Great Lakes Coastal Forecasting System: Next Generation

Making the transition from research to operations

Primary Investigator


Project Overview

GLCFS next generation figure 1 - model grids

Figure 1: The 3rd generation NOAA GLERL Great Lakes Coastal Forecasting System (GLCFS) uses an unstructured grid (i.e., triangular shapes of adaptable size) to better model physical processes

NOAA’s Great Lakes Coastal Forecasting System (GLCFS) is a set of hydrodynamic computer models that predict lake circulation and other physical processes (e.g. circulation, thermal structure, waves, ice dynamics) of the lakes and connecting channels in a real-time nowcast and forecast mode. These research models provide us with timely information on currents, water temperatures, short-term water level fluctuations (e.g. seiche, storm surge), ice, and waves for up to 120 hours into the future. Predictions of these lake conditions assist planners and managers in addressing several critical issues in the Great Lakes such as navigation, search and rescue,contaminant spill response, harmful algal blooms, beach management, and recreational use. The GLCFS also functions to provide NOAA’s National Weather Service (NWS) marine forecasters with information that improves the accuracy and efficiency of Great Lakes marine forecasts and warnings. The current 3rd generation of the GLCFS is run in near real-time at NOAA’s Great Lakes Environmental Research Laboratory (GLERL), and operationally at NOAA’s National Ocean Service (NOS) under the name Great Lakes Operational Forecast System (GLOFS).

The GLOFS is a NOAA automated model-based prediction system aimed at providing improved predictions (guidance) of water levels, water currents and water temperatures in the five Great Lakes (Erie, Michigan, Superior, Huron and Ontario) for the commercial, recreation, and emergency response communities. GLOFS generates hourly nowcast guidance (analyses) and four times daily forecast guidance of total water level, current speed and direction, and water temperature for each of the Great Lakes. The GLOFS predictions will enable users to increase the margin of safety and maximize the efficiency of commerce throughout the Great Lakes.

GLCFS next generation figure 2 - Currents

Figure 2: Snapshot of Great Lakes circulation using model-predicted vertically-averaged currents from the Great Lakes Coastal Forecasting System (GLCFS). See NOAA GLERL’s GLCFS Great Lakes Surface Currents Map for daily visualizations of lake current flow patterns based on simulations from the GLCFS.

Background and Current Operations

An overview of the evolution of the GLCFS from research to operations is presented in Table 1 (below). The GLERL GLCFS and NOS GLOFS are based on the Great Lakes Forecasting System (GLFS, generation 0), originally developed by The Ohio State University (OSU) and GLERL in the late 1980s and 1990s under the direction of Dr. Keith Bedford (OSU) and Dr. David Schwab (NOAA GLERL). The original forecasting systems utilize the Princeton Ocean Model (POM) and use a set of uniformly structured bathymetric grids (illustrated in the sidebar on structured versus unstructured grids for hydrodynamic computer modeling). The first routine nowcast, using a low-resolution grid for Lake Erie, began at OSU in 1992.

Starting in 2002, GLERL’s semi-operational GLCFS was expanded to five lakes using medium-resolution grids (5 – 10 km) and the 48-hr forecasts were added (generation 1). This version was successfully transferred from research to operations at NOS in 2010. The transition to operations at NOAA NOS was a joint effort between NOAA GLERL, NOS Center for Operational Oceanographic Products and Services (CO-OPS) and NOS Office of Coast Survey (OSC) Coast Survey Development Laboratory (CSDL), private industry (, and academia (OSU).

GLERL has continued to make improvements to the GLCFS; these include increasing the grid resolution (2 – 10 km), adding ice dampening and an ice model, and extending the forecasts to 120 hours during the period of 2006-2014 (generation 2). Examples of nowcasts and forecasts of Great Lakes conditions, such as waves, winds, and water temperature and ice (see the NOAA Great Lakes Coastal Forecasting System Factsheet).

Future Direction

The next generation of the GLCFS (generation 3) is being developed at NOAA GLERL through collaboration between GLERL, NOS CO-OPS and NOS CSDL. It is based on the latest version of the FVCOM, and uses a very high resolution (30 meters – 2 km) unstructured grid (i.e., triangular shapes of adaptable size) framework (illustrated in the sidebar on structured versus unstructured grids for hydrodynamic computer modeling). Examples of generation 3 FVCOM models are presented in Figures 1-3.

