribution by RPO
ACAD-ME
BRIG-NJ
LYBR-VT
SHEN-VA

12. 2002 Average SO4 Contributions (by state)

13. Example: 2018 Contributions
Ratio of 2018/2002

14. Proximity-Based Clusters Best and Worst Days
Acadia
Mohawk Mountain
Lyebrook
Highest
Sulfate
Sulfate= 3.19
Bext= 64.16
PM =8.19
OC = 1.91
Sulfate= 5.10
Bext= 91.95
PM =11.78
OC = 3.19
Sulfate= 3.92
Bext= 68.67
PM =9.11
OC = 1.72
Lowest
Sulfate
Sulfate= 2.01
Bext= 41.43
PM =5.35
OC =1.50
Sulfate= 1.40
Bext= 30.83
PM =4.09
OC = 1.04
Sulfate= 1.30
Bext= 25.92
PM =3.62
OC = 0.90

15. Proximity-based Clusters Acadia 2000-2004 Trajectories
CT(1)= acad dat
Cluster 2
CT(2)= acad dat
Cluster 3
CT(3)= acad dat
Cluster 4
CT(4)= acad dat
Cluster 5
CT(5)= acad dat
Original Cluster
Original Cluster
Original Cluster
Original Cluster
Original Cluster
Original Cluster
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Sulfate= 2.45
Bext= 50.35
PM =6.54
OC = 1.57
Sulfate= 3.14
Bext= 59.88
PM =7.92
OC = 1.52
Sulfate= 1.79
Bext= 39.91
PM =5.42
OC =1.49
Sulfate= 1.40
Bext= 30.83
PM =4.09
OC = 1.04
Sulfate= 3.13
Bext= 57.88
PM =7.80
OC = 1.58

16. Proximity-based Clusters Acadia 2000-2004 Trajectories
CT(6)= acad dat
Cluster 7
CT(7)= acad dat
Cluster 8
CT(8)= acad EDAS.dat
Cluster 9
CT(9)= acad EDAS.dat
Cluster 10
CT(10)= acad dat
Original Cluster
Original Cluster
Original Cluster
Original Cluster
Original Cluster
Original Cluster
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Rotated Coordinate
Sulfate= 3.19
Bext= 64.16
PM =8.19
OC = 1.91
Sulfate= 1.70
Bext= 33.20
PM =4.25
OC = 0.81
Sulfate= 2.21
Bext= 44.48
PM =6.11
OC =1.52
Sulfate= 2.10
Bext= 44.05
PM =5.95
OC = 1.77
Sulfate= 1.41
Bext= 28.96
PM =3.78
OC = 0.89

17. Lyebrook Cluster Weighted Probability Aggregated by State

18. Comparison of Techniques (Lyebrook)
IP500
IP1000
CWP Norm (RotCoor CWP)
CWP Regular

19. Source Models and Receptor Models Agree!

20. Attribution/Consultation
Average
REMSAD
Q/d
CALPUFF (VT)
CALPUFF (MD)
E x RTP
CANADA MA PA OH ME NY IN IL WV MI NH
CENRAP MD KY GA NC VA NJ WI DE TN SC AL CT VT RI MS DC
Acadia
Who do you want
to consult?

21. Acadia Uncertainty?

22. Attribution/Consultation
Average
REMSAD
Q/d
CALPUFF (VT)
CALPUFF (MD)
E x RTP PA OH WV
CANADA MD VA NY NC NJ IN KY DE IL GA MI MA
CENRAP TN SC AL WI CT NH DC ME MS RI VT
Brigantine
Who do you want
to consult?

23. Attribution/Consultation
Average
REMSAD
Q/d
CALPUFF (VT)
CALPUFF (MD)
E x RTP PA
CANADA OH NY IN WV MI
CENRAP KY MD IL MA GA VA NC WI NJ NH TN VT AL SC DE CT MS ME RI DC
Lye Brook
Who do you want
to consult?

24. Lye Brook Uncertainty?

25. Attribution/Consultation
Average
REMSAD
Q/d
CALPUFF (VT)
E x RTP
CANADA OH PA NY IN NH MA MI WV KY IL
CENRAP ME NJ WI NC GA VA MD TN AL VT SC CT DE RI MS DC
Great Gulf
Who do you want
to consult?

26. Attribution/Consultation
Average
REMSAD
Q/d
CALPUFF (VT)
E x RTP
CANADA ME PA OH MA NY IN IL MI WV NH GA KY MD NC
CENRAP VA WI NJ TN AL DE SC VT CT RI MS DC
Moosehorn/
Roosevelt-
Campobello
Who do you want
to consult?

27. Moosehorn Uncertainty?

28. Contribution by Month
Acadia NP
Lye Brook Wilderness

29. Contribution by Month
Brigantine NWR
Shenandoah NP

30. Contribution Day-by-Day
August 9-16, 2002
3 sites, 3 influences