modeling the atmospheric transport and deposition of ... · modeling the atmospheric transport and...
TRANSCRIPT
Modeling the Atmospheric Transport and Deposition of Mercury to the Great Lakes
Dr. Mark Cohen NOAA Air Resources Laboratory (ARL)
College Park, MD, USA
NOAA Air Resources Laboratory Seminar February 7, 2013, NCWCP, College Park MD
2
Acknowledgements
Roland Draxler Glenn Rolph Barbara Stunder Ariel Stein Fantine Ngan
HYSPLIT Development Team at ARL:
Winston Luke Paul Kelley Steve Brooks Xinrong Ren
Mercury Research
Team at ARL:
Great Lakes Restoration Initiative, via Interagency Agreement with USEPA
Funding:
Rick Jiang Yan Huang
IT Team at ARL:
+ numerous collaborations with external partners involving:
o emissions inventory data for model input, and
o atmospheric measurement data for model evaluation
3
Mercury in the Environment
Biodiversity Research Institute (2011): http://www.briloon.org/uploads/hgconnections/glmc/GLMC_FinalReport.pdf
Evers, D.C., et al. (2011). Great Lakes Mercury Connections: The Extent and Effects of Mercury Pollution in the Great Lakes Region. Biodiversity Research Institute. Gorham, Maine. Report BRI 2011-18. 44 pages.
4
Mercury in Great Lakes Fish 0.05 ppm level
recommended by the Great Lakes Fish Advisory
Workgroup (2007)
19
90
19
92
19
94
19
96
19
98
20
00
20
02
20
04
20
06
20
08
Temporal trends of mercury in Lake Erie 45−55 cm walleye collected between 1990−2007
(Bhavsar et al., Environ. Sci. Technol. 2010, 44, 3273-3279)
5
Atmospheric deposition is believed to be the largest current mercury loading pathway to the Great Lakes…
How much is deposited and where does it come from? (…this information can only be obtained via modeling...)
6
7
Different “forms” of mercury in the atmosphere
Elemental Mercury -- Hg(0)
• most of total Hg in atmosphere
• doesn’t easily dry or wet deposit
• globally distributed
Reactive Gaseous Mercury -- RGM
• a few % of total atmos. Hg
• oxidized Hg (HgCl2, others)
• very water soluble and “sticky”
• bioavailable
Particulate Mercury -- Hg(p)
• a few % of total atmos. Hg
• Hg in/on atmos. particles
• atmos. lifetime 1~ 2 weeks
• bioavailability?
?
Elemental Mercury -- Hg(0)
• most of total Hg in atmosphere
• doesn’t easily dry or wet deposit
• globally distributed
Reactive Gaseous Mercury -- RGM
• a few % of total atmos. Hg
• oxidized Hg (HgCl2, others)
• very water soluble and “sticky”
• bioavailable
Particulate Mercury -- Hg(p)
• a few % of total atmos. Hg
• Hg in/on atmos. particles
• atmos. lifetime 1~ 2 weeks
• bioavailability?
?
Type of Emissions Source
coal-fired power plants other fuel combustion waste incineration metallurgical manufacturing & other
Emissions (kg/yr)
10-50
50-100
100–300
5-10
300–500
500–1000
1000–3000
8
2005 Atmospheric Mercury Emissions from Large Point Sources
Starting point: where is mercury emitted to the air?
2005 Atmospheric Mercury Emissions (Direct Anthropogenic + Re-emit + Natural)
Policy-Relevant Scenario Analysis
9
Dry and wet
deposition of
the pollutants
in the puff are
estimated at
each time step.
The puff’s mass, size,
and location are
continuously tracked…
Phase partitioning and chemical
transformations of pollutants within the
puff are estimated at each time step
= mass of pollutant
(changes due to chemical transformations and
deposition that occur at each time step)
Centerline of
puff motion
determined by
wind direction
and velocity
Initial puff location
is at source, with
mass depending
on emissions rate
TIME (hours)
0 1 2
deposition 1 deposition 2 deposition to receptor
lake
HYSPLIT-Hg Lagrangian Puff Atmospheric Fate and Transport Model
10
Next step: What happens to the mercury after it is emitted?
Uncertainties
Uncertainties
Uncertainties
To simulate the global
transport of mercury,
puffs are transferred to
Eulerian grid after a
specified time downwind
(~2 weeks), and the
mercury is simulated on
that grid from then on…
When puffs grow to sizes
large relative to the
meteorological data grid, they
split, horizontally and/or
vertically
This is how we model the
local & regional impacts.
