EPA/600/A-95/115
2
The Modeling of Regional-Scale Atmospheric Mercury Transport and
Deposition Using RELMAP
O. Russell Bullock, Jr.*
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, NC 27711
William G. Benjey"
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, NC 27711
Martha H. Keating
Air Quality Strategies and Standards Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
* On assignment to the U.S. Environmental Protection Agency, National Exposure Research Laboratory
-------
1
Modeling Regional-Scale Atmospheric Mercury Using RELMAP
O. Russell Bullock, Jr.
Atmospheric Modeling Division
US EPA, MD-80
Research Triangle Park, NC 27711
Phone: 919-541-1349
Fax: 919-541-1379
-------
J
DISCLAIMER
The information in this document has been funded wholly or in part by the U.S.
Environmental Protection Agency. It has been subject to Agency review and has been
approved for publication. Mention of trade name or commercial products does not constitute
endorsement or recommendation for use.
-------
4
Abstract - The Regional Lagrangian Model of Air Pollution (RELMAP) is used to simulate
the emission, transport and diffusion, chemical transformation, and wet and dry deposition of
elemental mercury gas, divalent mercury gas and particulate mercury. Based on recent
modeling advances in Europe, the RELMAP has been modified to simulate a reduction-
oxidation (redox) balance for mercury dissolved in cloud and rain water. This redox balance
is used in the estimation of a variable precipitation scavenging ratio for elemental mercury.
The result is more effective removal of elemental mercury gas by precipitation processes than
is implied by its solubility in water. Wet removal of divalent mercury gas and particulate
mercury is modeled using constant wet scavenging ratios. Dry deposition is estimated using
deposition velocities developed from various modeling studies in the U.S. and in Europe. A
mercury air emission inventory for the continental United States is used to estimate the
releases of elemental mercury gas, divalent mercury gas and particulate mercury based on
emission speciation estimates for the various emitter types in the inventory. Average annual
concentrations and wet and dry deposition totals for these three forms of mercury have been
simulated. The results of the simulation are used to estimate the quantity of mercury emitted
to the air annually over the United States, and the amount that is subsequently deposited back
to U.S. soils and water bodies. An analysis of the modeling results also provides some
information about the areas of the country thought to have the most significant exposure from
all air emissions of mercury. This analysis contributes to the understanding of the key
variables, such as source location, chemical and physical form of emission, or meteorology,
that lead to these outcomes.
Key Words - Mercury, Atmospheric, Modeling, Deposition, Emission
-------
5
BACKGROUND INFORMATION ON RELMAP
During the mid-1970's, SRI International developed a Lagrangian puff air pollution
model called the EUropean Regional Model of Air Pollution (EURMAP) for the Federal
Environment Office of the Federal Republic of Germany [1], This regional model simulated
monthly sulfur dioxide (S02) and sulfate (SO;;") concentrations, wet and dry deposition
patterns, and generated matrices of international exchanges of sulfur for 13 countries of
western and central Europe. In the late-1970's, the US EPA sponsored SRI International to
adapt and apply EURMAP to eastern North America. The adapted version of this model was
called the Eastern North American Model of Air Pollution (ENAMAP) [2,3].
By 1985, simple parameterizations of processes involving fine (diameters < 2.5 |j,m)
and coarse (2.5 (am < diameters < 10.0 |o.m) particulate matter were incorporated into the
model. This version of the model, renamed the REgional Lagrangian Model of Air Pollution
(RELMAP), is capable of simulating concentrations and wet and dry deposition patterns of
S02, SO4" and fine and coarse particulate matter and can also generate source-receptor
matrices for user defined regions. Currently, the RELMAP is operated by EPA's National
Exposure Research Laboratory (NERL) on a wide variety of computing systems. The data
preparation for this modeling study was performed at EPA's National Environmental
Supercomputing Center and the model calculation was performed on the UNIX computing
workstation network at the Atmospheric Modeling Division of NERL. A complete scientific
specification of the RELMAP as used at EPA for atmospheric sulfur modeling has been
developed and published [4]. The next section will discuss the modifications made to the
original sulfur version of the RELMAP to enable the simulation of atmospheric mercury.
-------
6
MODELING STRATEGY
General Information
Previous versions of the RELMAP have been described and evaluated [4,5], The goal
of this study was to model the emission, transport and fate of airborne mercury over the
continental U.S. for the year of 1989. Our modifications to the RELMAP for atmospheric
mercury simulation were heavily based on recent Lagrangian model developments in Europe
[6], The mercury version of the RELMAP was developed to handle three species of mercury:
elemental vapor (Hg°), divalent vapor (the mercuric ion, Hg2+) and particulate Hg (HgP), and
also aerosol carbon soot. Recent experimental work indicates that ozone [7] and carbon soot
[8,9,10] are both important in determining the wet deposition of Hg°. Carbon soot, or total
carbon aerosol, was included as a modeled pollutant in the mercury version of RELMAP to
provide necessary information for the Hg° wet deposition parameterization. Observed ozone
(03) air concentration data were obtained from EPA's Aerometric Information Retrieval
System (AIRS) data base. Thus, it was not necessary to include 03 as an explicitly modeled
pollutant. Observed 03 air concentration data were objectively interpolated in time and space
for each 3-hour time step of the model simulation to produce analyses of 03 air concentration.
Methyl mercury was not included in the mercury version of RELMAP because it is not yet
known if it has a primary natural or anthropogenic source, or if it is produced in the
atmosphere.
RELMAP may be run in either of two modes. In the field mode, wet deposition, dry
deposition, and air concentrations are computed at user-defined time intervals. In the source-
-------
7
receptor mode, RELMAP also computes the contribution of each source cell to the deposition
and concentration at each receptor cell. For mercury, only the field mode of RELMAP
operation was used. With over 10,000 model cells in the high-resolution receptor grid and a
significant fraction of these cells also emitting mercury, the data accounting task of a source-
receptor run for all mercury sources could not be performed.
Unless specified otherwise in the following sections, the modeling concepts and
parameterizations described in the EPA users' guide [4] were preserved for the RELMAP
mercury modeling study.
Physical Model Structure
RELMAP simulations were originally limited to the area bounded by 25 and 55
degrees north latitude and 60 and 105 degrees west longitude, and had a minimum spatial
resolution of 1 degree in both latitude and longitude. For this study, the western limit of the
RELMAP modeling domain was moved out to 130 degrees west longitude, and the modeling
grid resolution was reduced to Vi degree longitude by 1/3 degree latitude (approximately 40 km
square) to provide high-resolution coverage over the entire continental U.S.
The original 3-layer puff structure of the RELMAP has been replaced by a 4-layer
"structure. The following model layer definitions were used for the RELMAP mercury
simulations:
Layer 1 top
Layer 2 top
30 to 50 meters above the surface (season-dependent)
200 meters above the surface
-------
Layer 3 top
Layer 4 top
700 meters above the surface
700 to 1500 meters above the surface (month-dependent)
8
Mercury and Carbon Aerosol Emissions
Area source emissions are introduced into the model in the lowest layer. Point source
emissions are introduced into model layer 2 to account for the effective stack height of the
point source type in question. Effective stack height is the actual stack height plus the
estimated plume rise. The layer of emission is inconsequential during the daytime when
complete vertical mixing is imposed throughout the 4 layers. At night, since there is no
vertical mixing, area source emissions to layer 1 are subject to dry deposition while point
source emissions to layer 2 are not. Large industrial emission sources and sources with very-
hot stack emissions tend to have a larger plume rise, and their effective stack heights might
actually be higher than the top of layer 2. However, since the layers of the pollutant puffs
remain vertically aligned during advection, the only significant process effected by the layer
of emission is nighttime dry deposition.
Global-scale natural emissions, recycled anthropogenic emissions and primary
anthropogenic emissions from outside the continental U.S. are accounted for by superimposing
a background atmospheric concentration of Hg° gas of 1.6 ng/m3. This use of a constant
background concentration is the same technique used by modelers in Europe [6], The same
model parameterizations used to simulate the deposition of Hg° from explicit anthropogenic
emissions are also used to simulate the deposition of Hg° from this constant background
concentration throughout the entire 3-dimensional model domain.
-------
Total carbon air concentrations have been shown to be highly correlated with sulfur
dioxide (S02) air concentrations from minor sources [11], and the emissions of total carbon
and S02 from minor point sources are assumed to be correlated as well, since both pollutants
result from the combustion of fossil fuel. A 35% proportionality constant for total carbon air
concentrations versus S02 air concentrations has been estimated [11]. For this study, total
carbon aerosol emissions were estimated using this 35% proportionality constant and S02
emissions data for minor sources obtained by the National Acidic Precipitation Assessment
Program (NAPAP). Many of these SO, emissions data had been previously analyzed for use
by the Regional Acid Deposition Model (RADM). For the portion of the RELMAP mercury
model domain not covered by the RADM domain, state by state totals of S02 emissions were
apportioned to the county level on the basis of weekday vehicle-miles-traveled data since
recent air measurement studies have indicated that aerosol elemental carbon can be attributed
mainly to transportation source types [12]. The county level data were then apportioned by
area to the individual RELMAP grid cells. Total carbon soot was assumed to be emitted into
the lowest layer of the model.
