somewhat less than that measurement-based deposition
estimate.  Nevertheless, it is encouraging to note that the
In  response  to  encouragement  from  several
overall model-predicted deposition is the same order of
magnitude as the measurement-based estimate, and the
quarters, including recommendations made by the
difference between the two is in the expected direction.
IJC in its Tenth Biennial Report on Great Lakes Water
Additional Model Evaluation Exercises Underway
Quality,  mercury  measurement  capabilities  have
been  added  to  several  stations  in  the  binational
A detailed evaluation procedure is underway for the NOAA-
HYSPLIT mercury modeling analysis in which model
Integrated Atmospheric Deposition Network.
outputs or predictions are being compared in detail to
available 1999 ambient mercury measurements.  There are
several data sets that are potentially available for this
Michigan with measurement-based estimates obtained
purpose, including the following: 1. weekly mercury wet
during the Lake Michigan Mass Balance Study (Landis and
deposition measured at 21 sites in the Mercury Deposition
Keeler, 2002; Vette et al., 2002).  An advantage of this
Network for some or all of 1996; 2. ambient concentrations
comparison is that it is for an entire lake (rather than a
and event-based wet deposition measurements at several
single location) and it is one of the Great Lakes.  This
sites in the Great Lakes region (Hoyer 1995; Keeler 2001);
comparison is shown in Figure 25 for estimated wet
3. ambient concentrations and event-based precipitation
deposition of Hg(II) and Hg(p) (combined), dry deposition
wet deposition measurements near Lake Champlain, NY
of Hg(II) and Hg(p), and total mercury deposition.  The
(Burke et al., 1995); 4. ambient concentrations and/or wet
ranges shown for the Lake Michigan Mass Balance Study’s
deposition measurements at several sites in Canada
data are the reported standard deviations (Landis and
(Kellerhals et al., 2000; Beauchamp et al., 1997; Poissant
Keeler, 2002) for the estimates, and are a measure of the
and Pilote, 1998); and 5. measurements of ambient
uncertainty associated with these estimates.  Analyses to
concentrations and wet deposition at several locations in
estimate uncertainties in the HYSPLIT-based model
the mid-Atlantic region (Mason et al., 1997ab).
estimates have not yet been carried out, but a rough
estimate of the uncertainty (+/- 25%) has been indicated for
In response to encouragement from several quarters,
these data in Figure 22.
including recommendations made by the IJC in its Tenth
Biennial Report on Great Lakes Water Quality, mercury
The Lake Michigan Mass Balance Study measurement-based
measurement capabilities have been added to several
estimate of net Hg  deposition is also shown – actually a net
stations in the binational Integrated Atmospheric Deposi-
evasion from the lake of 453 +/- 144 kg/yr (thus, it is shown
tion Network.  In Ontario, the Point Petre and Burnt Island
as a negative number, in contrast to the other values, which
master stations and the Egbert satellite station have been
represent downward deposition to the lake).  As discussed
measuring total gaseous mercury as of April 1998 or earlier
above, this quantity was not estimated in the modeling
and mercury in precipitation has been measured since on
analysis, and so a comparison is not possible.
or before November of 2001.  Speciation of reactive
gaseous mercury and total particulate mercury is now
Precise agreement between the two sets of estimates would
underway at the Point Petre site as well.  Plans to support
also not be anticipated, as the modeling results are for
mercury measurements at various  Integrated Atmospheric
1996, and the Lake Michigan Mass Balance Study’s mea-
Deposition Network stations in the United States are
surements were carried out in 1994-1995.  There were
awaiting the required financial support.
undoubtedly distinct weather patterns and different
precipitation amounts in these two periods, and perhaps
Both governments are participating in a Lake Ontario
even more importantly, mercury emissions during these
Atmospheric Deposition Study (LOADS), under which
two periods probably changed significantly.  For example,
mercury samples are being collected at Sterling, Potsdam
during the 1994-1996 period, there were significant
and Stockton, New York, as well as at the Point Petre station
reductions in emissions from medical waste incinerators
referred to earlier.  Shipboard mercury sampling is also an
(due to the closure of many facilities) and municipal waste
aspect of this program.
incinerators (due to closures, retrofits, and changes in
waste stream composition).  Moreover, the modeling
Given the uncertainties in the inventory and in the current
analysis did not include contributions from sources outside
state of understanding of mercury’s atmospheric behavior,
the United States and Canada, nor did it include the
and certain limitations of this modeling analysis (e.g. global
contribution of natural emissions.  For all these reasons,
sources have not been included; temporal variation in
one would expect that the model predictions would be
emissions are not available), it is not expected that the