But for global modeling,
puff splitting overwhelms
computational resources
Deposition explicitly modeled to actual lake/watershed areas
As opposed to the usual practice of ascribing portions of gridded deposition to these areas in a post-processing step
12
Lake Erie
Results scaled to actual RGM emissions of 43.6 g/hr 1 ng/m2-hr = 8.8 ug/m2-yr (if it persisted the entire year) Total deposition to Lk Erie is ~20 ug/m2-yr
Illustrative simulation of reactive gaseous mercury (RGM) emissions from one power plant on the shore of Lake Erie:
hourly deposition estimates for the first two weeks in May 2005
13
14
350,000 sources in global emissions inventory
typical one-year simulation takes ~96 processor hours
~3800 processor years, if ran explicit simulation for each source
Computational Challenge
Would like to keep track of each source individually
~240 years on 16-processor workstation
Spatial Interpolation
15
Chemical Interpolation
16
Standard Points in North America
17
For each standard source location, we do three unit-emissions simulations: o pure Hg(0), o pure HgII
(RGM) o pure Hg(p)
Standard Points Outside of North America
18
…three unit-emissions simulations for each location
19
This analysis done with 136 standard source locations
~4.5 processor years
Computational Solution
3 unit emissions simulations from each location (Hg(0), RGM, and Hg(p)
~3.5 months on 16-processor workstation
instead of 240 years … almost 1000x less!
Type of Emissions Source
coal-fired power plants other fuel combustion waste incineration metallurgical manufacturing & other
Emissions (kg/yr)
10-50
50-100
100–300
5-10
300–500
500–1000
1000–3000
20 2005 Atmospheric Mercury Emissions from Large Point Sources
an example for one source… the Monroe coal-fired power plant on the shore of Lake Erie
Detroit Edison Monroe coal fired power plant on the
shore of Lake Erie
Lake Erie
Monroe emitted 561 kg of mercury in 2005 (EPA’s National Emissions Inventory)
How much of this mercury was deposited into Lake Erie and its watershed?
21
Detroit Edison Monroe coal fired power plant on the
shore of Lake Erie
Lake Erie
Monroe emitted 561 kg of mercury in 2005 (EPA’s National Emissions Inventory)
Modeling results for this specific source:
• 24 kg (~4%) of this emitted mercury was deposited directly into Lake Erie
• 107 kg (~19%) of this emitted mercury was deposited in the Lake Erie Watershed
We make this same type of estimate for every source in the national and global emissions inventories used as model input… using spatial and chemical interpolation
22
HYSPLIT-Hg (with mercury-specific chemistry, …)
Unit Emissions Simulations of Hg(0), Hg(II) and Hg(p) from an array of standard source locations
Emissions Inventory – emissions of Hg(0), Hg(II), and Hg(p) from sources at specified latitudes and longitudes
“Multiplication” of emissions inventory by array of unit emissions simulations using spatial and chemical interpolation
Evaluate overall model results: compare against ambient measurements
Source-attribution results for deposition to selected receptors
HYSPLIT
Outline of Modeling Analysis
23
y = 0.95xR² = 0.59
y = 1.44xR² = 0.15
0
2
4
6
8
10
12
14
16
18
0 2 4 6 8 10 12
Mo
del
ed
Me
rcu
ry W
et
De
po
siti
on
(u
g/m
2-yr
)
Measured Mercury Wet Deposition (ug/m2-yr)
MDN sites in the "western" Great Lakes region
MDN sites in the "eastern" Great Lakes region
1:1 line
Linear (MDN sites in the "western" Great Lakes region)
Linear (MDN sites in the "eastern" Great Lakes region)
Error bars shown are the range in model predictions obtained
with different precipitation adjustment schemes (none, all,EDAS only, NCEP/NCAR only)
Error bars shown are the range in model predictions obtained
with different precipitation adjustment schemes (none, all,EDAS only, NCEP/NCAR only)
Modeled vs. Measured Wet Deposition of Mercury at Sites in the Great Lakes Region
24
After all the standard source simulations have been run, and the impacts of each of the ~350,000 sources worldwide are estimated using spatial and chemical interpolation, is the model giving reasonable results?
Standard source locations, MDN sites, and mercury emissions in the Great Lakes region
25
HYSPLIT-Hg (with mercury-specific chemistry, …)
Unit Emissions Simulations of Hg(0), Hg(II) and Hg(p) from an array of standard source locations
Emissions Inventory – emissions of Hg(0), Hg(II), and Hg(p) from sources at specified latitudes and longitudes
“Multiplication” of emissions inventory by array of unit emissions simulations using spatial and chemical interpolation
Evaluate overall model results: compare against ambient measurements
Source-attribution results for deposition to selected receptors
HYSPLIT
Outline of Modeling Analysis
26
2005 Atmospheric Mercury Emissions (Direct Anthropogenic + Re-emit + Natural)
Policy-Relevant Scenario Analysis
27
Geographical Distribution of 2005 Atmospheric Mercury Deposition Contributions to Lake Erie
Policy-Relevant Scenario Analysis
28
Keep track of the contributions from each source, and add them up
-
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
< 5
00
km
50
0 -
1,0
00
km
1,0
00
-3
,00
0 k
m
3,0
00
-1
0,0
00
km
10
,00
0 -
20
,00
0 k
m
Me
rcu
ry E
mis
sio
ns
(M
g/y
r)
Distance of Emissions Source from
the Center of Lake Erie
Emissions from Natural Sources
Emissions from Re-Emissions
Emissions from Anthropogenic
Sources
A tiny fraction of 2005 global mercury emissions within 500 km of Lake Erie
-
50
100
150
200
250
< 5
00
km
50
0 -
1,0
00
km
1,0
00
-3
,00
0 k
m
3,0
00
-1
0,0
00
km
10
,00
0 -
20
,00
0 k
m
De
po
siti
on
Co
ntr
ibu
tio
n (
kg/y
r)
Distance of Emissions Source from
the Center of Lake Erie
Contributions from Natural Sources
Contributions from Re-Emissions
Contributions from Anthropogenic
Sources
Modeling results show that these “regional” emissions are responsible for a large fraction of the modeled 2005 atmospheric deposition
Important policy implications!
