as well as the relative concentrations of mercury in the
lake and in the air above the lake.
Mercury, like other pollutants, is also subject to dry
Estimates for Lake Michigan (Vette et al., 2002; Landis
deposition phenomena, by which it is transported
and Keeler, 2002) based on measurements made in the
down  to  the  earth’s  surface  by  atmospheric
Lake Michigan Mass Balance Study suggest that the net
direction of the elemental mercury Hg  flux during that
dispersion and then some portion of it adheres to
study was upwards (i.e. out of the lake); this may also
various earth surfaces such as water, trees, other
be the case for Lake Michigan currently.
vegetation or buildings.
Overall, the mercury in the lakes can be considered to
be dry and wet deposited downward in the form of
Hg(II) and Hg(p), and some portion of it revolatilized
upwards in the form of Hg0; if more is deposited than
The most significant of these regarded the treatment of
revolatilized, then the mercury concentrations in the
atmospheric chemistry.
lake ecosystem will increase, and vice versa.
Simulating the Atmospheric Chemistry of Mercury
Modeling Methodology
In the dioxin modeling, only a gas-phase reaction with OH
and gas- and particle-phase photolysis were considered in
Atmospheric Transport and Dispersion Model
the chemistry module, and it was assumed that no conver-
sion from one congener to another occurred.  However, an
The NOAA-HYSPLIT_4 model (Hybrid Single-Particle
adequate simulation of mercury’s atmospheric chemistry
Lagrangian Integrated Trajectory model, Version 4) (Draxler
requires a more sophisticated analysis.
and Hess, 1998), augmented by a spatial and chemical
interpolation procedure (Cohen et al., 2002), was used to
The mercury chemical equilibrium and reaction scheme
estimate source-receptor relationships.  In this application
used in this application is similar to that currently being
of the HYSPLIT model, puffs of pollutant were considered
employed in other atmospheric mercury models (e.g.
to be emitted from each given source location.  The
Ryaboshpko et al., 2002, 2003) and is summarized concep-
subsequent advection and dispersion of the pollutant puffs
tually in Figure 4.
was then simulated using meteorological data supplied to
the model.  A full year (1996) of meteorological output
Ambient concentrations of O3, SO2, and soot and the pH
from NOAA’s Nested Grid Model (Rolph 1997) was used for
and aqueous concentration of Cl-1;were estimated from
the year-long simulations in this study.  These data had a
ambient data, and concentrations of other key reactants
horizontal resolution of 180 km, 11 vertical levels up to
(OH, Cl2, etc.) were estimated using other empirically-
6000 meters elevation, and a two-hour temporal resolu-
based procedures.
Simulating Dry and Wet Deposition of Atmospheric
The use of more highly resolved meteorological data would
no doubt improve the accuracy of the simulation, especially
in the prediction of concentrations and deposition at
As noted previously, elemental mercury is only sparingly
specific locales (e.g. in the model evaluation exercises
soluble in water, and it is not efficiently incorporated into
described below).  However, it is not likely that the overall
wet deposition.  As a result, the preponderance of mercury
deposition and the source-receptor relationships for the
in wet deposition is in the oxidized or particulate forms.
Great Lakes estimated here would be significantly affected
Mercury, like other pollutants, is also subject to dry
by use of more highly resolved meteorological data.  This
deposition phenomena, by which it is transported down to
supposition could be confirmed in future work through the
the earth’s surface by atmospheric dispersion and then
use of such data, which have recently become available.
some portion of it adheres to various earth surfaces such as
water, trees, other vegetation or buildings.  In addition to
The HYSPLIT_4 model has recently been used to simulate
this downward dry deposition component, there is also
atmospheric fate and transport of dioxin to the Great Lakes
generally an upward flux of mercury from land and waters
(Cohen et al., 2002) and many of the model modifications
(e.g. from natural sources or previously deposited anthro-
made for that study have been retained in the present
pogenic emissions).
application.  Several mercury-specific changes and addi-
tions were incorporated into the model for this analysis.