Ozone Concentration
Hourly 03 concentration data were obtained from EPA's Aerometric Information
Retrieval System (AIRS). Any observations of 03 concentration below 20 ppb were treated as
missing. For each RELMAP grid cell, the 03 concentrations were computed for the two
midday time steps by using the mean concentration value during two corresponding time
periods (1000-1300 and 1400-1600 local time). The mean of these two midday values was
-------
10
used to estimate the 03 concentration for the time steps after 1600 local time and before 1000
local time the next morning. This previous-day average was used at night since ground-level
03 data are not valid for the levels aloft, where the wet removal of Hg° was assumed to be
occurring. Finally, an objective interpolation scheme was used to produce complete 03
concentration grids for each time step, with a minimum value of 20 ppb imposed to represent
an average global background value.
Lagrangian Transport and Deposition
In the RELMAP model, each pollutant puff begins with an initial mass equal to the
total emission rate of all sources in the source cell multiplied by the model time-step length.
Emission rates for each source cell were defined from input data and a time step of three
hours was used. The initial horizontal area of each puff was set to 1200 km2, instead of the
standard initial size of 2500 km2, in order to accommodate the finer grid resolution used for
the mercury modeling study. However, the standard horizontal expansion rate of 339 km2 per
hour was not changed.
Although each puff is defined with four separate vertical layers, all of these layers
are advected through the model cell array by the same wind velocity field. Thus, the layers
of each puff always remain vertically stacked. Wind field initialization data for a National
Weather Service prognostic model, the Nested Grid Model (NGM), were obtained from the
NOAA Atmospheric Research Laboratory for the entire year of 1989. Wind analyses for the
NGM vertical model layer near 1000 m above ground level (a.g.l.) were used to define puff
advection throughout the simulation, except for the months of January, February and
-------
11
December, when winds for a layer near 500 m a.g.l. were used to better represent the more
shallow mixed layer heights of the winter season.
Pollutant mass was removed from each puff by the processes of wet deposition, dry
deposition and diffusive air exchange between the surface-based mixed layer and the free
atmosphere. Observed precipitation data for the entire year of 1989 were obtained from the
National Climatic Data Center and used to estimate wet deposition of all mercury species.
Wet and dry deposition totals were accumulated and average surface-level concentrations were
calculated on a monthly basis for every model cell, except for the cells in the far southwest
and eastern corners of the model domain for which NGM-derived wind data were not
available. When the mass of pollutant in a puff declined through deposition or vertical
diffusion to, a prescribed minimum value, or when a puff moved out of the model grid, the
puff and its pollutant load were no longer tracked. The amount of pollutant in the terminated
puff was taken into account in monthly mass balance calculations so that the integrity of the
model simulation was assured.
Vertical Exchange of Mass with the Free Atmosphere
Due to the long atmospheric lifetime of mercury, the RELMAP was adapted to allow a
treatment of the exchange of mass between the surface-based mixed layer and the free
atmosphere above. As an intuitive approximation, a pollutant depletion rate of 5 percent per
3-hour time step was chosen to represent this diffusive mass exchange. When compounded
over a 24-hour period, this depletion rate removes 33.6% of an inert, non-depositing pollutant.
Since all three of the modeled mercury species deposit to the surface to some degree, their
-------
effective diffusion rate out of the top of the model is somewhat less than 33.6% per day.
MODEL PARAMETERIZATIONS
Chemical Transformation and Wet Deposition
In the mercury version of RELMAP, HgP and total carbon are each modeled explicitly
as inert pollutant species. The RELMAP was originally developed to simulate sulfur
deposition, and the algorithm for transformation of sulfur dioxide to sulfate was independent
of wet deposition. For gaseous mercury, however, the situation is more complex. Since there
are no gaseous chemical reactions of mercury in the atmosphere which appear to be
significant [6], mercury chemistry is treated only in the aqueous medium. Hg° has a very low
solubility in water, while oxidized forms of mercury and particulate mercury readily find their
way into the aqueous medium through dissolution and particle scavenging, respectively.
Swedish measurements of large north-to-south gradients of mercury concentration in rainwater
without corresponding gradients of atmospheric mercury concentration suggest the presence of
physical and chemical interactions with other pollutants in the precipitation scavenging
process [8]. Aqueous chemical reactions incorporated into the mercury version of RELMAP
are based on research efforts in Sweden [7,10,13,14,15] and Canada [16,17],
Unlike other pollutants that have been modeled with RELMAP, mercury has wet
deposition and chemical transformation processes that are interdependent. A combined
transformation/wet-removal scheme is used [6]. In this scheme, the following aqueous
chemical processes are modeled when and where precipitation is present.
-------
13
1) oxidation of dissolved Hg° by 03 yielding Hg2+
2) catalytic reduction of this Hg2+ by ubiquitous sulfite ions
3) adsorption of Hg2" onto carbon soot particles suspended in the aqueous medium
These three simultaneous reactions are considered in the formulation of a variable wet
scavenging ratio for elemental mercury gas as follows:
W
-------
03
gas
spring and autumn, and 0.317 in summer as calculated from [19]). csoot is obtained from the
simulated air concentration of total carbon aerosol using a scavenging ratio of 5.0 x 105.
The model used in [6] defined one-layer cylindrical puffs, and the Hg° scavenging
layer was defined as the entire vertical extent of the model. The RELMAP defines 4-layer
puffs to allow special treatment of surface-layer and nocturnal inversion-layer processes. Due
to the low solubility of Hg° in water, it was assumed for this study that the scavenging
process outlined above would only take place effectively in the cloud regime where the water
droplet surface-area to volume ratio is highest, and not in falling raindrops. Thus the Hg° wet
scavenging process was applied only in the top two layers on RELMAP, which extends from
200 meters above the surface to the model top.
For the modeling study described in [6], the wet deposition of Hg2* was treated
separately from that of Hg°. Obviously, any Hg2+ dissolved into the water droplet directly
from the air could affect the reduction-oxidation balance between the total concentration of
Hg° and Hg2+ in the droplet. Since the solubility and scavenging ratio for Hg2" is much larger
than that for Hg°, and since air concentrations of Hg° are typically much larger than those of
Hg2+, separate treatment of Hg2+ wet deposition is used in the RELMAP also. Thus, as was
done in [6], the catalytic reduction of Hg2+ is only considered as a moderating factor for the
oxidation of dissolved Hg°.
Although Hg2+ is recognized as a reactive species in aqueous phase redox reactions, it
is modeled in the RELMAP as an inert species similar to HgP and total carbon soot. Given
the rapid rate at which the aqueous Hg2+ reduction reaction is believed to occur in the
-------
15
presence of sulfite, it is possible that an interactive cloud-water chemical mechanism could
produce significant conversion of scavenged Hg2" to Hg°, with possible release of that Hg°
into the gaseous medium. It is anticipated that future improvements to the RELMAP will
include a more sophisticated treatment of this interacting aqueous chemistry.
Wet deposition of Hg2+, HgP and total carbon soot in the mercury version of the
RELMAP are modeled with the same scavenging ratios used in [6]. These fixed scavenging
ratios were applied to all four layers of the RELMAP in the calculation of pollutant mass
scavenging by precipitation.
Dry Deposition
Recent experimental data indicate that Hg° vapor does not exhibit a net dry
depositional flux to vegetation until the atmospheric concentration exceeds a compensation
point of 10 ng/m3 or more, where emission and deposition forces are equal [20]. This
compensation point is apparently dependent on the surface or vegetation type and represents a
balance between emission from humic soils and dry deposition to leaf surfaces [20,21], Since
the emission of mercury from soils is accounted for by a global-scale ambient concentration
and not an actual emission of Hg°, for consistency, there is no explicit simulation of the dry
deposition of Hg°.
During simulated daylight hours, dry deposition velocity tables previously developed
based on HN03 data [22,23] are used to estimate Hg2+ dry depostion. This use of HN03 data
to estimate Hg2+ dry deposition is similar to the technique used in [6], However, in this case,
the dry deposition velocity of Hg2+ is allowed to vary based on surface characteristics and
-------
16
meteorological conditions. The dry deposition velocity tables provide season-dependent
values for 11 land-use types under six different Pasquill stability categories. The indicated
values for Hg2+ dry deposition velocity vary between 0.13 cm/s and 4.83 ctn/s, For nighttime,
a value of 0.3 cm/s is used for all grid cells since the RELMAP does not have the capability
of applying land-use dependent dry deposition at night. Since the nighttime dry deposition is
applied only to the lowest layer of the model and no vertical mixing is assumed for nighttime
hours, all Hg2+ would be quickly depleted from the lowest model layer by larger dry
deposition velocities.