29
Results can be shown in many ways…
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1 10 100 1000 10000
Cumulative Fraction of Total
Modeled Mercury
Deposition to Lake Erie
(2005)
Rank of Source's Atmospheric Mercury Deposition Contribution to Lake Erie
30
…350,000
DET
RO
IT E
DIS
ON
MO
NR
OE
PO
WER
CO
NES
VIL
LE REL
IAN
T EN
ERG
Y A
VO
N L
AK
E
FIR
STEN
ERG
Y C
OR
P E
AST
LAK
E
REL
IAN
T EN
ERG
Y K
EYST
ON
E
Nan
tico
ke G
ener
atin
g St
atio
n
PA
PO
WER
CO
BR
UC
E M
AN
SFIE
LD P
LT
DET
RO
IT E
DIS
ON
TR
ENTO
N C
HA
NN
EL
CA
RD
INA
L P
OW
ER P
LAN
T
REL
IAN
T EN
ERG
Y SH
AW
VIL
LE
W. H
. SA
MM
IS P
LAN
T
DU
NK
IRK
STE
AM
GEN
ERA
TIN
G S
TATI
ON
HO
MER
CIT
Y O
L H
OM
ER C
ITY
GEN
STA
CLE
VEL
AN
D E
LEC
TRIC
ASH
TAB
ULA
TOLE
DO
ED
ISO
N C
O. B
AY
SHO
RE
OH
IO P
OW
ER -
MIT
CH
ELL
PLA
NT
ALL
EGH
ENY
ENER
GY
HA
TFIE
LDS
FER
RY
MO
NO
NG
AH
ELA
PO
WER
FO
RT
MA
RTI
N
J R
WH
ITIN
G C
O
DET
RO
IT E
DIS
ON
RIV
ER R
OU
GE
OH
IO P
OW
ER -
KA
MM
ER P
LAN
T
ALL
EGH
ENY
ENER
GY
AR
MST
RO
NG
OR
ION
PO
WER
NEW
CA
STLE
DET
RO
IT E
DIS
ON
ST.
CLA
IR
MO
UN
T ST
OR
M P
OW
ER P
LAN
T
R. E
. BU
RG
ER P
LAN
T
HU
NTL
EY S
TEA
M G
ENER
ATI
NG
STA
TIO
N
AP
PA
LAC
HIA
N P
OW
ER -
JOH
N E
AM
OS
Do
mta
r P
aper
Jo
hn
son
bu
rg M
ill
CLE
VEL
AN
D E
LEC
TRIC
LA
KE
SHO
RE
PLA
NT
Lam
bto
n G
ener
atin
g St
atio
n
PP
G IN
DU
STR
IES
OH
IO C
IRC
LEV
ILLE
Essr
oc
Cem
ent
Co
rp.
LAFA
RG
E SY
STEC
H E
NV
IRO
NM
ENTA
L C
OR
P.
Nia
gara
Fal
ls
RO
SS IN
CIN
ERA
TIO
N S
ERV
ICES
INC
.
Bio
med
ical
Was
te In
cin
erat
ion
_43
20
Bio
med
ical
Was
te In
cin
erat
ion
_42
45
STER
ICYC
LE IN
C
VO
N R
OLL
AM
ERIC
A IN
C
Bio
med
ical
Was
te In
cin
erat
ion
_43
19
CEN
TRA
L W
AYN
E EN
ERG
Y R
ECO
VER
Y
Rep
ub
lic E
ngi
nee
red
Ste
els
Inc
NO
RTH
STA
R B
LUES
CO
PE
STEE
L LL
C
AK
STE
EL C
OR
PO
RA
TIO
N
No
rth
Sta
r St
eel
ISG
CLE
VEL
AN
D IN
C.
