Water Resources Issues of the Laurentian Great Lakes
155
www.glerl.noaa.gov/eegle/). Many of the earlier hydrodynamic models developed
for the lakes are summarized by Hayashida et al. (2000).
The most detailed hydrodynamic models have for the most part been
applications of the Princeton Ocean Model (POM) , modified for the closed
boundary conditions of lake modeling. Such models have been developed for
Lakes Michigan, Erie and Ontario (Schwab, pers. commun.). Podber and Bedford
(1999) also developed a finite difference model that could be applied to the lakes,
with the interesting feature of an automatically adjusted variable vertical grid
resolution, to allow detailed transport calculations in the neighborhood of the
seasonal thermocline. In addition, finite element models have been proposed
(Hayashida et al. 2000), which provide additional flexibility in modeling certain
regions of a lake with a finer spatial grid. In this latter case, for example, interest
was primarily in modeling the region near the mouth of the Niagara River in Lake
Ontario.
With several exceptions, most modeling efforts in the lakes have, in general,
not directly combined hydrodynamic and water-quality models. To a large extent,
this is because of the different temporal and spatial scales normally used in these
two types of models. The most common water-quality model that has been applied
for Great Lakes problems is W ASP4 (Ambrose et al. 1988), which considers the
physical system as a series of mixed reactors. Because of the complexity and
number of water-quality processes of interest in Great Lakes modeling, the system
is usually divided into a relatively small number of reactors, with a resulting
spatial scale much greater than the typical scale used for a hydrodynamic model.
Ideally, a hydrodynamic model would be run to provide advective and dispersive
transport terms for the water-quality model, but then issues of scale matching need
to be considered (Atkinson et al. 1998). Typical variables of interest in Great
Lakes models include traditional parameters such as temperature, turbidity,
dissolved oxygen, nutrient levels, etc., as well as newer parameters such as
particulate and dissolved concentrations of heavy metals and hydrophobic organic
contaminants (e.g., PCBs).
In order to model the fate and transport of persistent toxic chemicals,
sediment/water interactions must be incorporated. This is partly because of the
role played by suspended particles in the transport of these chemicals, and also
because sediments tend to serve as a significant source for these compounds (see
below). Toxic contaminants pose a threat to human health as well as to other
species, since humans are usually at the top of the food chain. In addition, the
problem of bioaccumulation must be considered and there is interest in linking
water-quality models with food-chain models, in order to evaluate the potential
risks involved, for example, with eating fish caught in the Great Lakes (although
there are advisories for eating fish caught in all five lakes, it is generally
considered safe to eat occasional meals oflake-caught fish).
An interesting aspect of modeling sediment transport and distribution in large
lakes is the formation of benthic nepheloid layers (BNLs). These are regions near
the bottom of the lake, several tens of meters thick, where suspended particle
concentrations may be nearly an order of magnitude higher than in the overlying
water column. They are thought to form either as a natural consequence of a
155
www.glerl.noaa.gov/eegle/). Many of the earlier hydrodynamic models developed
for the lakes are summarized by Hayashida et al. (2000).
The most detailed hydrodynamic models have for the most part been
applications of the Princeton Ocean Model (POM) , modified for the closed
boundary conditions of lake modeling. Such models have been developed for
Lakes Michigan, Erie and Ontario (Schwab, pers. commun.). Podber and Bedford
(1999) also developed a finite difference model that could be applied to the lakes,
with the interesting feature of an automatically adjusted variable vertical grid
resolution, to allow detailed transport calculations in the neighborhood of the
seasonal thermocline. In addition, finite element models have been proposed
(Hayashida et al. 2000), which provide additional flexibility in modeling certain
regions of a lake with a finer spatial grid. In this latter case, for example, interest
was primarily in modeling the region near the mouth of the Niagara River in Lake
Ontario.
With several exceptions, most modeling efforts in the lakes have, in general,
not directly combined hydrodynamic and water-quality models. To a large extent,
this is because of the different temporal and spatial scales normally used in these
two types of models. The most common water-quality model that has been applied
for Great Lakes problems is W ASP4 (Ambrose et al. 1988), which considers the
physical system as a series of mixed reactors. Because of the complexity and
number of water-quality processes of interest in Great Lakes modeling, the system
is usually divided into a relatively small number of reactors, with a resulting
spatial scale much greater than the typical scale used for a hydrodynamic model.
Ideally, a hydrodynamic model would be run to provide advective and dispersive
transport terms for the water-quality model, but then issues of scale matching need
to be considered (Atkinson et al. 1998). Typical variables of interest in Great
Lakes models include traditional parameters such as temperature, turbidity,
dissolved oxygen, nutrient levels, etc., as well as newer parameters such as
particulate and dissolved concentrations of heavy metals and hydrophobic organic
contaminants (e.g., PCBs).
In order to model the fate and transport of persistent toxic chemicals,
sediment/water interactions must be incorporated. This is partly because of the
role played by suspended particles in the transport of these chemicals, and also
because sediments tend to serve as a significant source for these compounds (see
below). Toxic contaminants pose a threat to human health as well as to other
species, since humans are usually at the top of the food chain. In addition, the
problem of bioaccumulation must be considered and there is interest in linking
water-quality models with food-chain models, in order to evaluate the potential
risks involved, for example, with eating fish caught in the Great Lakes (although
there are advisories for eating fish caught in all five lakes, it is generally
considered safe to eat occasional meals oflake-caught fish).
An interesting aspect of modeling sediment transport and distribution in large
lakes is the formation of benthic nepheloid layers (BNLs). These are regions near
the bottom of the lake, several tens of meters thick, where suspended particle
concentrations may be nearly an order of magnitude higher than in the overlying
water column. They are thought to form either as a natural consequence of a