Marine Ecosystem Dynamics Modeling Laboratory


We started involving in the Great Lake research 10 years ago. NSF CooP Program and NOAA Coastal Ocean Program funded the Keweenaw Interdisciplinary Transport Experiment in Superior (called KITE) to study the physical, chemical, and biological effects, which arise from the formation, evolution, and perturbation of the thermal bar and Keweenaw Current, on the distinct nearshore and offshore environments in Lake Superior. The modeling was focused on (1) developing a physical model with capability of simulating the formation, evolution, and perturbation of thermal bar and Keweenaw Current, (2) coupling an ice model and an nutrient-phytoplankton-zooplankton-detritus (NPZD) model into the physical model, and (3) examining the cross-margin and vertical transports of momentum, solutes, particulates, and organisms associated with the thermal bar and Keweenaw Current.

Many scientists participated in modeling efforts, including Dr. Ralph at University of Minnesota, Drs. Green and Budd at Michigan Technological University, Jim Churchill at Woods Hole Oceanographic Institution, etc. The first question raised in the first funding year is why all previous modeling efforts fail to resolve the intensity of the Keweenaw Current-a strong current jet with a cross-shelf scale of a few kilometers and a speed of over 60 cm/s. We argued that the failure is due to the limitation of the model resolution and poor resolving of the coastal geometry. For this reason, we modified ECOM-si (a modified version of POM) into the non-orthogonal coordinate system, so that we could make a better fitting the local coastal geometry. The modified high-resolution ECOM-si did a pretty good job to resolve the intensity of the current jet and thermal front for the 1973 and 1999 experiments, but fails to resolve the current separation and eddies under the wind environement along the Keweenaw Current.

In the 2001 we used FVCOM, an unstructured grid finite-volume coastal ocean model to replace ECOM-si for Lake Superior. By using the unstructured triangular meshes, FVCOM provides an accurate fitting of coastal geometry and also is capable to link to small lakes with Laker Superior. We did some experiments using the same resolution as ECOM-si, and found that FVCOM provides a better structure of the current jet and thermal front than structured grid finite-difference models. We have prepared all meteorological, satellite , current meter data and made them ready for FVCOM to simulate the circulation and stratification in Lake Superior for the KITE experiment years. Unfortunately, we only did for selected months, because our NSF project ended. We did the current separation experiment in Narragansett Bay/Mount Hope Bay and found whether or not a model could resolve the current separation and eddy formation depends on both horizontal resolution and geometric resolving. FVCOM reproduced the current separation in that bay as the horizontal resolution inceases to 50 m. Experiments made by the Japanese scientists with FVCOM on the eddy formation of the Kurioshio also reports that FVCOM very well reproduces the current separation process and eddy shedding. Current version of FVCOM in Lake Superior is about 500 m along the coast. We expect that FVCOM should be capable to resolve the current separatio of the Keweenaw Current as we increases its horizontal resolution, to about 20-50 m, which should be very easy to achieve in an unstructured grid model like FVCOM.

In this web site, we attempt to provide the public a brief view of our previous modeling accomplishments in Lake Superior. We do find that we have undergone a learning process and are trying hard to find a beter solution for modeling in this lake. We believe that FVCOM is capable to address all technical difficulities in Lake Superior modeling. We are working hard to find a fund that allow us to set up an intergrated model system for Lake Superior.

FVCOM Testing Results
Unstructured grids of Lake Superior FVCOM. Horizontal resolution is about 500 m m near the coast and about 10 km in the interior. This model mesh links the Portage Lake and Torch Lake through the Waterway on the Keweenaw coast.
In order to resolve the current separation, we are increasing the horizontal resolution to 20-m within the width of 20 m along the Keweenaw.
Click here to view the full-size figure that includes amplified view of meshes in the Portage and Torch Lakes.

Non-Orthogonal ECOM-si Results
Non-orthogonal structured grids of Lake Superior non-orthogonal coordinate ECOM-si. Our previous process-oriented modeling experiments for KITE Program were conducted using this model with understanding of its limitation for the real-time simulation due to inflexibility of geometrical fitting. Non-orthogonal coordinate used in this model provides a reasonable fitting of the coastline along the Keweenaw coast, but not for other coasts. It is very difficult to adjust the grid to increase along-shelf resolution. This model resolve the current spearation and eddy formation at the headland area of the Keweenaw Peninsula but not for the case with winds. This is one of the reason we replace ECOM-si by FVCOM.
Click here to view the grids.

Lake Superior Meteorological Model (MM5/WRF)
We have set up a meso-scale meteorological model (called MM5) over Lake Superior. This is a nested model with two-way nesting from 3 domains. Lake Superior is covered by domain 3 with the horizontal resolution of 5 km. Since the ETA has provided the wind field with a horizontal resolution of about 27 km, 3 domains are modified to be 27, 9, and 3 km. It means that our current MM5 provides a meterological forcing with a 3 km resolution. All buoys and coastal weather station measurement data are assimilated into MM5 to provide a reasonable wind and heat flux flieds.
We are also working on shifting MM5 to WRF.
Click here to learn the details for this model.

Bathymetry and Other Maps
Click here to view details

«Previous Posted on January 15, 2014