Van
dM
STA
R
ASH
TA C
HEM
ICA
LS IN
C
GE
RA
VEN
NA
LA
MP
PLA
NT
MI
OH
OH
OH
OH
OH
PA O
N
PA M
I OH P
A OH N
Y PA OH O
H WV IN P
A WV OH M
I MI
WV NY PA PA M
I
WV OH OH OH OH NY
OH
WV
ON ON PA
OH
OH
ON
OH
ON M
I
OH
OH
OH M
I
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
0 5 10 15 20 25 30 35 40 45 50
Cu
mu
lati
ve F
ract
ion
of
Tota
l M
od
ele
d D
ep
osi
tio
n (
20
05
)
Rank of Source's Atmospheric Mercury Deposition Contribution to Lake Erie
Top 50 Atmospheric Deposition Contributors to Lake Erie
coal fired power plants
other fuel combustion
waste incineration
metallurgical
manufacturing and other
Based on estimated 2005 mercury emissions, e.g., from the 2005 USEPA National Emissions Inventory, and atmospheric fate and transport simulations with the NOAA HYSPLIT-Hg model
31
Natural 23%
Ocean
Re-emission 14%
U.S. 32%
China 14%
Canada 3%
India 2%
Other Countries 12%
Sources of Mercury Deposition to the Great Lakes Basin 2005 Baseline Analysis
Total = 11,300 kg/yr
32
Natural 17%
Ocean Re-emission
10%
U.S. 49%
China 10%
Canada 4%
India 1%
Other Countries 9%
Sources of Mercury Deposition to the Lake Erie Basin
2005 Baseline Analysis
Total = 2,300 kg/yr
33
Natural 23%
Ocean
Re-emission 14%
U.S. 32%
China 14%
Canada 3%
India 2%
Other Countries
12%
Sources of Mercury Deposition to the Great Lakes Basin 2005 Baseline Analysis
Total = 11,300 kg/yr
Natural 17%
Ocean Re-emission
10%
U.S. 49%
China 10%
Canada 4%
India 1%
Other Countries
9%
Sources of Mercury Deposition to the Lake Erie Basin
2005 Baseline Analysis
Total = 2,300 kg/yr
34
Jan 1, 2010
Jan 1, 2011
Jan 1, 2012
Jan 1, 2013
Jan 1, 2014
Jan 1, 2009
Jan 1, 2015
Jan 1, 2016
ARL’s GLRI Atmospheric Mercury Modeling Project
FY12 $ Scenario Analysis
FY13 $ (proposed) Update Analysis (~2008)
FY11 $ Sensitivity Analysis + Extended Model Evaluation
FY10 $ Baseline Analysis for 2005
Initial Inter- and Intra-Agency Planning for FY10 GLRI Funds
FY14 $ (proposed) Update Analysis (~2011)
35
A multi-phase project
Using 2005 meteorological data and emissions, the deposition and source-attribution for this deposition to each Great Lake and its watershed was estimated
2005 was chosen as the analysis year, because 2005 was the latest year for which comprehensive mercury emissions inventory data were available at the start of this project
Phase 1: Baseline Analysis for 2005 (Final Report Completed December 2011)
The model results were ground-truthed against 2005 Mercury Deposition Network data from sites in the Great Lakes region
36
Modeling Atmospheric Mercury Deposition to the Great Lakes. Final Report for work conducted with FY2010 funding from the
Great Lakes Restoration Initiative. December 16, 2011. Mark Cohen, Roland Draxler, Richard Artz. NOAA Air Resources Laboratory, Silver Spring, MD, USA. 160 pages.
http://www.arl.noaa.gov/documents/reports/GLRI_FY2010_Atmospheric_Mercury_Final_Report_2011_Dec_16.pdf
http://www.arl.noaa.gov/documents/reports/Figures_Tables_GLRI_NOAA_Atmos_Mercury_Report_Dec_16_2011.pptx
37
One-page summary: http://www.arl.noaa.gov/documents/reports/GLRI_Atmos_Mercury_Summary.pdf
Some Key Features of this Analysis
Deposition explicitly modeled to actual lake/watershed areas
Uniquely detailed source-attribution information is created
As opposed to the usual practice of ascribing portions of gridded deposition to these areas in a post-processing step
deposition contribution to each Great Lakes and watersheds from each source in the emissions inventories used is estimated individually
The level of source discrimination is only limited by the detail in the emissions inventories
Source-type breakdowns not possible in this 1st phase for global sources, because the global emissions inventory available did not have source-type breakdowns for each grid square
Combination of Lagrangian & Eulerian modeling
allows accurate and computationally efficient estimates of the fate and transport of atmospheric mercury over all relevant length scales – from “local” to global.
38
Some Key Findings of this Analysis
Regional, national, & global mercury emissions are all important contributors to mercury deposition in the Great Lakes Basin
For Lakes Erie and Ontario, the U.S. contribution is at its most significant
For Lakes Huron and Superior, the U.S. contribution is less significant.
Local & regional sources have a much greater atmospheric deposition contributions than their emissions, as a fraction of total global mercury emissions, would suggest.
“Single Source” results illustrate source-receptor relationships
For example, a “typical” coal-fired power plant near Lake Erie may contribute on the order of 1000x the mercury – for the same emissions – as a comparable facility in China.
39
Some Key Findings of this Analysis (…continued)
Reasonable agreement with measurements
Despite numerous uncertainties in model input data and other modeling aspects
Comparison at sites where significant computational resources were expended – corresponding to regions that were the most important for estimating deposition to the Great Lakes and their watersheds – showed good consistency between model predictions and measured quantities.
For a smaller subset of sites generally downwind of the Great Lakes (in regions not expected to contribute most significantly to Great Lakes atmospheric deposition), less computational resources were expended, and the comparison showed moderate, but understandable, discrepancies.
40
Ground-truthing the model against additional ambient monitoring data, e.g., ambient mercury air concentration measurements and wet deposition data not included in the Mercury Deposition Network (MDN)
Examining the influence of uncertainties on the modeling results, by varying critical model parameters, algorithms, and inputs, and analyzing the resulting differences in results
Phase 2: Sensitivity Analysis + Extended Model Evaluation (current work, with GLRI FY11 funding)
41
0%
2%
4%
6%
8%
10%
12%
14%
bas
elin
e
afte
r cr
ash
; ne
w c
om
pile
rar
ray
fix
mat
h o
pt.