In [6], a dry deposition velocity of 0.2 cm/s was used for Hgp at all times and
locations. Dry deposition of HgP seems to be dependent on foliar activity [24], In the
RELMAP mercury model, the daytime dry deposition velocities for HgP are calculated using a
FORTRAN subroutine developed by the California Air Resources Board (CARB) [25]. A
particle density of 2.0 g/cm3 and diameter of 0.3 |im is assumed. A wind speed of 10 m/s is
assumed for Pasquill stability category A, 5 m/s for categories B and C, 2.5 m/s for categories
D and E, and 1 m/s for category F. Table 1 shows the roughness lengths (Zo) used for each
land-use category and season. At night, all cells use 0.02 cm/s as the dry deposition velocity
for Hgp. A value of 0.003 cm/s has been suggested for non-vegetated land [24], but since the
RELMAP can not model land-use dependent dry deposition at night, the value of 0.02 cm/s is
used for these cells by necessity.
For total carbon aerosol, daytime dry deposition velocities are also calculated using the
CARB subroutine. A particle density of 1.0 g/cm3 and diameter of 1.0 (im is assumed for
these aerosol soot particles. For nighttime, a dry deposition velocity of 0.07 cm/s is used for
all seasons and land-use types for total carbon aerosol.
-------
17
Mercury Air Emissions
The atmospheric mercury emission inventory used for this modeling study was
developed by the EPA Office of Air Quality Planning and Standards (OAQPS) and is
described in Volume II of the Mercury Study Report to Congress which is due to be
submitted to the U.S. Congress by OAQPS in late 1995. Data from this inventory were used
to compile estimates of the mercury emissions from 7 major stationary source types (point
sources) and from a group of minor source types for which individual emission site locations
were not available (area sources). The 8 emission source types resolved for input to the
RELMAP mercury model were; electric-utility fossil-fueled boilers, non-utility fossil-fueled
boilers, municipal solid waste combustion, medical waste incineration, chlor-alkali factories,
non-ferrous metal smelting, all other point sources, and area sources. For each of the source-
specific point source types, a base-case estimate of the mercury emission speciation profile
was adopted from [6] to represent the best engineering estimate of the chemical and physical
forms of the Hg emissions. Since there remains considerable uncertainty about the actual
chemical and physical forms of mercury emitted and about the importance of Hg2" vapor
attachment to particles, an alternate emission speciation was also developed to measure the
sensitivity of the RELMAP results to these uncertainties. The alternate speciation represents a
complete transfer of all Hg2+ vapor emissions to a particle-bound form. The alternate
speciation is not intended to serve as a credible alternative to the base speciation, but only as
a perturbation for model sensitivity testing.
The base-case and alternate mercury emission speciation profiles for the 7 point source
types are shown in Table 2. The area sources are modeled as having emissions of Hg° only.
-------
18
The area source data used for input to the RELMAP model simulation include estimates of
the emission of mercury from latex paints, which were not included in the assessment of
emissions described in Volume II of the Mercury Study Report to Congress. These latex
paint emission estimates for the continental U.S. total 4 metric tons per year. The compiled
nationwide patterns of Hg°, Hg2~ and HgP emissions from all point and area source types
using the base-case emission speciation profiles are illustrated in Figures 1, 2 and 3.
respectively.
Table 3 shows the results of applying the base-case emission speciation profiles to the
8 mercury emission source types resolved by the RELMAP mercury model. These emissions
are for the lower 48 states only. Thus, some of the specific-source totals differ slightly from
the national totals shown in Volume II of the Mercury Study Report to Congress. This
analysis indicates that of the total anthropogenic emissions from the lower 48 states, 26% is
from medical waste incineration, 22% is from municipal waste combustion, 22% is from
electric utility boilers, 13% is from fossil fuel combustion other than that by electric utilities,
4% is from non-ferrous metal smelting, and 3% is from chlor-alkali factories. The
atmospheric emissions from all other point source types represent 7% of the total
anthropogenic emission in the lower 48 states, while area sources represent 3% of this total.
As a whole, large-scale fossil fuel combustion represents about 35% of the total anthropogenic
mercury emissions to the atmosphere in the lower 48 states. Because of differences in the
behavior of the various chemical and physical forms of mercury in the atmosphere, it is
important to determine their relative contributions to the total mass of mercury emitted. Table
3 indicates that, based on the base-case emission speciation profiles, about 41% of all
anthropogenic mercury emitted to the air in the continental United States is in the form of Hg°
-------
vapor, 41% is in the form of Hg2+ vapor, and 17% is emitted as HgP.
INTERPRETIVE ANALYSIS OF THE RESULTS
Mass Balances of Mercury within the Model Domain
The general mass balance of Hg°, Hg2+ and HgP from the RELMAP simulation using
the base-case emission speciation is shown in Table 4. The mass-balance accounting for the
simulation of the year 1989 shows a total of 223.8 metric tons of mercury emitted to the
atmosphere from anthropogenic sources in the continental United States. The base-case
simulation accounting also shows that 77.9 metric tons of these anthropogenic mercury
emissions deposited to the surface within the model domain and that 0.6 metric tons remained
in the air within the model domain at the end of the simulation. The balance, about 145.3
metric tons, was transported outside the model domain and conceptually became part of the
global atmospheric background. This estimated annual contribution of continental U.S.
anthropogenic sources to the global mercury cycle amounts to about 2% of the total global
reservoir. A rough calculation of the total mass of the global atmospheric mercury reservoir
based on a sea-level concentration of 1.6 ng/m3 (1.6 g/kmJ), a total atmospheric scale height
of 9 km and a global surface area of 5.1 x 108 km2 yields a figure of 7.3 x 109 g, or 7300
metric tons. The simulation mass balance also indicates that 33.0 metric tons of mercury is
deposited annually within the model domain from this global atmospheric reservoir.
As shown in Table 5, the alternate case emission speciation profiles result in a
noticeably different mass balance. By assuming that all of the Hg2+ emitted quickly becomes
-------
20
bound to existing particles, the total deposition of anthropogenic Hg to the surface is reduced
by about 40%. This is primarily due to the fact that the modeled dry deposition velocity of
particulate matter in the assumed size range is much smaller than the modeled dry deposition
velocity for Hg2+ vapor. A particle diameter of 0.3 (j.m was assumed to be representative of
continental air masses with moderate urban influences. Less efficient wet scavenging of
particulate versus gaseous Hg2+ also contributes to the lower total deposition using the
alternate emission speciation profiles. While the assumption of total Hg2+ attachment to
particles is only intended as a bounding exercise, the results show the importance of an
accurate determination of the mass of HgP emitted and formed during transport.
Of the total simulated anthropogenic mercury deposited to the surface in the model
domain, 80% is estimated to come from Hg2+ emissions, 18% from HgP emissions and 2%
from Hg° emissions when the base-case emission speciation profiles are used. When the
deposition of Hg° from the global background is considered in addition to anthropogenic
sources in the lower 48 states, the species fractions become 56% Hg2", 31% Hg° and 13%
HgP. The vast majority of mercury already in the global atmosphere is in the form of Hg°
and, in general, the anthropogenic Hg° emissions do not greatly increase the existing Hg°
concentration. Although Hg° is removed from the atmosphere very slowly, the global
background reservoir is large and extraction of mercury from it is significant in terms of the
total deposition. It should be reiterated that dry deposition of Hg° vapor appears to be
significant only at very high concentrations and has not been included in the RELMAP
simulations. Wet deposition is the only major pathway for removal of Hg° from the
atmosphere. This removal pathway simulated by the RELMAP involves oxidation of mercury
by 03 in an aqueous solution; thus, the Hg° that is extracted from the atmosphere by the
-------
modeled precipitation process would actually be deposited primarily in the form of Hg2\
Results from the RELMAP simulation show that 98% of the deposited anthropogenic
mercury was emitted in the form of Hg2+ or HgP. Thus, a strong argument can be made that
the combined Hg2+ and HgP component of all anthropogenic mercury emissions can be used as
an indicator of eventual deposition of those emissions to the lower 48 states and surrounding
areas. The emission inventory and base-case chemical/physical speciation profiles indicate
that of all combined Hg2+ and HgP emissions, about 36% is from medical waste incineration,
30% is from Municipal Waste Combustion, 18% is from electric utility boilers, 11% is from
combustion of fossil fuel other than by electric utilities, 1% is from Chlor-alkali factories, 1%
is from non-ferrous metal smelting, and 2% is from all other sources.
Qualitative Description of Mercury Concentration Results-
Annual average surface-level concentration fields for Hg°, Hg2\ and HgP have been
obtained from the RELMAP simulation of 1989. Figure 4 shows the annual average Hg°
concentration at ground level from anthropogenic sources obtained by using the base-case
emission speciation. It shows that anthropogenic Hg° concentrations remain less than 0.1
ng/m3 over most of the investigation area. The areas where the average anthropogenic Hg°
concentrations exceed 0.1 ng/m3 are mostly confined to the highly industrialized locations in
the eastern Midwest and the Northeast regions. Compared to the estimated average global
background concentration of 1.6 ng/m3, this 0.1 ng/m3 elevation of Hg° concentration by
anthropogenic emissions is rather small.