The 3rd generation of the GLCFS provides several features to enhance modeling capability in the Great Lakes and connecting channels, including:

These new features allow for improved predictions that better equip planners and managers responsible for components of the Great Lakes ecosystem affected by lake circulation and other physical process, such as lake level fluctuations, the transport of toxic materials, and nutrient enrichment processes. Current and future improvements to the GLCFS models have the capability to address many emerging issues in the Great Lakes, such as pathogen forecasting, harmful algal bloom (HAB) forecasts, contaminant spill tracking, coupled biological processes (e.g. invasive mussel impacts), ice coverage and thickness, and fisheries.

GLCFS next generation fig 3 - particle simulation

Figure 3: Dye tracers and particle simulations can be used to represent river water concentration, bacteria, and contaminants in conjunction with the GLCFS (3rd generation) to produce forecasts of material transport. Note the triangular elements used in the unstructured grid become increasingly dense closer to shore and near river mouths.

Project Timeline

Development of the generation 3 GLCFS and its transfer to operations at NOS GLOFS is part of a multi-year model implementation plan. Project research and development began in 2012 and the transfer to NOS operations has been initiated in 2015 with a projected completion in 2019. The models will be developed and implemented at the rate of one lake per year, starting with the Lake Erie Operational Forecasting System (LEOFS), followed by Lake Michigan-Huron (a combined-lake system), Lake Superior, and Lake Ontario. The transition from research to operations is an ongoing collaboration between NOAA’s OAR, the NOS CO-OPS and NOS OSC and NOS CSDL.

Project Design

The new GLCFS is based on the Finite Volume Coastal Ocean Model (FVCOM), a community oceanographic computer models that uses triangular elements (Figure 3) to represent conditions in the Great Lakes and connecting channels. FVCOM solves the three-dimensional (3-D), integral form of the equations of motion. Further information on the FVCOM is provided in the journal Oceanography article An Unstructured Grid, Finite Volume Coastal Ocean Model (FVCOM) System (Chen, et al., 2006).

At its core, FVCOM is a hydrodynamic computer model that predicts currents, temperature, and water levels. The FVCOM model is driven by hourly meteorological variables, such as air temperature, dew point, cloud cover, and wind speed/direction; data for these variables are collected from numerous observation stations around the Great Lakes. Model predictions are calibrated to observations from NOAA’s NOS gauges and NWS stations at several points:

Using skill assessment metrics (statistical analysis for accuracy of predictions) as defined by NOS, model performance is determined for each predicted variable. Upon successful simulation for several hindcast years (historical simulations), the model is implemented into a real-time nowcast (present conditions) and forecast (future conditions) mode, and then transferred to NOAA operations within NOS. The ultimate goal for model performance is to expend the least amount of resources as required to make acceptable predictions of physical conditions and processes in the Great Lakes.

Project Discoveries

Prediction of flow and oscillating currents in the Straits of Mackinac

The 3rd generation GLCFS includes a hydrodynamic model of Lake Michigan-Huron that uses observed and forecasted meteorology to predict flow and oscillating currents in the Straits of Mackinac, one of the most physically-complex areas in the Great Lakes. The inclusion of the Straits in the GLCFS model will help strengthen our understanding of the water exchange between Lake Michigan and Lake Huron, where there is a high level of current oscillation (Figure 4).

GLCFS next generation fig 4 - Volumetric Flow at the Straits

Figure 4: Volumetric Flow at the Straits where negative values represent flow from Huron to Michigan; and positive values represent flow from Michigan to Huron

Analysis of these weather driven forcing conditions and flow between the lakes reveals a Helmholtz mode - the natural resonance of the lake - that exists within the combined Lake Michigan-Huron system and is responsible for oscillation of currents between the lakes. Essentially, Lake Michigan-Huron acts like a constantly forced harmonic oscillator similar to blowing air across the opening of a bottle.

The FVCOM-based GLCFS model functions to simulate this resonance and predict the oscillating, bi-directional flow that occurs in the Straits. As illustrated in Figure 4, negative values represent flow from lakes Huron to Michigan and positive values represent flow from lakes Michigan to Huron.

Particle trajectory simulations in the Straits of Mackinac

To illustrate the effect of oscillating flow and the potential transport of a substance (e.g. oil, sediment, or nutrients), GLERL researchers conducted particle trajectory simulations in the Straits. These simulations (Figure 5) show that the hydrodynamics in the lakes transport particles from the Straits to northern Lake Michigan (west of the Straits) and to the southern shore of Lake Huron (east of the Straits).