= 0
(vs
2),
arr
ay f
ixar
ray
pat
ch, g
rid
exp
ansi
on
= f
ull
(vs
1.5
x)m
ath
op
t. =
0 (
vs 2
), a
rray
fix
, gri
d e
xp =
fu
ll (v
s.…
rele
ase
ele
vati
on
= 5
0 m
(vs
25
0)
EDA
S re
gio
nal
me
t d
ata
(vs
NA
RR
)
tim
e st
ep
= 2
0 m
in (
vs 1
2)
tim
e st
ep
= 3
0 m
in (
vs 1
2)
max
pu
ff li
feti
me
= 2
wks
(vs
3)
max
pu
ff li
feti
me
= 4
wks
(vs
3)
PU
F +
GEM
(vs
PU
F o
nly
)
max
nu
mb
er
of
pu
ffs
= 1
0K
(vs
20
K)
max
nu
mb
er
of
pu
ffs
= 4
0K
(vs
20
K)
1 p
uff
pe
r 7
hrs
(vs
3),
sp
lit a
fter
7 h
rs (
vs 2
4)
1 p
uff
pe
r 7
hrs
(vs
3),
sp
lit a
fter
48
hrs
(vs
24
)
Hg0
dry
de
p =
0 (
vs m
od
ele
d)
wet
dep
was
ho
ut
par
ame
ter
*0
.5w
et d
ep w
ash
ou
t p
aram
ete
r *
2
hv
red
uct
ion
aq
* 2
hv
red
uct
ion
aq
* 0
.5o
3 o
xid
atio
n g
as *
2o
3 o
xid
atio
n g
as *
0.5
oh
oxi
dat
ion
aq
* 2
oh
oxi
dat
ion
aq
* 0
.5o
h o
xid
atio
n g
as *
2o
h o
xid
atio
n g
as *
0.5
so2
red
uct
ion
aq
* 2
so2
red
uct
ion
aq
* 0
.5so
ot
par
titi
on
eq
lbrm
* 2
soo
t p
arti
tio
n e
qlb
rm *
0.5
soo
t p
arti
tio
n r
ate
* 2
soo
t p
arti
tio
n r
ate
* 0
.5
Fraction of RGM
emissions deposited in
Lake Erie from a
hypothetical source
in Detroit
Fort
ran
co
mp
ilati
on
Ch
em
istr
y an
d
Par
titi
on
ing
De
po
siti
on
m
od
elin
g
Tim
e s
tep
Pu
ff li
feti
me
Nu
mb
er
of
pu
ffs
Met data (EDAS vs. NARR)
Release height 50 m vs 250 m
42
0.000%
0.001%
0.002%
0.003%
0.004%
0.005%
0.006%
bas
elin
e
afte
r cr
ash
; ne
w c
om
pile
r
rele
ase
ele
vati
on
= 5
0 m
(vs
25
0)
max
nu
mb
er
of
pu
ffs
= 1
0K
(vs
20
K)
max
nu
mb
er
of
pu
ffs
= 4
0K
(vs
20
K)
Hg0
dry
de
p =
0 (
vs m
od
ele
d)
wet
dep
was
ho
ut
par
ame
ter
*0
.5
wet
dep
was
ho
ut
par
ame
ter
*2
hv
red
uct
ion
aq
* 2
hv
red
uct
ion
aq
* 0
.5
o3
oxi
dat
ion
gas
* 2
o3
oxi
dat
ion
gas
* 0
.5
oh
oxi
dat
ion
aq
* 2
oh
oxi
dat
ion
aq
* 0
.5
oh
oxi
dat
ion
gas
* 2
oh
oxi
dat
ion
gas
* 0
.5
so2
red
uct
ion
aq
* 2
so2
red
uct
ion
aq
* 0
.5
soo
t p
arti
tio
n e
qlb
rm *
2
soo
t p
arti
tio
n e
qlb
rm *
0.5
soo
t p
arti
tio
n r
ate
* 2
soo
t p
arti
tio
n r
ate
* 0
.5
Fraction of Hg(0)
emissions deposited in
Lake Erie from a
hypothetical source
in China
Co
mp
ilati
on
Ch
em
istr
y an
d
Par
titi
on
ing
De
po
siti
on
m
od
elin
g
Nu
mb
er
of
pu
ffs
Emit
he
igh
t
43
We will work with EPA and other Great Lakes Stakeholders to identify and specify the most policy relevant scenarios to examine
A modeling analyses such as this is the only way to quantitatively examine the potential consequences of alternative future emissions scenarios
Phase 3: Scenarios (next year’s work, with GLRI FY12 funding)
For each scenario, we will estimate the amount of atmospheric deposition to each of the Great Lakes and their watersheds, along with the detailed source-attribution for this deposition
44
Thanks!