Figure 5 shows annual average Hg2+ air concentrations, also using the base case
-------
22
speciation. These values are significantly lower than for anthropogenic Hg°, and there are
some new areas of maximum concentration. The higher concentration areas have values from
0.05 to just over 0.1 ng/m3 and are mostly confined to Florida, the Midwest and the Northeast
urban corridor. The background atmospheric mercury loading is assumed to be completely in
the elemental form, so there is no background contribution to the Hg2+ concentrations. In
most areas, the anthropogenic component of the Hg° concentrations (Figure 4) are at least 3 to
5 times higher than the Hg2+ concentrations (Figure 5). Since the national total of Hg2"
emissions are about equal to those for Hg°, these much lower average annual Hg2+
concentrations must be due to more rapid depletion from wet and dry deposition.
The RELMAP-simulated Hg° and Hg2+ air concentrations taken together with the
assumed background Hg° concentration of 1.6 ng/m3 agree well with observations of vapor-
phase Hg air concentration in Minnesota [26], in Vermont [27] and in Wisconsin [28]. These
observations showed that annual average vapor-phase Hg concentrations were near the levels
found over other remote locations in the northern hemisphere, from 1.6 to 2.0 ng/m3.
Measurements taken for a two-week period at three sites in Broward County, Florida, [29]
show slightly elevated vapor-phase Hg air concentrations for two of those sites downwind of
industrial activities. These two sites had average vapor-phase Hg air concentrations of 3.3
and 2.8 ng/m3. The RELMAP simulation results for the Fort Lauderdale area show only
about a 0.2 ng/m3 elevation of the annual average vapor-phase Hg (Hg° plus Hg2+)
concentration over the 1.6 ng/mJ background value assumed. However, the measurements in
Broward County did not extend for a significant portion of the year and there was no
discrimination between Hg° and Hg21" forms. The third site for their observations had an
average vapor-phase air concentration of 1.8 ng/mJ, exactly what the RELMAP simulation
-------
23
suggests. Obviously, a more comprehensive air monitoring program is required before an
evaluation of the RELMAP results for Florida can be performed.
HgP emissions are thought to be a small fraction of the total Hg emissions for most
source types. For the base-case emission speciation, 20% is the largest particulate fraction of
mercury emissions for any source type. Figure 6 indicates that the simulated annual average
HgP concentrations are even lower than those for Hg2+ vapor. The maximum annual average
values are around 50-100 pg/m3 (0.05-0.1 ng/m3) in the urban centers of the Northeast. A
study in urban Detroit found instantaneous HgP concentrations during March of 1992 of over
1 ng/m3 and average concentrations over an 18-day period of 94 pg/m3 [30]. The RELMAP
simulation suggests an annual average HgP concentration in the Detroit area of about 50
pg/m3. However, given the 40-km horizontal scale of the RELMAP computational grid, one
can not expect the simulation to reflect these extreme local-scale measurement results.
Average HgP concentrations of between 34 and 51 pg/m3 were measured in Broward County,
Florida, at three sites from 25 August to 7 September of 1993 [29]. The RELMAP
simulation results agree well with these observations around the city of Fort Lauderdale.
Researchers have found annual average HgP air concentrations of 10.5 pg/m3 in Pellston,
Michigan, 22.4 pg/m3 in South Haven, Michigan, and 21.9 pg/m3 in Ann Arbor, Michigan,
from April 1993 to April 1994, and 11.2 pg/m3 in Underhill, Vermont, for the year of 1993
[30]. The RELMAP simulation results agree quite well with these observations also.
Table 6 shows a percentile analysis of the simulated concentration results from the
RELMAP grid cells within the lower 48 United States. This table shows that the Hg°
concentrations never exceeded the assumed background level of 1.6 ng/m3 by a large relative
amount. It also shows that Hg2+ and to a lesser degree HgP air concentrations were highly
-------
24
elevated in only a few grid cells. There is an order of magnitude difference in the Hg2+
concentrations at the 90th percentile level and the maximum level, with a factor of 5
difference for HgP.
For the alternate emission speciation tests, the Hg2+ vapor emission fraction was
redistributed to the HgP fraction, simulating the complete attachment of Hg2~ vapor to ambient
particulate matter. The annual average Hg° and HgP concentration fields from this test (not
shown) indicate that, as one would expect, there was no change to the Hg° results, but the HgP
concentrations are increased nearly to the level of anthropogenic Hg°, with maximum
concentrations over 100 pg/m3 (0.1 ng/m3) for the larger urban areas of Florida and the
Midwest and for nearly all of the Northeast urban corridor.
Description of Mercury Wet Deposition Simulation Results
Figure 7 shows the simulated wet deposition of Hg° from continental U.S.
anthropogenic sources during 1989 using the base emission speciation factors. Figure 8 shows
the simulated annual wet deposition of Hg° assuming only a non-depleting global background
concentration of 1.6 ng/m3. Both of these wet deposition results are influenced by 03 and
carbon soot concentrations due to the chemical transformations modeled by the RELMAP.
Emission patterns influence the primary anthropogenic Hg° wet deposition pattern and it is
obvious that total annual precipitation is a strong positive factor in wet deposition from the
global background concentration. It is widely accepted that deposition of Hg° occurs on
continental and global scales, and the RELMAP simulation shows areas of significant Hg° wet
deposition occurring in remote areas. The base-case wet deposition results for Hg2" vapor
-------
shown in Figure 9 indicate maximum deposition areas that are highly coincident to the
emission source areas. There are many model cells in urban areas with wet deposition totals
of Hg2+ vapor over 20 |ig/m2, while most of the cells in the non-urban areas have wet
depositions of less than 5 p.g/m2. This indicates that Hg2+ vapor wet deposits more on the
local scale, and not on regional or global scales and that its wet removal from the atmosphere
is much more rapid than for Hg°. This is an expected result due to the higher water solubility
of most mercuric salts compared to mercury in the elemental form. Figure 10 shows that, for
the base-case emission speciation, the maximum simulated wet deposition of HgP is about half
of that for Hg2" vapor. This is partly due to differences in the total mass of HgP emitted
compared to Hg2~; but, it is also due to the less efficient wet scavenging of HgP. The areas of
HgP wet deposition are also more widespread than for Hg2" due to the slower depletion of HgP
and greater opportunity for long-range transport.
The total wet deposition of mercury emitted in all three forms is shown in Figure 11.
This illustration shows significant wet deposition of mercury over most of the eastern half of
the U.S. For the simulated year of 1989, nearly the entire eastern half of the nation has a wet
deposition total of over 5 jag/m2 and values exceed 20 p.g/m2 over much of the urban
northeast U.S. In fact, the largest simulated wet deposition exceeded 100 |ig/m2 in the grid
cell containing New York City. We have not designed Figure 11 to highlight these extreme
maximum wet deposition results because at this time such high wet deposition rates for total
mercury can not be substantiated by observations. In the RELMAP simulation, the most
impacted areas are being subjected to wet deposition of mercury mainly from emissions of
Hg2+ vapor. We suspect that the RELMAP model for mercury may still be significantly
incomplete, and that other chemical and/or physical transformations may occur which
-------
26
moderate the wet deposition of Hg2+ vapor and possibly HgP.
Measurements of mercury wet deposition at three locations in northeastern Minnesota
during 1989 indicated annual wet deposition rates of 6.5 p.g/m2 at Duluth, 13.5 ug/m2 at
Marcell and 41.9 |ag/m2 at Ely [31]. A later study measuring annual wet deposition of Hg
during 1990, 1991 and 1992 at Ely, Duluth and seven other sites in Minnesota, upper
Michigan and northeastern North Dakota found all annual wet deposition totals to be within
the range of 3.8 to 9.7 (ig/'m2 [32]. Measurements at Little Rock Lake, in northern
Wisconsin, of Hg in snow during February and March, 1989, and in rain from May to
August, 1989, have been used to estimate annual Hg depositions in rain and snow of 4.5 and
2.3 p.g/m2, respectively [26]. This suggests a total annual Hg wet deposition of 6.8 ^g/m2 at
Little Rock Lake. Measurements at Presque Isle, also in northern Wisconsin, from 1993 to
1994 suggested a wet deposition rate for total Hg of 5.2 pg/m2/yr [28], somewhat less than
the measurements at Little Rock Lake. The extremely heavy rainfall during the summer of
1993 in the mid-west states to the south and west of Presque Isle may be responsible for the
lower wet deposition. The RELMAP simulation results for 1989 indicate 2 to 10 pg/m2 wet
deposition of total Hg over most of the area represented by these studies, the major exception
being the Minneapolis-Saint Paul metropolitan area where the RELMAP indicates over 20
jig/m2.
There were also some Hg wet deposition measurement programs conducted during the
early 1990's in somewhat less remote sites in Michigan and Vermont. Observations during
two years of event precipitation sampling at three sites in Michigan show evidence for a
north-to-south gradient in Hg wet deposition [33]. From March 1992 to March 1993, the
total Hg wet deposition observed at South Haven, in southwest Michigan, was 9.45 |ig/m2. At
-------
Pellston, in the northern part of the lower peninsula of Michigan, the wet deposition was 5.79
ug/m2. At Dexter, in southeast Michigan about 100 km west of Detroit, the wet deposition
was 8.66 jig/m2. From March 1993 to March 1994, wet deposition at South Haven was 12.67
(ag/m2, significantly higher than for the previous year, while measurements at Pellston and
Dexter remained about constant at 5.54 and 9.11 (.ig/m2, respectively. The higher second-
year wet deposition at South Haven has been attributed to increased precipitation rate, and
measurements at Underhill, Vermont, by [27] are cited as further evidence of the importance
of precipitation amount [33], From December 1992 to December 1993, the average volume-
weighted Hg concentration at Underhill (8.3 ng/L) was similar to that observed at Pellston
(7.9 ng/L). However, with more precipitation during that period, the total Hg wet deposition
at Underhill was 9.26 jag/nr, significantly higher than at Pellston. The RELMAP simulation
for 1989 shows 5-10 fig/m2 wet deposition of total Hg at the Pellston site, which agrees well
with the 1992 to 1994 observations there. At Underhill, the RELMAP simulation indicates
10-20 p.g/m2 wet deposition for 1989 which is slightly larger than the observation in 1993.
