Table 1
Predictions for the Great Lakes Region
Hadley Centre Model
Canadian Model
Temperature change by 2030
+1.2oC (Lake Superior)
+2.0oC (Lake Superior)
Temperature change by 2090
+2.9oC (Lake Superior)
+5.4oC (Lake Superior)
Precipitation -summer
Slight decrease
0-20% decrease
Precipitation -winter
Slight increase
0-20% increase
Annual precipitation
Increase
Increase
Runoff
1-10% increase
10-12% decrease
Soil Moisture
Increase
20-40% decrease
Evaporation
Slight increase
Increase
(Frederick and Gleick, GCSI and the Meteorological Service of Canada, 1999;
Legates 2002; 1995 Lofgren et al., 2002b)
the two most commonly used General Circulation Models,
The comparison between the Hadley and Canadian models
the Hadley and Canadian models.  The conditions used
also includes runs to evaluate the ability of the models to
were the same for both General Circulation Models, though
simulate past and present climates.  The National Oceanic
they were not reported in the literature.  Although similar
and Atmospheric Administration and the Great Lakes
in purpose and methodology, the two models produce
Environmental Research Laboratory have conducted trial
somewhat different results for the Great Lakes area.  While
runs of the 1961 to 1990 base period and compared the
both predict a slight temperature increase and agree that
results to observed mean temperatures.  The Canadian
evaporation is likely to decrease, they do not agree on the
model showed a strong cold bias during the winter and
direction of precipitation change, runoff or the effect on
spring, and a warm bias over the fall and summer for the
ground water (See Table and Figure 1).
Lake Superior area.  Predicted averages showed differences
of up to five degrees Celsius in daily minimum tempera-
tures.  The Hadley model showed a warm bias during the
winter and early spring, and was not able to predict an
adequate range of daily maximum and minimum tempera-
tures in the summer.  However, the Hadley model showed
biases usually less than three degrees Celsius.
Precipitation patterns also were predicted for the 1961 to
1990 run.  The Hadley model proved better at duplicating
early summer and late fall observations in precipitation,
however, it did produce overestimates in late winter and
spring, and underestimates in late summer and early
winter.  The Canadian model produces an estimate of
summer precipitation that greatly exceeds the observed
values and a small underestimate for fall and early winter.
Overall, the Hadley model provides a more accurate
representation of the observed precipitation averages.
The results of the trial run show that neither model is a
completely accurate predictor of average temperature or
precipitation.  Still, differences in the models may be able
to account for the different biases.  The Hadley model
FIGURE 1
exhibited a smaller bias overall, and this may be due to the
Comparison between present precipitation and future
crude representation of the Great Lakes in the model.  It
projections using the Canadian and Hadley models
uses 3 grid cells to represent the Great Lakes, whereas the
(Adams and Gleick, 2000)
Canadian model has no representation of the lakes at all
96