45
46
EXTRA SLIDES
0
2
4
6
8
10
12
14
16
18
20
Erie Ontario Michigan Huron Superior
Atm
osp
he
ric
Me
rcu
ry D
ep
osi
tio
n(u
g/m
2-yr
)
Great Lake
Natural
Ocean Re-Emit
All Other Countries
Canada
China
United States
47
CLOUD DROPLET
cloud
Primary Anthropogenic
Emissions
Hg(II), ionic mercury, RGM
Elemental Mercury [Hg(0)]
Particulate Mercury [Hg(p)]
Re-emission of previously deposited anthropogenic
and natural mercury
Hg(II) reduced to Hg(0) by SO2 and sunlight
Hg(0) oxidized to dissolved Hg(II) species by O3, OH,
HOCl, OCl-
Adsorption/ desorption of Hg(II) to /from soot
Natural emissions
Upper atmospheric halogen-mediated oxidation?
Polar sunrise “mercury depletion events”
Br
Dry deposition
Wet deposition
Hg(p)
Vapor phase: Hg(0) oxidized to RGM and Hg(p) by O3, H202, Cl2, OH, HCl
Multi-media interface
Atmospheric Mercury Fate Processes
Reaction Rate Units Reference
GAS PHASE REACTIONS
Hg0 + O3 Hg(p) 3.0E-20 cm3/molec-sec Hall (1995)
Hg0 + HCl HgCl2 1.0E-19 cm3/molec-sec Hall and Bloom (1993)
Hg0 + H2O2 Hg(p) 8.5E-19 cm3/molec-sec Tokos et al. (1998) (upper limit
based on experiments)
Hg0 + Cl2 HgCl2 4.0E-18 cm3/molec-sec Calhoun and Prestbo (2001)
Hg0 +OH Hg(p) 8.7E-14 cm3/molec-sec Sommar et al. (2001)
Hg0 + Br HgBr2
AQUEOUS PHASE REACTIONS
Hg0 + O3 Hg+2 4.7E+7 (molar-sec)-1 Munthe (1992)
Hg0 + OH Hg+2 2.0E+9 (molar-sec)-1 Lin and Pehkonen(1997)
HgSO3 Hg0 T*e((31.971*T)-12595.0)/T) sec-1
[T = temperature (K)]
Van Loon et al. (2002)
Hg(II) + HO2 Hg0 ~ 0 (molar-sec)-1 Gardfeldt & Jonnson (2003)
Hg0 + HOCl Hg+2 2.1E+6 (molar-sec)-1 Lin and Pehkonen(1998)
Hg0 + OCl-1 Hg+2 2.0E+6 (molar-sec)-1 Lin and Pehkonen(1998)
Hg(II) Hg(II) (soot) 9.0E+2 liters/gram;
t = 1/hour
eqlbrm: Seigneur et al. (1998)
rate: Bullock & Brehme (2002).
Hg+2 + hv Hg0 6.0E-7 (sec)-1 (maximum)
Xiao et al. (1994);
Bullock and Brehme (2002)
(Evolving) Atmospheric Chemical Reaction Scheme for Mercury
?
?
?
new
What year to model?
Mercury Emissions Inventory
Meteorological Data to drive model
Ambient Data for Model Evaluation
Need all of these datasets for the
same year
Dataset Available for 2005
U.S. anthropogenic emissions inventory Canadian anthropogenic emissions inventory Mexican anthropogenic emissions inventory Global anthropogenic emissions inventory Natural emissions inventory Re-emissions inventory
Wet deposition (Mercury Deposition Network) “Speciated” Air Concentrations
NCEP/NCAR Global Reanalysis (2.5 deg) NCEP EDAS 40km North American Domain North American Regional Reanalysis (NARR)
2005 chosen for baseline
analysis
50
51
Getting good ground-truthing results harder
than estimating deposition to the Great Lakes One Standard Source Location (green dot) would do a decent job of estimating deposition to the receptor, for all of the hypothetical, “actual” source locations shown (numbered boxes) But the same Standard Source Location would be completely inadequate to estimate deposition and concentrations at the monitoring site (red star)
52
Standard Source Locations for Illustrative Modeling Results
53
1.0E-11
1.0E-10
1.0E-09
1.0E-08
1.0E-07
1.0E-06
1.0E-058 97
86
10
91 7 9 1 90 6 87
94 5 33
11
7
88 4 93
35
74
32
26
22
20
12
19
21
17
37
16
27
18
10
5
10
6
10
4
54
50
72
14
70
13
67
15
68
66
56
84
49
Great Lakes Regional Inset Map North American Regional Inset Map Global Map
Tran
sfe
r Fl
ux
Co
eff
icie
nt
to L
ake
Eri
e (
1/k
m2
)
Standard Source Location Number
Transfer Flux Coefficient to Lake Erie for a "Typical" Coal-Fired Power Plant
Transfer Flux Coefficient to Lake Erie for a Coal-Fired Power Plant with a higher RGM emissions fraction
The "Transfer Flux Coefficient" is calculated as the atmospheric deposition flux to a given receptor (in this case, Lake Erie) in units of g/km2-yr, divided by the total emissions from the source, in units of g/yr.
With this transfer flux coefficient, if one knows the emissions of the source in the given location, then the atmospheric dep osition fluximpact of the source on the receptor can be estimated, by simply multiplying the emissions by the transfer flux coefficient.