At the South Haven and Dexter sites, the RELMAP appears to be estimating nearly 20 (ig/m2
wet deposition of total Hg for 1989 which is significantly larger than the measurements of
1992 to 1994.
The very large total Hg wet deposition values (>50 p.g/m2) from the RELMAP
simulation for some of the larger urban centers in the Ohio Valley and Northeast regions can
not be evaluated thoroughly due to a lack of long-term precipitation event sampling at those
locations. Precipitation event sampling was performed from 19 August to 7 September of
1993 at 4 sites in Broward County, Florida, in and around the city of Fort Lauderdale [29],
During the 20-day sampling period, total Hg mean concentrations in precipitation were 35, 57,
-------
40 and 46 ng/L at the 4 sites. Given the average annual precipitation of 150 cm per year
typical of that area, the resulting annual wet deposition estimates at these 4 sites are 52.5,
85.5, 60 and 69 jig/nr. Since most of the annual rainfall in Broward County occurs in warm
tropical conditions of the March to October wet season, this extrapolation from 20 days
during the wet season to an annual estimate may be considered reasonable. However,
additional urban measurement studies will certainly be required to allow any credible
evaluation of RELMAP wet deposition results in heavily populated, industrialized areas.
Figure 12 shows the wet deposition of total mercury from the alternate emission
speciation and offers some measure of the sensitivity of the RELMAP simulation results to
the emission speciation estimates used. This figure shows basically the same large-scale
pattern as for the base-case emissions, but in general the amount of wet deposition is
increased. Since the wet scavenging ratio for HgP is less than one-third of that for Hg2*, we
would expect the wet deposition to be reduced when using the alternate emission speciation
profiles which reallocate all Hg2+ vapor emissions to the HgP form. This unexpected increase
is the result of an interaction of the wet deposition processes with those for dry deposition.
The alternate emission speciation profiles result in greatly reduced dry deposition compared to
the base-case speciation. This results in much slower depletion of mercury by dry deposition
and provides a greater opportunity for wet deposition to occur.
The percentile analysis of the wet deposition simulation results in Table 7 shows that,
for the base-case emission speciation, 50 percent of the land area of the continental U.S. has
an annual wet deposition of total mercury of more than about 3.4 }J,g/m2, and 10 percent of
the area has more than 16 jag/m2. However, due to rapid wet deposition of Hg2" and HgP
there are select areas where wet deposition may be significantly higher. For the area of the
-------
29
U.S. east of 90 degrees west longitude, the 50th and 90th percentile levels for simulated
annual total mercury wet deposition are about 12 (ig/m2 and 25 |ig/m2, respectively.
Qualitative Description of Mercury Dry Deposition Results
As indicated previously, the RELMAP mercury model assumes that Hg° is not
effectively dry deposited. The percentile analysis of the simulated dry deposition using the
base-case emission speciation profiles shown in Table 8 indicates the strong local dry
deposition of Hg2+ vapor as parameterized in the RELMAP mercury model. There is
considerable uncertainty regarding the dry deposition velocity of Hg2" and we have little
confidence in the extremely high local depositions indicated from the simulation.
Figure 13 shows the simulated annual dry deposition patterns for Hg2+ using the base
emission speciation. Dry deposition of Hg2+ appears to occur primarily on the local scale,
within one or two grid cells (40 to 80 km) from the source, much like the Hg2+ wet
deposition. The magnitude of the dry deposition of Hg2+ is similar to that for wet deposition,
with urban areas showing values in excess of 20 |J.g/m2. As was the case for wet deposition,
dry deposition of Hg2+ vapor in heavily populated urban centers is relatively high, exceeding
100 |ag/m2 in the model grid cell containing New York City. Again, it must be stressed that
dry deposition of Hg2+ vapor is not well understood. We have used nitric acid vapor as a
surrogate for Hg2+ vapor. The authors have been unable to find observations of Hg2" dry
deposition to compare with the RELMAP simulation results. Dry deposition rates for vapor-
phase Hg have been estimated from vertical eddy flux calculations at a single site, but these
calculations estimate the combined effects of both Hg° and Hg2+ vapors [21]. The relatively
-------
30
high solubility and reactivity of Hg2+ compounds suggest that dry deposition of total vapor-
phase Hg may be strongly driven by the Hg2+ component.
Figure 14 shows the simulated annual dry deposition totals for HgP using the base
emission speciation. For this regional-scale modeling study, the dry deposition velocity
estimates for HgP were made based on the assumption that the particulate mass was
concentrated around a 0.3 (im diameter size. The patterns show less intense local dry
deposition of HgP than for Hg2', but the dry deposition still appears to occur primarily within
a few hundred km of the source areas. This slower dry deposition combined with relatively
smaller quantities of HgP emission results in maximum dry deposition values for HgP of only
around 0.5 jig/m2. In urban areas where larger particle sizes are more prevalent, these
estimates of HgP dry deposition are probably too low. However, the RELMAP could treat
only one particle size and could not separately model large-particle deposition on the urban
scale.
Figures 15 and 16 show the simulated annual dry deposition for total mercury in all
forms using the base-case and alternate emission speciations, respectively. A comparison of
these figures clearly demonstrates the sensitivity of the modeling results to changes in the
emission speciation profiles. Dry deposition is not a major pathway for removal of the
atmospheric mercury burden when the alternate emission speciation profiles are employed.
This result indicates that dry deposition is much less important when greater transfer of Hg2"
to HgP is occurring through enhanced particle adsorption or condensation. Thus, it is very
important that our understanding of the physical transformations of Hg in the atmosphere be
complete and accurate.
-------
31
DISCUSSION OF IMPORTANT MODELING UNCERTAINTIES
Transport
The RELMAP model top is defined to be the maximum vertical extent of the
convectively driven mixed layer. Monthly values defined from mixed-laver-height
climatology are rough estimates of a meteorological phenomenon that may not exist in many
situations. Even when a surface-based mixed layer is well defined, pollutants that persist in
the atmosphere for long periods of time are certain to mix to some degree into the free
troposphere above the mixed-layer top. Extremely persistent compounds such as
chlorofluorocarbons (CFCs) are even known to diffuse into the stratosphere. Elemental
mercury deposits relatively slowly through precipitation processes due to its low water
solubility, and its dry deposition appears to be minimal since it is in vapor form under normal
atmospheric conditions. Thus, the diffusion of elemental mercury into the atmosphere above
the mixed layer is certainly an important transport pathway.
Since the RELMAP only simulates the flux of pollutant through the top of the mixed
layer due to turbulence and not that due to large-scale vertical motions, the use of horizontally
divergent or convergent wind fields to define the motion of the pollutant puffs can sometimes
result in unrealistic instantaneous concentration fields. Horizontally convergent winds will
tend to concentrate puffs at the point of convergence, resulting in artificially high modeled
concentrations when the effects of the puffs are summed together. Ordinarily, horizontal
convergence in the surface-based mixed layer would push the mixed-layer top higher into the
atmosphere. This higher mixed-layer top would compensate for the greater pollutant mass
-------
loading per unit area from the converging puffs. The RELMAP, with its monthly definition
of the mixed layer height, was not designed to provide instantaneous realizations of pollutant
concentration fields. Rather, it was designed for seasonal and annual simulations where the
effects of convergent and divergent wind fields can balance one another on the whole.
Nonetheless, there is some uncertainty as to whether this balance occurs for all situations.
Aqueous Chemistry
The aqueous reduction-oxidation chemistry mechanism in the mercury version of the
RELMAP was applied only to the Hg° dissolved from the air into water droplets. Where
significant concentrations of Hg2+ from emissions exist in air, this Hg2+ could also be
dissolved into the water droplet and inhibit the scavenging of Hg°. The RELMAP results
described above indicate that Hg2+ air concentrations are certainly lower than those for Hg° at
the length scales of the RELMAP grid cells; however, the magnitude of the effect of ambient
Hg2+ air concentrations on the wet scavenging of Hg° is not yet well understood.
Another source of modeling uncertainty in aqueous chemistry relates to the fact that
the aqueous chemical mechanisms in the RELMAP are invoked only when and where
precipitation is known to have occurred. Precipitation fields are mostly defined over land
areas where precipitation observations are routinely available. Significant wet transformation
and removal of mercury may occur over oceanic areas were precipitation observations are not
available. It is also possible that significant aqueous chemistry is occurring in non-
precipitating clouds throughout the entire model domain.
