The Transport and Deposition of Dioxinto Lake Michigan: A Case Study

Dr. Mark CohenNOAA Air Resources Laboratory

SSMC3, R/ARL, Room 33161315 East West Highway, Silver Spring MD, 20910

ph: 301-713-0295 x122; fax: 301-713-0119mark.cohen@noaa.gov

Presentation at:

“Using Models to DevelopAir Toxics Reduction Strategies:Lake Michigan as a Test Case”

Milwaukee, Wisconsin, Nov 8-9, 2000

The views expressed in this presentation are those of the author, and do notnecessarily represent the views of the Air Resources Laboratory, NOAA,

the Department of Commerce, or the United States Government.

The Transport and Deposition of Dioxinto Lake Michigan: A Case Study

Presentation Outline

ØØ Policy Making Contextand the potential role of models

ÙÙ regarding atmospheric deposition, What Do We Need to Know?

and How Well Do We Need to Know It?

ÚÚ Atmospheric Deposition of Dioxin to Lake Michigan

ÛÛ Uncertainty Analysis

ÜÜ Conclusions and Recommendations

POLICY MAKING CONTEXT

1. EFFECTS?Harmful effects on wildlife, public health?[what is the exposure? consequences of thisexposure?]

2. CAUSES ?

What is the relative contribution ofdifferent loadings pathways contributing tothe harmful effects?

And, for any given significant loadingpathway, what are the relativecontributions of different sources?

(THIS TALK: ATMOSPHERIC PATHWAY DETAILS)

3. COSTS ?

What are the technical options involved inreducing or eliminating the contributionsfrom major sources?

What are the costs to implement theseoptions?

-- Decision Making: Need Info in All Three AreasOr, it doesn’t necessarily do you much good to have precise informationin one area, if one or more of the other areas remain very uncertain

-- How much do we know regarding these questionsfor dioxin in Lake Michigan?

THE ROLE AND POTENTIAL VALUE OF MODELS

ëë MODELS are mathematical/conceptual descriptions of real-world phenomena

è Necessarily a simplification; the real world is very complicated

è Key processes must be sufficiently characterized

ëë MODELS are POTENTIALLY VALUABLE for:

è Examining different large-scale scenarios that cannot be easily testedin the real world (e.g., different emissions reduction scenarios).

è Interpreting measurements

è Filling in spatial and temporal gaps between measurements

ëë MODELS are a TEST of our KNOWLEDGE:

è Attempts to synthesize everything important about a given system.

è If a model fails, we don’t understand enough about the system.

ëë MODELS are USED IN developing approximate answers inALL THREE fundamental policy information areas

0 - 0.10.1 - 2525 - 5050 - 7575 - 100100 - 300300 - 50005000 - 335000

Areal Density of Dioxin Emissions (µgrams TEQ/km²-yr)

No Data Available

N

500 0 500 1000Miles

500 0 500 1000Kilometers

Figure 1: Total Dioxin Emissions for 1996

No Data Available

Contribution to Deposition(µgrams TEQ/km²-yr)

Mid-Range Estimate of the Contribution to 1996 Atmospheric Deposition of Dioxin to Lake Michigan (µgrams TEQ/km²-yr)

500 0 500 1000 Miles

500 0 500 1000 Kilometers

N

0 - 0.010.01 - 0.10.1 - 11 - 1010 - 100100 - 10001000 - 7000

Estimates of the Percent of Lake Michigan Dioxin Loadings

Attributable to the Atmospheric Deposition Pathway

Study Fraction of Current Loadings Contributed

Through the Atmospheric Pathway

Cohen et al. PCDD/F TEQ: 50-100

(central estimate ~ 88)

Pearson et al. PCDD: 50-100

PCDF: 5-35

ë Cohen, M., et al., 1995. Quantitative Estimation of the Entry of Dioxins, Furans, andHexachlorobenzene into the Great Lakes from Airborne and Waterborne Sources. Flushing, NY: CBNS, Queens College. Final Report to the Joyce Foundation.

ë Pearson, R.F., D.L. Swackhamer, S.J. Eisenreich, and D.T. Long (1998). “AtmosphericInputs of Polychlorinated Dibenzo-p-dioxins and Dibenzofurans to the Great Lakes:Compositional Comparison of PCDD and PCDF in Sediments.” J. Great Lakes Research24(1): 65-82.

What Do We Need to Know, and.How Well Do We Need to Know It?

For most policy considerations, the exact contributions of individual sourcesdo not need to be known.

It is generally sufficient to know about:

ëë The geographical extent of the problem

C relative impact of local, regional, national, continental,and/or global sources

C don’t need exact answers, e.g., if 70% or 50% of thecontributing air sources arise from within 100 km of theLake – the policy response will be similar in either case.

C Only if the estimates are grossly incorrect will policydeliberations be seriously affected.

ëë Which source categories are the most significantcontributors?

C don’t need exact answers; e.g., it does not matter that muchwhether municipal solid waste incinerators contribute 20% or40% to the deposition – the policy response will likely bevery similar.

C Again, the estimates will be of little or no use only if they areextremely inaccurate.