Lake Erie Transfer Flux Coefficients for two kinds of Generic Coal-Fired Power Plants (logarithmic scale)
54
0.0E+00
2.0E-07
4.0E-07
6.0E-07
8.0E-07
1.0E-06
1.2E-06
1.4E-06
1.6E-06
1.8E-06
2.0E-068 97
86
10
91 7 9 1 90 6 87
94 5 33
11
7
88 4 93
35
74
32
26
22
20
12
19
21
17
37
16
27
18
10
5
10
6
10
4
54
50
72
14
70
13
67
15
68
66
56
84
49
Great Lakes Regional Inset Map North American Regional Inset Map Global Map
Tran
sfe
r Fl
ux
Co
eff
icie
nt
to L
ake
Eri
e (
1/k
m2
)
Standard Source Location Number
Transfer Flux Coefficient to Lake Erie for a "Typical" Coal-Fired Power Plant
Transfer Flux Coefficient to Lake Erie for a Coal-Fired Power Plant with a higher RGM emissions fraction
The "Transfer Flux Coefficient" is calculated as the atmospheric deposition flux to a given receptor (in this case, Lake Erie) in units of g/km2-yr, divided by the total emissions from the source, in units of g/yr.
With this transfer flux coefficient, if one knows the emissions of the source in the given location, then the atmospheric dep osition fluximpact of the source on the receptor can be estimated, by simply multiplying the emissions by the transfer flux coefficient.
Lake Erie Transfer Flux Coefficients for two kinds of Generic Coal-FIred Power Plants (linear scale)
55
In order to conveniently compare different model results, a “transfer flux coefficient” X will be used,
defined as the following:
56
57
Transfer Flux Coefficients For Pure Elemental Mercury Emissions at an Illustrative Subset of Standard Source Locations, for Deposition Flux Contributions to Lake Erie
58
Transfer Flux Coefficients For Pure Reactive Gaseous Mercury Emissions at an Illustrative Subset of Standard Source Locations, for Deposition Flux Contributions to Lake Erie
59
1.0E-11
1.0E-10
1.0E-09
1.0E-08
1.0E-07
1.0E-06
1.0E-058 97
86
10
91 7 9 1 90 6 87
94 5 33
11
7
88 4 93
35
74
32
26
22
20
12
19
21
17
37
16
27
18
10
5
10
6
10
4
54
50
72
14
70
13
67
15
68
66
56
84
49
Great Lakes Regional Inset Map North American Regional Inset Map Global Map
Tran
sfe
r Fl
ux
Co
eff
icie
nt
to L
ake
Eri
e (
1/k
m2
)
Standard Source Location Number
Transfer Flux Coefficient to Lake Erie for Pure Hg(II) Emissions
Transfer Flux Coefficient to Lake Erie for Pure Hg(p) Emissions
Transfer Flux Coefficient to Lake Erie for Pure Hg(0) Emissions
The "Transfer Flux Coefficient" is calculated as the atmospheric deposition flux to a given receptor (in this case, Lake Erie) in units of g/km2-yr, divided by the total emissions from the source, in units of g/yr.
With this transfer flux coefficient, if one knows the emissions of the source in the given location, then the atmospheric dep osition fluximpact of the source on the receptor can be estimated, by simply multiplying the emissions by the transfer flux coefficient.
Transfer Flux Coefficients For Hg(0), Hg(II), and Hg(p) to Lake Erie (logarithmic scale)
60
0.0E+00
5.0E-07
1.0E-06
1.5E-06
2.0E-06
2.5E-06
3.0E-068 97
86
10
91 7 9 1 90 6 87
94 5 33
11
7
88 4 93
35
74
32
26
22
20
12
19
21
17
37
16
27
18
10
5
10
6
10
4
54
50
72
14
70
13
67
15
68
66
56
84
49
Great Lakes Regional Inset Map North American Regional Inset Map Global Map
Tran
sfe
r Fl
ux
Co
eff
icie
nt
to L
ake
Eri
e (
1/k
m2
)
Standard Source Location Number
Transfer Flux Coefficient to Lake Erie for Pure Hg(II) Emissions
Transfer Flux Coefficient to Lake Erie for Pure Hg(p) Emissions
Transfer Flux Coefficient to Lake Erie for Pure Hg(0) Emissions
The "Transfer Flux Coefficient" is calculated as the atmospheric deposition flux to a given receptor (in this case, Lake Erie) in units of g/km2-yr, divided by the total emissions from the source, in units of g/yr.
With this transfer flux coefficient, if one knows the emissions of the source in the given location, then the atmospheric dep osition fluximpact of the source on the receptor can be estimated, by simply multiplying the emissions by the transfer flux coefficient.