-------
:>:>
Horizontal Diffusion
Since the RELMAP vertical domain covers only the surface-based mixed layer and
vertical wind shear is small when the mixed-layer is well defined, under ideal conditions of
strong vertical mixing, horizontal diffusion is probably handled adequately. When the
surface-based mixed layer is not well defined, vertical gradients in the speed and/or direction
of the wind may be present which can not be represented by the motion of individual
Lagrangian puffs whose layers remain vertically stacked. There are two techniques that might
be used to represent vertical wind shear in the RELMAP: puff splitting and wind-shear-
dependent puff expansion. Puff splitting is computationally expensive due to the large
number of puffs that result from the process, and wind-shear-dependent puff expansion does
not resolve the direction of the shear. The most complete solution to the problem of vertical
wind shear is the use of an Eulerian reference frame for numerical modeling.
Boundary Fluxes of Pollutants
Due to the fact that the RELMAP simulates atmospheric pollutant loading as the
combined effect of a population of discrete Lagrangian puffs originating from inside the
horizontal domain, natural and recycled anthropogenic mercury emissions from the oceans and
land surfaces and fresh anthropogenic emissions from outside the model domain could not be
explicitly modeled. Instead, a background Hg° air concentration of 1.6 ng/m3 was used to
represent these sources of mercury to the atmosphere. Simulated puffs of oceanic and
terrestrial mercury emissions could be generated from all model grid cells, but their effects
-------
34
would be artificially concentrated in the eastern sections of the model domain due to the
prevailing west-to-east wind flow over the continental United States. The puffs that would
impact the western areas would primarily be those originating from farther west in the model
domain, while the eastern areas would be impacted by puffs from all parts of the model
domain. Puffs arriving from outside the model domain can not be simulated by the
RELMAP. The use of an Eulerian-type model structure would allow a more realistic
definition of these lower and lateral boundary fluxes of mercury.
Chemical and Physical Forms of Mercury Air Emissions
The RELMAP modeling results show that 98% of the total mercury deposition to the
surface is from emissions of Hg2~ and HgP. Obviously, an accurate model of atmospheric
mercury deposition requires an accurate determination of the chemical and physical forms of
all air emissions of mercury. The current estimates for these chemical and physical
speciations are based on measurements of hot materials inside exhaust stacks and estimates of
the various effects of cooling and dilution of those exhaust materials after they are released to
the atmosphere. If these estimates of the cooling and dilution effects are inaccurate, then an
accurate model simulation is not possible.
SUMMARY
At this time there is significant uncertainty regarding the chemical and physical forms
of mercury air emissions and their chemical and physical transformations in the atmosphere.
-------
35
This modeling study has relied heavily on the assumptions of a previous Lagragian modeling
study in Europe regarding emission speciation estimates and chemical and physical pathways
for atmospheric mercury deposition [6]. The modeling results from this previous study were
compared to measurements of Hg° and HgP air concentration and wet deposition in northern
Europe, showing that the simulation agreed with measurements to within a factor of 2 in most
cases. While the climate of northern Europe may be quite different from that of many
locations in North America, it is assumed that the predominant chemical and physical
mechanisms for mercury transport, transformation and deposition should be the same for both
regions.
This modeling study of atmospheric mercury emission, transport and deposition
indicates that nearly all areas of the continental United States are subject to atmospheric
deposition of mercury from current anthropogenic activities and that wet deposition of
mercury from the global atmospheric reservoir is a significant part of the total deposition.
The results from the RELMAP simulation of atmospheric mercury agree with actual
measurements of air concentration and deposition within a factor of 2 in nearly all cases. The
RELMAP annual wet deposition estimates of 50 to 100 jag/m2 in some of the more highly
industrialized areas of the United States are quite significant and worthy of concern, but there
are very few measurements with which one can evaluate these results. The RELMAP
simulation results also indicate that the importance of dry versus wet deposition processes may
be dependent on the fraction of mercury that becomes bound to particles before deposition.
Very few direct measurements of the dry deposition of gaseous and particulate Hg have been
made to date. Vertical concentration gradients and eddy flux correlations have been used to
estimate the dry flux of total gaseous Hg, but no discrimination has been made between Hg°
-------
36
and Hg2" forms.
Many of the measurement studies performed up until recent years are now suspected
of having been subject to laboratory contamination upon analysis of their samples. It is only
recently that, by employing ultra-clean laboratory techniques, mercury measurement studies
have been able to accurately assess atmospheric concentrations and deposition quantities of
mercury in near-background conditions. Even now, it is very difficult to obtain an accurate
assessment of the chemical forms of mercury in typical ambient air samples. The RELMAP
air concentration results seem quite plausible, with the vast majority of atmospheric mercury
estimated to be in the Hg° form, but the individual air concentrations of Hg°, Hg2" and HgP
can not be simulated with much confidence until a more complete understanding is established
of all pertinent chemical and physical processes in the atmosphere.
Acknowledgement - The authors would like to thank George Mapp and Jang-Tai Lin for their efforts
in the installation and operation of the RELMAP modeling codes. Their enduring cooperation and
enthusiasm for the project were instrumental in the successful completion of this study.
-------
37
REFERENCES
1. Johnson, W. B., D. E. Wolf and R. L. Mancuso. 1978. Long-term regional patterns
and transfrontier exchanges of airborne sulfur pollution in Europe. Atmos. Environ.
12: 511-527.
2. Bhumralkar, C. M., R. L. Mancuso, D. E. Wolf, R. H. Thuillier, K. D. Nitz and
W. B. Johnson. 1980. Adaptation and application of a long-term air pollution model
ENAMAP-1 to Eastern North America. EPA-600/4-80/039, Final Report, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
3. Johnson, W. B. 1983. Interregional exchanges of air pollution: Model types and
applications. J. Air Pollut. Control Assoc. 33: 563-574.
4. Eder, B. K., D. H. Coventry, T. L. Clark and C. E. Bollinger. 1986. RELMAP:
A regional Lagrangian model of air pollution - users guide. EPA/600/8-86/013,
Project Report, U.S. Environmental Protection Agency, Research Triangle Park, NC.
5. Clark, T. L., P. Blakely and G. Mapp. 1992. Model calculations of the annual
atmospheric deposition of toxic metals to Lake Michigan. Proceedings, 85th Annual
Meeting of the Air and Waste Management Assoc., Kansas City, MO, June 23-37,
1992.
-------
38
6. Petersen, G., A. Iverfeldt and J. Munthe. 1995. Atmospheric mercury species over
Central and Northern Europe. Model calculations and comparison with observations
from the Nordic Air and Precipitation Network for 1987 and 1988. Atmos, Environ.
29:47-68.
7. Munthe, J. 1992. The aqueous oxidation of elemental mercury by ozone. Atmos.
Environ. 26A: 1461-1468.
8. Iverfeldt, A. 1991. Occurrence and turnover of atmospheric mercury over the nordic
countries. Wat. Air Soil Pollut. 56:251-265.
9. Brosset, C. and E. Lord. 1991. Hg in precipitation and ambient air. Anew
scenario. Wat. Air Soil Pollut. 56:493-506.
10. Lindqvist, O., K. Johansson, M. Aastrup, A. Andersson, L. Bringmark, G.
Hovsenius, L. Hakanson, A. Iverfeldt, M. Meili and B. Timm. 1991. Mercury in
the Swedish environment. Transformation and deposition processes. Wat. Air Soil
Pollut. 55:49-63.
11. Penner, J. E., H. Eddleman and T. Novakov. 1993. Towards the development of a
global inventory for black carbon emissions. Atmos. Environ. 27A: 1277-1295.
12.
Keeler, G. J., S. M. Japar, W, W. Brachaczek, R. A. Gorse, Jr., J. M. Norbeck
-------
39
and W. R. Pierson. 1990. The sources of aerosol elemental carbon at Allegheny
Mountain. Atmos. Environ. 24A:2795-2805.
13. Iverfeldt, A. and O. Lindqvist. 1986. Atmospheric oxidation of elemental mercury
by ozone in the aqueous phase. Atmos. Environ. 20:1567-1573.
14. Munthe, J., Z. F. Xiao and O. Lindqvist. 1991. The aqueous reduction of divalent
mercury by sulfite. Wat. Air Soil Pollut. 56:621-630.
15. Munthe, J. and W. McElroy. 1992. Some aqueous reactions of potential importance
in the atmospheric chemistry of mercury. Atmos. Environ. 26A: 553-557.
16. Schroeder, W. H. and R. A. Jackson. 1987. Environmental measurements with an
atmospheric mercury monitor having speciation capabilities. Chemosphere 16:183-199.
17. Schroeder, W. H., G. Yarwood and H. Niki. 1991. Transformation processes
involving mercury species in the atmosphere - Results from a literature survey. Wat.
Air Soil Pollut. 56:653-666.
18. Sanemasa, I. 1975. The solubility of elemental mercury vapor in water. Bull.
Chem. Soc. Japan 48:1795-1798.
