percent of all snowfall in the Great Lakes (Goyette et al.
(Croley and Luukkonen, 2002).  The absence of the Great
2000), the ability of the Regional Climate Models to
Lakes in the Canadian model and the representation of the
simulate this is very important.  During a simulation of one
Great Lakes area as a continuous land surface could lead
10 day period in December 1985, the model was able to
the model to predict more convective precipitation in the
basically reproduce the distribution of lake-effect precipita-
summer and omit the effect of lake-effect precipitation in
tion over the Great Lakes basin (Bates et al, 2002).  In the
the winter (Lofgren et al. 2002a).  On the other hand, it has
major snow belt areas, precipitation was under-forecast at
been criticized that this could not be the only reason for
all four model resolutions, but predictions became more
the differences in predictions, as there have been differ-
4.3
accurate at finer resolutions (Bates et al, 2002).  It is
ences of similar extent in the two models elsewhere in
expected that with climate change there will be almost no
North America.
change in the northernmost snow belts, but an approxi-
mately 50 percent decrease in the amount of lake-effect
Regional Climate Models
snow in southernmost belts (Kunkel et al, 2002).
Regional Climate Models were developed to account for
The Hadley Centre in the United Kingdom has developed
the omission of regional land forms in General Circulation
an Regional Climate Model, “Providing Regional Climates
Models.  By scaling down the grid area to typically 50 km,
for Impact Studies” (PRECIS), which is now being offered
as compared to 300 km in a General Circulation Model, the
free to developing countries.  The advantage of this
Regional Climate Model can better predict regional climate
Regional Climate Model is that it can be run on a relatively
changes (Hadley Centre 2001).  Once known as Limited
inexpensive personal computer, as compared to the
Area Models, Regional Climate Models do a much better
supercomputers required for General Circulation Models
job of simulating convection processes, which typically
(Hadley Centre 2001).  Each 30 year simulation takes about
occur on the scale of a few kilometres.  As convection is the
four to six months to complete.  Simulations over the
mechanism responsible for most summer precipitation and
United Kingdom, South Africa and Japan have shown a
severe storm weather, these smaller scale models should be
better ability to match past precipitation averages, which
more accurate at predicting summer precipitation, an area
increases confidence in the future success of this model
where General Circulation Models are very inefficient.
(Hadley Centre 2001).
Regional Climate Models are nested into an existing
General Circulation Model and the predictions for a certain
Most Importantly, Regional Climate Models can work with
climate scenario generated by the General Circulation
other models to predict climate changes other than
Model are the input data for the Regional Climate Model.
temperature and precipitation.  The PRECIS model has
Problems with this method are obvious, as miscalculations
been developed to work with models to predict storm
and mistakes in the predictions from the General Circula-
surges (Hadley Centre 2002b).  Using an Regional Climate
tion Model are amplified within the Regional Climate
Model will give a better indication as to the frequency and
Model.  However, these problems are overlooked, as the
intensity of future storm surges.  Development of a
purpose is to simulate climate processes influenced by
hydrological model of ground water using Regional
regional land forms like the Great Lakes.
Climate Model inputs for the Great Lakes basin could
suggest if ground water resources will increase or decrease
Results of the Canadian Regional Climate Model simula-
as our climate changes.
tions are not available at this time however; initial trial runs
from another regional model have seen valuable results.
Historical Analogues
Observations from a 10 day and 24 month period corre-
lated well with results from separate model runs using
Historical analogues, observations about the way climate
input data from documented weather reports.  By using an
has varied and changed in the past, are good indicators as
Regional Climate Model rather than a General Circulation
to how climate may change in the future.  By examining the
Model, the changes in lake surface temperatures and ice
climate record from the past, we can estimate variability
cover were replicated rather well.  Important feedback
and tell when an extreme is outside the range of the natural
cycles between land, lake and atmosphere were simulated
variability.  To use historical analogues, one would choose
at the finer scale, allowing the model to closely predict a
warmer than average years from the historical record and
similar climate to the observed conditions (Goyette et al.
see how the increases in temperature may have affected
2000).
precipitation, drought conditions and crop growth.
Historical analogues are a reliable way to discover the
Regional Climate Models have proven to be particularly
consequences of climatic change for any region where
good at simulating the interactions between lake, land and
long-term records exist.
atmosphere to produce predictions of  lake-effect precipita-
tion.  As lake effect snowfall accounts for about 30 to 50
97