GLCFS next generation fig 5 - Particle trajectory simulations

Figure 5: Particle trajectory simulations in the Straits of Mackinac

Drifter buoys released in the summer of 2014 confirmed this general circulation and particle pathways. In each release from the Straits, the drifter buoys made several passes back and forth through the channel (reaching speeds up to 50 cm/s, see Figure 6).

GLCFS next generation fig 6 - Pathway of drifter buoys

Figure 6: Pathway of drifter buoys (driven by current oscillation) in the Straits of Mackinac

Governmental and Societal Relevance of GLCFS Research

To reduce risks to and improve the health of the Great Lakes and their coastal communities, economy and ecosystems, we need to strengthen our understanding of the lakes’ complex and constantly changing physical, chemical and biological conditions. The GLCFS and its evolution has supported the transition of research results to operational products and tools helping us to better understand and predict how these changing conditions affect the Great Lakes system. In doing so, GLCFS modeling plays a critical role in promoting the health and beneficial use of the Great Lakes, including safe recreational swimming and boating, contaminant spill response, management of nutrient load targets, search and rescue, beach management, among others.

GLERL’s GLCFS nowcasts and forecasts of physical variables such as currents, temperature, ice cover, and water levels provide the public and decision makers with real-time information to aid in the usage and management of the Great Lakes. This knowledge is critically important given the role that physical conditions, such as current oscillation, plays in water quality and the transport of contaminants. The relevance of GLCFS research is demonstrated by GLERL's water quality projects that are currently focused on harmful algal blooms, hypoxia (oxygen depleted waters), and beach pathogens.

Through increasing model grid resolution and expanding the model domain, the new GLCFS (generation 3), and future generations to come, will continue to strengthen GLERL’s capacity to provide more accurate ecological forecasts for resource managers and the public.

Featured Publications and Other Related Products

Anderson, E.J., and D.J. Schwab. 2013. Predicting the oscillating bi-directional exchange flow in the Straits of Mackinac, Journal of Great Lakes Research 39(4):663-671 (2014).

Anderson, E.J., and D.J. Schwab. Contaminant transport and spill reference tables for the St. Clair River. Marine Technology Society Journal 46(5):34-47 (2012).

Anderson, E.J., D.J. Schwab and G.A. Lang. Real-time hydraulic and hydrodynamic model of the St. Clair River, Lake St. Car, Detroit River system. Journal of Hydraulic Engineering August 2010: 507-518 (2010).

Chen, C., R. C. Beardsley, and G. Cowles. 2006. An unstructured-grid, Finite-Volume Coastal Ocean Model (FVCOM) system. Advances in Computational Oceanography Vol. 19, No 1: 78-89 (2006).

Schwab, D.J., and K.W. Bedford. Initial implementation of the Great Lakes Forecasting System: A real-time system for predicting lake circulation and thermal structure. Water Pollution Research Journal of Canada 29 (2/3):203-220 (1994).

glcfs Research to Operations Poster

Table 1 (click to enlarge): GLCFS Research to Operations. Download PDF Copy


FVCOM: Finite Volume Coastal Ocean Model

GLCFS: Great Lakes Coastal Forecasting System

GLFS: Great Lakes Forecasting System

GLOFS: Great Lakes Operational Forecast System

HAB: Harmful Algal Bloom

NOAA: National Oceanic and Atmospheric Administration

NOAA GLERL: Great Lakes Environmental Research Laboratory

NOAA LEOFS: Lake Erie Operational Forecasting System

NOAA NDBC: National Data Buoy Center

NOAA NOS: National Ocean Service

NOAA NOS CO-OPS: Center for Operational Oceanographic Products and Services

NOAA NOS CSDL: Coastal Survey Development Laboratory

NOAA NOS OSC: Office of Coast Survey

NOAA OAR: Oceanic and Atmospheric Research

NOAA NWS: National Weather Service

NOAA NWS EMC: Environmental Modeling Center

NOAA NWS MMAP: Marine Modeling & Analysis Programs

NOAA NWS NCEP: National Centers for Environmental Prediction

OSU: The Ohio State University

POM: Princeton Ocean Model

R2O: Research to Operations