Figure 10. Percent of Total Emissions or Total Deposition of Dioxin (1996)Arising from Within Different Distance Ranges From Each of the Great Lakes

0 - 100 - 200 - 400 - 700 - 1000 - 1500 - 2000 - 2500 - > 3500100 200 400 700 1000 1500 2000 2500 3500

Distance Range from Lake (km)

Emissions Deposition

0

10

20

30

40

50

Lake Michigan

0

10

20

30

40

50

Lake Erie

0

10

20

30

40

50

Lake Superior

0

10

20

30

40

50

Lake Huron

0

10

20

30

40

50

Lake Ontario

Figure 12. Contriibution of Different Source Sectors to Atmospheric Deposition of Dioxin( pg TEQ deposition / km2 ) / ( person - year )

(Each country's annual deposition flux contribution amount normalized by their total population)

United States Canada

"incin" = waste incineration; "metals" = metallurgical processing; "fuel" = fuel combustion

incin metals fuel

Emissions Sector

0.00.10.20.30.40.50.60.7

Lake Erie

incin metals fuel

Emissions Sector

0.00.10.20.30.40.50.60.7

Lake Michigan

incin metals fuel

Emissions Sector

0.0

0.1

0.2

0.3

0.4

0.5

Lake Superior

incin metals fuel

Emissions Sector

0.00.10.20.30.40.50.6

Lake Huron

incin metals fuel

Emissions Sector

0.0

0.2

0.4

0.6

0.8

Lake Ontario

incin metals fuel

Emissions Sector

0.00.10.20.30.40.50.60.7

Great Lakes Average

Major Sources of Uncertaintyin Atmospheric Dioxin Modeling

ëë Emissions Inventory

ëë Meteorological Data Used as Input to the Model

ëë Atmospheric Dispersion Simulation

ëë Atmospheric Fate Processes

ëë Vapor/Particle Partitioningëë Chemical Transformationsëë Wet Depositionëë Dry Deposition

Other than uncertainties in the emissions inventory, the dry deposition modeling

methodology is probably the most important factor influencing the results...

1 2 3 4 5 6

Dry Deposition Algorithm

0.0

0.5

1.0

1.5

2.0

2.5

to D

epo

siti

on

wit

h C

han

ged

Par

amet

er V

alu

e

Rat

io o

f D

epo

siti

on

wit

h D

efau

lt P

aram

eter

Val

ue

Dry Deposition Algorithms: 1 - HYSPLIT_4 default; 2 - vpfin5f; 3 - Slinn & Slinn, 1980, no RH correction, for water, vpfin5f for other processes4 - EPA "best method" (ADOM-II) for particles, vpfin5f for gases; 5 - Williams (1982) for particles over water, vpfin5f for other processes;6 - same as 3, but with RH correction, i.e., Slinn and Slinn (1980) with RH correction for particles over water, vpfin5f for other processes

Sensitivity Analysis for Model Estimated Deposition to Lake MichiganFactor Varied: Dry Deposition Algorithm (note: Default = #6)

Continuous Year Long Source of 2,3,7,8-TCDD at Center of Modeling Domain

What are the most important factors in the simulation uncertainty

(other than the dry deposition methodology and the emissions inventory)?

wetr_

tcdd_c

ente

r

wetr_

tcdd_m

ilw

wetr_

tcdd_s

f

times

tep

phot_tc

df_sf

partn

um

phot_tc

df_ce

nter

emit

freq

psd_t

cdd_m

ilw

phot_tc

dd_sf

phot_tc

dd_cen

ter

psd_t

cdd_s

f

phot_tc

df_m

w

phot_tc

dd_mw

psd_t

cdd_c

ente

r

phot_ocd

d_sf

height_

tcdd_c

ente

r

phot_ocd

d_cen

ter

height_

tcdd_m

ilw

phot_ocd

d_mw

Sensitivity Analysis Series

0%

5%

10%

15%

20%

Var

iatio

n in

Par

amet

er

Rat

io o

f Var

iatio

n in

Res

ults

to

Source Locations: "center" = center of modeling domain (lat/long = 40,95); "sf" = San Francisco; "milw" = Milwaukee.Parameters: "wetr" = in-cloud particle wet deposition; "phot" = photolysis rate; "psd" = ambient particle conc. (affects V/P partitioning);

"height" = height of source; "timestep" = model time step; "partnum" = # of puffs simulated; "emitfreq" = puff emissions frequency.