Transfer Flux Coefficients For Hg(0), Hg(II), and Hg(p) to Lake Erie (linear scale)
61
Anthropogenic Mercury Emissions (ca. 2005)
Natural mercury emissions
Figure 55. Mercury Deposition Network Sites in the Great Lakes Region Considered in an Initial Model Evaluation Analysis
NOAA Air Resources Laboratory 64 Discussion July 5, 2012
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0N
Y68
PA7
2
NY2
0
IN2
8
IN2
6
PQ
04
PA6
0
PA4
7
IN2
1
PA9
0
PA0
0
PA3
7
PA1
3
PA3
0
PQ
05
IL1
1
IN2
0
OH
02
MN
27
WI3
6
MN
23
MN
22
MI4
8
MN
18
MN
16
WI9
9
WI3
1
WI2
2
WI0
9
IN3
4
WI3
2
ON
07
Me
asu
red
an
d M
od
ele
d P
reci
pit
atio
n (m
/yr)
max sample or rain gauge precip
sample precip
rain gauge precip
NCEP/NCAR Reanalysis precip
EDAS precip
Figure 56. Comparison of Total 2005 Precipitation Measured at each of the Great-Lakes Region MDN Sites with the Precipitation in the Meteorological Datasets Used as Inputs to this Modeling Study
NOAA Air Resources Laboratory 65 Discussion July 5, 2012
Comparison of 2005 precipitation total as measured at MDN sites in the Great Lakes region (circles) with precipitation totals assembled
by the PRISM Climate Group, Oregon State University
NOAA Air Resources Laboratory 66 Discussion July 5, 2012
y = 1.11x
0
2
4
6
8
10
12
14
16
18
0 2 4 6 8 10 12
Mo
del
ed
Me
rcu
ry W
et
De
po
siti
on
(u
g/m
2-yr
)
Measured Mercury Wet Deposition (ug/m2-yr)
model result (136 std pts)
1:1 line
Linear Regression
Error bars shown are the range in model predictions obtained
with different precipitation adjustment schemes (none, all,EDAS only, NCEP/NCAR only)
Modeled vs. Measured Wet Deposition of Mercury at Sites in the Great Lakes Region
NOAA Air Resources Laboratory 67 Discussion July 5, 2012
0
2
4
6
8
10
12
14
16
18
IN2
1
IN2
6
IN2
8
MN
27
OH
02
WI2
2
MN
22
MN
23
IL1
1
WI9
9
IN2
0
WI3
1
IN3
4
MI4
8
WI3
2
WI0
9
WI3
6
MN
18
MN
16
WI0
8
NY6
8
PA3
0
PA6
0
PA0
0
PA7
2
PA4
7
PQ
04
PA3
7
PA9
0
PA1
3
NY2
0
ON
07
2005
To
tal W
et
Me
rcu
ry D
ep
osi
tio
n(u
g/m
2-yr
)
Mercury Deposition Network Site
measurement
model result (136 std pts)
Error bars shown are the range
in model predictions obtainedwith different precipitation
adjustment schemes (none, all,EDAS only, NCEP/NCAR only)
MDN sites in the "western"
Great Lakes region
MDN sites in the "eastern"
Great Lakes region
Modeled vs. Measured Wet Deposition of Mercury at Sites in the Great Lakes Region
NOAA Air Resources Laboratory 68 Discussion July 5, 2012
Inventory domain Number
of records
Hg(0) emissions
(Mg/yr)
RGM emissions
(Mg/yr)
Hg(p) emissions
(Mg/yr)
Total mercury
emissions (Mg/yr)
U.S. Point Sources United States 19,353 50.6 35.5 9.1 95
U.S. Area Sources United States 44,848 4.5 1.8 1.1 7.4
Canadian Point Sources Canada 166 3.0 1.7 0.4 5.1
Canadian Area Sources Canada 12,372 1.0 0.96 0.42 2.4
Mexican Point Sources Mexico 268 28 0.81 0.46 29
Mexican Area Sources Mexico 160 1.25 0.38 0.25 1.9
Global Anthropogenic Sources not in U.S., Canada, or Mexico
Global, except for the U.S., Canada, and Mexico
52,173 1,239 434 113 1,786
Global Re-emissions from Land
Global land (and freshwater) surfaces
129,180 750 0 0 750
Global Re-emissions from the Ocean
Global oceans 43,324 1,250 0 0 1,250
Global Natural Sources Global 64,800 1,800 0 0 1,800
Total 366,804 5,127 475 125 5,728
Summary of Mercury Emissions Inventories Used in GLRI Analysis
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
Un
ited
Sta
tes
Ch
ina
Can
ada
Ind
ia
Ru
ssia
Mex
ico
Ind
on
esia
Sou
th K
ore
a
Jap
an
Co
lom
bia
Tota
l Atm
osp
he
ric
De
po
siti
on
to
the
Gre
at L
ake
s B
asin
(kg
/yr)
anthropogenic re-emissions
from land
direct anthropogenic
emissions
Model-estimated 2005 deposition to the Great Lakes Basin from countries with the highest modeled contribution from direct and re-emitted anthropogenic sources
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
Un
ited
Sta
tes
Ch
ina
Can
ada
Ind
ia
Ru
ssia
Mex
ico
Ind
on
esia
Sou
th K
ore
a
Jap
an
Co
lom
bia
Tota
l Atm
osp
he
ric
De
po
siti
on
to
th
e
Gre
at L
ake
s B
asin
(u
g/y
r-p
ers
on
)
anthropogenic re-emissions
from land
direct anthropogenic
emissions
Model-estimated per capita 2005 deposition to the Great Lakes Basin from countries with the highest modeled contribution from direct & re-emitted anthropogenic sources