measurements.  Indirect atmospheric deposition would
At any given location and time, the relative magnitude of
account for a portion of these contributions, but the fraction
the downward and upward components will vary, and this
is unknown.  The METAALICUS project, described in section
relative magnitude will generally be different for each form
1.4 of this report, represents an attempt to further under-
of mercury.  The general consensus of the scientific
stand these processes.  The knowledge gained in that
community is that the upward component of the surface-
project should support development of more accurate
exchange flux of Hg(II) and Hg(p) at most locations is
estimates of such phenomena for the Great Lakes.
relatively insignificant, (i.e. for these two forms, the net flux
is almost always in the direction of deposition (down), and,
Methodology to Estimate Source-Receptor Relation-
as a reasonable, simplifying assumption, the upward flux
phenomena for these two species is generally ignored).
This approach has been taken in the atmospheric mercury
A procedure was developed to establish linkages between
modeling analysis discussed here; that is, only the down-
significant point and areal sources of mercury throughout
ward flux of Hg(II) and Hg(p) has been modeled.
the United States and Canada, and the subsequent deposi-
tion of this contaminant to the Great Lakes basin.  In this
In addition, as a simplification of the deposition process,
technique (Cohen et al., 2002) an interpolation procedure
the approach of Bullock (2002) and others was followed, in
is used to estimate detailed source-receptor relationships.
which it was assumed that the total deposition of elemental
To conduct the analysis, explicit HYSPLIT modeling of
mercury from direct anthropogenic emissions was roughly
emissions was performed for a limited number of selected
balanced by the volatilization of previously deposited
source locations.  The impact of any given source on the
emissions of elemental mercury from both natural and
Great Lakes was estimated based on a weighted average of
anthropogenic sources.  That is, the net deposition of
the impact of the four explicitly modeled locations that
elemental mercury was assumed to be zero.  A more
would be nearest to the source in question.
sophisticated treatment of these phenomena would no
doubt be beneficial, but the limited understanding of the
To account for the varying proportions of different Hg0
relevant processes and a lack of measurement data to
species being emitted from different sources, separate unit-
evaluate estimates make their inclusion somewhat impracti-
emissions simulations of Hg(II), Hg0 , and Hg(p) emissions
cal.  Nevertheless, as discussed in more detail later, the
were made at each standard source location.  The impact of
ability of this and other models to satisfactorily explain
a source emitting a mixture of Hg , Hg(II), and Hg(p) was
ambient mercury concentrations and deposition provides
evidence that this simplified approach is reasonably
estimated based on a linear combination of these pure-
consistent with the overall net cycling of elemental mercury
component unit emissions simulations.  In sum, both spatial
between the atmosphere and the earth’s surface.
and chemical interpolation procedures were used to
estimate the impact of each source in the inventory on each
Only the direct deposition to the lake surfaces has been
of the Great Lakes.
estimated in this modeling analysis.  Indirect atmospheric
contributions, resulting from deposition to a lake’s water-
Validity of the Interpolation Procedure
shed and subsequent transfer to the lake, have not been
estimated as part of this modeling.  Because their water-
The spatial and chemical interpolation methodology used
sheds are smaller, relative to their size, than many other
relies on the assumption that the atmospheric fate and
lakes, the Great Lakes probably receive proportionally less
transport of Hg from any given source is not influenced by
mercury from this indirect route than many other lakes.
the emissions from any other source.  This assumption is
believed to be generally valid, based on the following
Estimating the amount of mercury loading to the Great
Lakes contributed through this indirect pathway is very
uncertain, as there are few measurements of runoff and
Mercury is present at extremely minute levels in the
tributary inputs to the lakes.  Moreover, determining the
atmosphere.  As a consequence, it will not affect
portion of such input that arises as a result of atmospheric
meteorology.  Thus, meteorological parameters, such as
deposition as opposed to direct discharges to the tributaries
wind speed and direction, temperature, humidity,
or natural mercury present in the ecosystem is difficult.
precipitation, etc., can be estimated independently and
Rolfhus et al. (2003) have estimated that total tributary
provided to the model.
input to Lake Superior represented approximately 27
percent of the total loading to the lake, based on measure-
Also, most species that react with mercury compounds
ments made in year 2000.  Landis and Keeler (2002)
(e.g. SO2) are generally present at much higher concen-
estimate that tributary inputs account for approximately 16
trations than the mercury compounds.  Other species
percent of the loading to Lake Michigan, based on 1994-95
(e.g. OH) generally react with many compounds other