19. Seinfeld, J. H. 1986. Atmospheric Chemistry and Physics of Air Pollution, John
-------
40
Wiley and Sons, New York, NY
20. Hanson, P. J., S. E. Lindberg, T. A. Tabberer, J. G, Owens and K. -H. Kim.
1995. Foliar exchange of mercury vapor: Evidence for a compensation point. Wat.
Air Soil Pollut. 80:373-382.
21. Lindberg, S. E., T. P. Meyers, G. E. Taylor, Jr., R. R, Turner and W. H,
Schroeder. 1992. Atmosphere-surface exchange of mercury in a forest: Results of
modeling and gradient approaches. J. Geophys, Res. 97:2519-2528.
22. Walcek, C. J., R. A. Brost, J. S. Chang and M. L. Wesely, 1985. S02, sulfate and
HN03 deposition velocities computed using regional land use and meteorological data.
Atmos. Environ. 20:949-964.
23. Wesely, M. L. 1986. On the parameterization of dry deposition of acidifying
substances for regional models. Interagency Agreement DW89930060-01 to the U.S.
Department of Energy, Internal Report (Nov. 1986), U. S. Environmental Protection
Agency, Research Triangle Park, NC
24. Lindberg, S. E., R. R. Turner, T, P. Meyers, G. E. Taylor, Jr. and W. H.
Schroeder. 1991. Atmospheric concentrations and deposition of Hg to a deciduous
forest at Walker Branch Watershed, Tennessee, USA. Wat. Air Soil Pollut. 56:577-
594.
-------
41
25. California Air Resources Board. 1987. Deposition rate calculations for air toxics
source assessments. Report date September 16, 1987, California Air Resources Board,
Sacramento, CA.
26. Fitzgerald, W. F., R. P. Mason and G. M, Vandal. 1991. Atmospheric cycling and
air-water exchange of mercury over mid-continental lacustrine regions. Wat. Air Soil
Pollut. 56:745-767.
27. Burke, J., M. Hoyer, G. Keeler and T. Scherbatskoy. 1995. Wet deposition of
mercury and ambient mercury concentrations at a site in the Lake Champlain basin.
Wat. Air Soil Pollut. 80:353-362.
28. Lamborg, C. H., W. F. Fitzgerald, G. M. Vandal and K. R. Rolfbus. 1995.
Atmospheric mercury in northern Wisconsin: Sources and species. Wat. Air Soil
Pollut. 80:189-198.
29. Dvonch, J. T., A. F. Vette, G. J. Keeler, G. Evans and R. Stevens. 1995. An
intensive multi-site pilot study investigating atmospheric mercury in Broward County,
Florida. Wat. Air Soil Pollut. 80:169-178.
30.
Keeler, G., G. Glinsorn and N. Pirrone. 1995. Particulate mercury in the
atmosphere: Its significance, transport, transformation and sources. Wat. Air Soil
Pollut. 80:159-168.
-------
42
31. Glass, G. E., J. A. Sorenson K. W. Schmidt, G. R. Rapp, Jr., D. Yap and D.
Fraser. 1991. Mercury deposition and sources for the upper Great Lakes region.
Wat. Air Soil Pollut. 56:235-249.
32. Sorenson, J, A., G. E. Glass and K. W. Schmidt. 1994. Regional patterns of wet
mercury deposition. Environ. Sci. Technol. 28:2025-2032.
33. Hoyer, M., J. Burke and G. Keeler. 1995. Atmospheric sources, transport and
deposition of mercury in Michigan: Two years of event precipitation. Wat. Air Soil
Pollut. 80:199-208.
-------
43
Table 1. Roughness Lengths Used in the Dry Deposition Velocity Calculations
Land-Use Category Roughness Length (meters)
Autumn-Winter Spring-Summer
Urban
0.5
0.5
Agricultural
0.15
0.05
Range
0.12
0.1
Deciduous Forest
0.5
0.5
Coniferous Forest
0.5
0.5
Mixed Forest/Wetland
0.4
0.4
Water
lO'6
10"6
Barren Land
0.1
0.1
Non-forested Wetland
0.2
0.2
Mixed Agricultural/Range
0.135
0.075
Rocky Open Areas
0.1
0.1
-------
44
Table 2. Emission Speciation Profiles for the Point Source Types Defined
Base Speciation (%) Alternate Speciation (%)
Point Source Type
Hg°
Hg2+
Hgp
Hg°
Hg2+
1 'cp
Electric Utility Boilers
50
30
20
50
0
50
Non-utility Fossil Fuel Combustion
50
30
20
50
0
50
Municipal Waste Combustion
20
60
20
20
0
80
Medical Waste Incineration
20
60
20
20
0
80
Non-ferrous Metal Smelting
85
10
5
85
0
15
Chlor-alkali Factories
70
30
0
70
0
30
Other Point Sources
80
10
' 10
80
0
20
-------
45
Table 3. Mercury Emissions Inventory Totaled by Source Type Using Base-Case Emission
Speciation Profiles (metric tons per year)
Source Type Hg° Hg2+ Hgpart Total Hg
Medical Waste Incineration 11.7 35.1 11.7 58.6
Municipal Waste Combustion 10.0 29.9 10.0 49.8
Electric Utility Boilers 24.3 14.6 9.7 48.5
Non-Utility Fossil Fuel 14.3 8.6 5.7 28.5
Non-Ferrous Smelting 7.4 0.9 0.4 8.7
Chlor-alkali Factories 4.6 1.9 0.0 6.5
Other Point Sources 13.0 1.6 1.6 16.2
Area Sources 6.9 0.0 0.0 6.9
Total 92.0 92.6 39.1 223.8
-------
46
Table 4. Modeled Mercury Mass Budget in Metric Tons for 1989
*
Using the Base-Case Emission Speciation Profiles
(all figures rounded to the nearest tenth of a metric ton)
Source/Fate
era
O
Hg2+
Hgpart
Total Hg
Total U.S. anthropogenic emissions
92.0
92.6
39.1
223.8
Mass advected from model domain
90.4
29.9
25.0
145.3
Dry deposited anthropogenic emissions
0.0
39.0
0.6
39.6
Wet deposited anthropogenic emissions
1.2
23.6
13.4
38.3
Remaining in air at end of simulation
0.4
0.1
0.1
0.6
Total deposited anthropogenic emissions
1.2
62.6
14.1
77.9
Deposition from background Hg°
33.0
0.0
0.0
33.0
Mercury deposited from all sources
34.2
62.6
14.1
111.0
-------
47
Table 5. Modeled Mercury Mass Budget in Metric Tons for 1989
Using the Alternate Emission Speciation Profiles
(All figures rounded to the nearest tenth of a metric ton)
Source/Fate
Hg°
Hg2*
Hgpart
Total Hg
Total U.S. anthropogenic emissions
92.0
0.0
131.7
223.8
Mass advected from model domain
90.4
0.0
84.5
174.9
Dry deposited anthropogenic emissions
0.0
0.0
2.1
2.1
Wet deposited anthropogenic emissions
1.2
0.0
44.9
46.1
Remaining in air at end of simulation
0.4
0.0
0.2
0.6
Total deposited anthropogenic emissions
1.2 '
0.0
47.0
48.2
Deposition from background Hg°
33.0
0.0
0.0
33.0
Mercury deposited from all sources
34.2
0.0
47.0
81.3
-------
Table 6. Percentile Analysis of RELMAP Simulated Concentration Results
for the Continental U. S. Using the Base-Case Emissions Speciation
48
Variable Min 10th 50th 90th Max
FULL AREA
Hg° concentration (ng/m3)
1.602
1.606
1.619
1.681
1.899
Hg2+ concentration (pg/m3)
0.265
0.804
. 3.368
14.72
149.1
HgP concentration (pg/m3)
0.391
1.426
5.183
19.18
99.33
Total (ng/m3)
1.603
1.608
1.627
1.715
2.114
EAST OF 90°W
Hg° concentration (ng/m3)
1.615
1.638
1.665
1.705
1.899
Hg2+ concentration (pg/m3)
1.486
4.745
9.966
25.25
149.1
HgP concentration (pg/m3)
4.058
8.815
14.88
28.01
99.33
Total (ng/m3)
1.622
1.655
1.691
1.755
2.114
WEST OF 90°W
Hg° concentration (ng/mJ)
1.602
1.605
1.612
1.636
1.818
Hg2+ concentration (pg/m3)
0.265
0.687
1.909
6.303
29.80
HgP concentration (pg/m3)
0.391
1.243
3.327
8.565
28.73
Total (ng/mJ)
1.603
1.607
1.618
1.651
1.860
-------
49
Table 7. Percentile Analysis of RELMAP Simulated Wet Deposition
for the Continental U.S. Using the Base-Case Emissions Speciation
Variable Min 10th 50th 90th Max
FULL AREA
Hg° wet dep. (jag/m2/yr)
0.022
0.590
2.143
6.306
10.66
Hg2+ wet dep. (|ig/m2/yr)
0.002
0.087
0.749
6.217
125.1
HgP wet dep. (jjg/m2/yr)
0.001
0.067
0.502
3.618
37.72
Total Hg (|j.g/m2/yr)
0.025
0.792
3.365
15.85
173.5
EAST OF 90°W
Hg° wet dep. (jig/m2/yr)
0.540
3.099
5.382
7.406
10.66
Hg2+ wet dep. (|ig/m2/yr)
0.242
1.837
4.269
12.40
125.1
HgP wet dep. (jig/m2/yr)
0.191
1.252
' 2.607
6.482
37.72
Total Hg (jjg/m2/yr)
0.979
6.846
12.40
25.42
173.5
WEST OF 90°W
Hg° wet dep. (p.g/m2/yr)
0.022
0.512
1.337
3.995
7.854
Hg2+ wet dep. (jj.g/m2/yr)
0.002
0.067
0.313
1.767
13.93
HgP wet dep. (|j.g/m2/yr)
0.001
0.050
0.253
1.146
6.550
Total Hg (|ig/m2/yr)
0.025
0.686
1.993
6.936
23.87
-------
50
Table 8. Percentile Analysis of RELMAP Simulated Dry Deposition
for the Continental U. S. Using the Base-Case Emission Speciation
Variable Min 10th 50th 90th Max
FULL AREA
Hg2+ dry dep. (ug/nr/yr)
0.113
0.412
1.641
8.500
153.5
HgP dry dep. (|_ig/m2/yr)
0.002
0.010
0.035
0.130
0.749
Total Hg (jig/m2/yr)
0.117
0.425
1.669
8.629
154.2
EAST OF 90*W
Hg2+ dry dep. (|^g/m2/yr)
0.434
2.649
6.263
15.53
153.5
HgP dry dep. (jj.g/m2/yr)
0.017
0.049
0.104
0.189
0.749
Total Hg (p.g/m2/yr)
0.451
2.699
6.373
15.73
154.2
WEST OF 90'W
Hg2* dry dep. (ug/m2/yr)
0.113
0.342
0.923
3.614
29.85
HgP dry dep. (|ig/m2/yr)
0.002
0.009
0.024
0.063
0.236
Total Hg (|ig/m2/yr)
0.117
0.352
0.948
3.679
30.03
-------
Legends for figures
51
Fig. 1. Estimated elemental mercury vapor emissions from all anthropogenic sources in the
continental United States in kilograms per year resolved to the RELMAP horizontal grid.