Summary of Sensitivity Analyses (for Factors Other than Dry Deposition Methodology or Emissions)

0 - 100100 - 200

200 - 400400 - 700

700 - 10001000 - 1500

1500 - 20002000 - 2500

2500 - 3500

Distance Range from Lake (km)

0%

10%

20%

30%

40%

50%

60%

Per

cen

t o

f T

ota

l Dep

osi

tio

n

drydep = 1drydep = 2

drydep = 4drydep = 5

drydep = 6no photolysis

particle wet deposition x 4

Dry Deposition Algorithms: 1 - HYSPLIT_4 default; 2 - vpfin5f; 3 - Slinn & Slinn, 1980, no RH correction, for water, vpfin5f for other processes4 - EPA "best method" (ADOM-II) for particles, vpfin5f for gases; 5 - Williams (1982) for particles over water, vpfin5f for other processes;6 - same as 3, but with RH correction, i.e., Slinn and Slinn (1980) with RH correction for particles over water, vpfin5f for other processes

Effect of Different Deposition and Fate Methodologieson Percent Dioxin Deposition Contribution

to Lake Michigan From Different Distance Ranges

Mu

nicip

al Waste In

cineratio

n

Iron

Sin

tering

Med

ical Waste In

cineratio

n

Cem

ent K

ilns B

urn

ing

Haz W

aste

Backyard

Waste B

urn

ing

Seco

nd

ary Co

pp

er Sm

elting

Seco

nd

ary Alu

min

um

Sm

elters

Resid

ential W

oo

d C

om

bu

stion

Mo

bile S

ou

rces

Utility C

oal C

om

bu

stion

Electric A

rc Fu

rnaces

Hazard

ou

s Waste In

cineratio

n

Sew

age S

lud

ge In

cineratio

n

Cem

ent K

ilns N

ot B

urn

ing

Haz W

as

Pu

lp-P

aper: H

og

Fu

el / Slu

dg

e Co

Resid

ential F

uel C

om

bu

stion

Ind

ustrial W

oo

d C

om

bu

stion

Grey Iro

n F

ou

nd

ries

Pu

lp-P

aper: K

raft Black L

iqu

or

Resid

ential O

il Co

mb

ustio

n

Co

mm

ercial Fu

el Co

mb

ustio

n

Seco

nd

ary Lead

Sm

elting

Base M

etal Sm

elting

Ind

ustrial F

uel C

om

bu

stion

Ag

ricultu

ral Fu

el Co

mb

ustio

0%10%20%30%40%

Pe

rce

nt

of

To

tal

drydep = 1drydep = 2

drydep = 4drydep = 5

drydep = 6photolysis = 0

particle wet dep x 4

Dry Deposition Algorithms: 1 - HYSPLIT_4 default; 2 - vpfin5f; 3 - Slinn & Slinn, 1980, no RH correction, for water, vpfin5f for other;4 - EPA "best method" (ADOM-II) for particles, vpfin5f for gases; 5 - Williams (1982) for particles over water, vpfin5f for other;6 - same as 3, but with RH correction, i.e., Slinn and Slinn (1980) with RH correction for particles over water, vpfin5f for other.

Results from Different Sensitivity Simulations (using 28 standard source locations)on Emissions Sector Contributions to Lake Michigan Atmospheric Dioxin Deposition (1996)

Mu

nicip

al Waste In

cin.

Iron

Sin

tering

Med

ical Waste In

cin

Cem

ent K

ilns B

urn

ing

HW

Backyard

Waste B

urn

ing

Seco

nd

ary Co

pp

er Sm

elting

Seco

nd

ary Alu

min

um

Sm

elti

Resid

ential W

oo

d B

urn

ing

Mo

bile S

ou

rces

Utility C

oal C

om

bu

stion

Electric A

rc Fu

rnaces

Hazard

ou

s Waste In

cin.

0%

10%

20%

30%

40%

Per

cen

t o

f T

ota

l

drydep = 1

drydep = 2

drydep = 4

drydep = 5

drydep = 6

photolysis = 0

particle wet dep x 4

Dry Deposition Algorithms: 1 - HYSPLIT_4 default; 2 - vpfin5f; 3 - Slinn & Slinn, 1980, no RH correction, for water, vpfin5f for other;4 - EPA "best method" (ADOM-II) for particles, vpfin5f for gases; 5 - Williams (1982) for particles over water, vpfin5f for other;6 - same as 3, but with RH correction, i.e., Slinn and Slinn (1980) with RH correction for particles over water, vpfin5f for other.

Results from Different Sensitivity Simulations (using 28 standard source locations)on Emissions Sector Contributions to Lake Michigan Atmospheric Dioxin Deposition

(for top 12 contributing sectors, 1996)

OntarioErie

SuperiorHuron

Michigan

1

10

100

1000

Dep

osit

ion

(g T

EQ

/yea

r)

central estimate range due to emissions uncertainty fate variations

Figure 3. Ranges in Estimated Deposition Arising fromUncertainties in Emisisons and Fate Simulation

What could we do to improve theaccuracy of atmospheric loadingestimates?

ë More information on non-atmospheric loadingpathways needs to be collected in order to moreaccurately place the atmospheric contributions intheir proper context.

ë Ambient monitoring for dioxin must be increasedin the Great Lakes region. This will allow modelevaluation and independent semi-empiricalestimates of atmospheric deposition to be made.

ë Additional efforts to improve the accuracy ofemissions inventories – including timely updates– must be made. Timely (e.g., annual) updates forat least the largest sources in the inventory wouldbe extremely helpful, because often, these largestsources tend to drive the analysis. If they can bebetter characterized, the accuracy of the overallanalysis can be greatly (and relatively easily)improved.