Fig. 2. Estimated divalent mercury vapor emissions from all anthropogenic sources in the
continental United States in kilograms per year resolved to the RELMAP horizontal grid.
Fig. 3. Estimated particulate mercury vapor emissions from all anthropogenic sources in the
continental United States in kilograms per year resolved to the RELMAP horizontal grid.
Fig. 4. Annual average elemental mercury vapor air concentration from continental U. S.
emissions as simulated by the RELMAP using the base-case emission speciation estimates.
Concentrations are in units of nanograms per cubic meter (ng/m3). The assumed global
background elemental mercury concentration of 1.6 ng/m3 is not included.
Fig. 5. Annual average divalent mercury vapor air concentration from continental U. S.
emissions as simulated by the RELMAP using the base-case emission speciation estimates.
Concentrations are in units of nanograms per cubic meter (ng/m3).
Fig. 6. Annual average particulate mercury air concentration from continental U. S.
emissions as simulated by the RELMAP using the base-case emission speciation estimates.
Concentrations are in units of nanograms per cubic meter (ng/m3).
-------
Legends for figures (continued)
52
Fig. 7. Annual total wet deposition of elemental mercury vapor from continental U. S.
emissions as simulated by the RELMAP using the base-case emission speciation estimates.
Depositions are in units of micrograms per square meter (jag/m2).
Fig. 8. Annual total wet deposition of elemental mercury vapor from the global background
air concentration of 1.6 nanograms per cubic meter (ng/mJ) as simulated by the RELMAP.
Depositions are in units of micrograms per square meter (jag/m2).
Fig. 9. Annual total wet deposition of divalent mercury vapor from continental U. S.
emissions as simulated by the RELMAP using the base-case emission speciation estimates.
Depositions are in units of micrograms per square meter (^g/m2).
Fig. 10. Annual total wet deposition of particulate mercury from continental U. S. emissions
as simulated by the RELMAP using the base-case emission speciation estimates. Depositions
are in units of micrograms per square meter (p.g/m2).
Fig. 11. Annual total wet deposition of all forms of mercury from continental U. S.
emissions and from the global background concentration as simulated by the RELMAP using
the base-case emission speciation estimates. Depositions are in units of micrograms per
square meter (|ig/m2).
-------
53
Legends for figures (continued)
Fig. 12. Annual total wet deposition of all forms of mercury from continental U. S.
emissions and from the global background concentration as simulated by the RELMAP using
the alternate emission speciation estimates. Depositions are in units of micrograms per square
meter (fag/m2).
Fig. 13. Annual total dry deposition of divalent mercury vapor from continental U. S.
emissions as simulated by the RELMAP using the base-case emission speciation estimates.
Depositions are in units of micrograms per square meter (|j.g/m2).
Fig. 14. Annual total dry deposition of particulate mercury from continental U. S. emissions
as simulated by the RELMAP using the base-case emission speciation estimates. Depositions
are in units of micrograms per square meter (jag/m2).
Fig. 15. Annual total dry deposition of all forms of mercury from continental U. S. emissions
as simulated by the RELMAP using the base-case emission speciation estimates. Depositions
are in units of micrograms per square meter (jag/m2).
Fig. 16. Annual total dry deposition of all forms of mercury from continental U. S. emissions
as simulated by the RELMAP using the alternate emission speciation estimates. Depositions
are in units of micrograms per square meter (jag/m2).
-------
F'3
-------
-------
~ 3 - 10
~ lO -30
-100
100 - 300
> 300
3
-------
~ 0.005 -0.01
~ 0.01 -0.02
B 0.02 -0.05
10.05 -0.1
¦ > 0.1
-------
0.002 -0.005
0.005 -0.01
0.01 -0.02
0.02 -0.05
> 0.05
-------
U 0.002 -0.005
~ 0.005 -0.01
00.01 -0.02
¦ 0.02 -0.05
¦ > 0.05
-------
0.05 -0.1
0.1 -0.2
2 -0.5
0.5 - 1.0
> 1.0
Hg-7
-------
Do.5 ~ 1.0
~ l.O -2.0
32.0 -5.0
¦ 5.0 - 10.0
¦ > 10.0
-------
1.0 -2.0
12.0 -5.0
5.0 - 10.0
10.0 -20.0
> 20.0
-------
i o
-------
-------
-------
~ l.O -2.0
~ 2.0 -5.0
HI 5.0 - 10.0
¦ lO.O -20.0
¦ > 20.0
-------
0.02 -0.05
-0.1
0.1 -0.2
0.2 -0.5
> 0.5
-------
~ l.O -2.0
32.0 -5.0
¦ 5.0 - 10.0
¦ lO.O -20.0
¦ > 20.0
-------
~ l.O -2.0
2.0 -5.0
5.0 - 10.0
10.0 - 20.0
> 20.0
fij. i y
-------
TECHNICAL REPORT DATA
1. REPORT NO.
EPA/600/A-95/115
2 .
3 ,
4. TITLE AND SUBTITLE
The modeling of Regional-Scale Atmospheric Mercury
Transport and Deposition Using RELMAP
5.REPORT DATE
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
0. Russell Bullock, Jr
William G. Benjey
Martha H. Keating
8. PERFORMING ORGANIZATION REPORT
NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
NERL/Atmospheric Modeling Division
Mail Drop 80
Research Triangle Park, NC 2 7711
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
NATIONAL EXPOSURE RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
13.TYPE OF REPORT AND PERIOD
COVERED
Book Chapter
14. SPONSORING AGENCY CODE
EPA/600/9
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The Regional Lagrangian Model of Air Pollution (RELMAP) is used to simulate the
emission, transport and diffusion, chemical transformation, and wet and dry-
deposition of elemental mercury gas, divalent mercury gas and particulate mercury.
Based on recent modeling advances in Europe, the RELMAP has been modified to
simulate a reduction-oxidation (redox) balance for mercury dissolved in cloud and
rain water. This redox balance is used in the estimation of a variable
precipitation scavenging ratio for elemental mercury. Wet removal of divalent
mercury gas and particulate mercury is modeled using constant wet scavenging
ratios. Dry deposition is estimated using deposition velocities developed from
various modeling studies in the U.S. and in Europe. " A mercury air emission
inventory for the continental United States is used to estimate the releases of
elemental mercury gas, divalent mercury gas and particulate mercury based on
emission speciation estimates for the various emitter types in the inventory.
Average annual concentrations and wet and dry deposition totals for these three
forms of mercury have been simulated. -The results of the simulation are used to
estimate the quantity of mercury emitted to the air annually over the United States
and the amount that is subsequently deposited back to U.S. soils and water bodies.
An analysis of the modeling results also provides some information about the areas
of the country thought to have the most significant exposure from all air emissions
of mercury. This analysis contributes to the understanding of the key variables,
such as source location, chemical and physical form of emission, or meteorology,
that lead to these outcomes.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED
TERMS
C.COSATI
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This
Report)
UNCLASSIFIED
21.NO. OF PAGES
20. SECURITY CLASS (This
Page)
UNCLASSIFIED
22. PRICE
------- |