Close Window

Univ. of Michigan: " Improving marine forecasts over the Great Lakes"

Final Report


The overall project objective is the implementation within the National Weather Service Forecast Office(s) of the Great Lakes Region a modified version of an existing marine forecast system for the Great Lakes region for the purpose of improving wind and wave forecasts over the Great Lakes.

One secondary objective is to have a locally run, high resolution, regional numerical weather prediction model and marine forecasting system with output that will be accessible to NWS forecasters within the office AWIPS (Automated Weather Information and Processing System) environment.

Another secondary objective is to further develop the coupling aspects among the different model components in the existing system to account for changes in lake surface temperatures during the forecast and to account for the impact of waves on surface roughness and resultant surface fluxes.

This project is a collaborative effort between the Detroit/Pontiac National Weather Service Forecast Office (DTX), the Great Lakes Environmental Research Laboratory (GLERL), and the University of Michigan (UM). The University component has been transferred to Michigan State University (MSU).

DTX responsibilities include implementing a workstation version of the Eta model and then running it on a 2X per day schedule, using the wind output to drive a wave model, displaying that wave model output within AWIPS, and using the graphical capabilities to assist with marine forecasting over Lake Huron.

GLERL responsibilities include providing the wave model as well as assistance/expertise with implementing it and understanding its output.

University responsibilities include evaluating the accuracy of the wave model output using the higher resolution (Eta06) wind information.

To date, the Eta06 has been implemented within the DTX environment and is being used to drive the wave model. The information is available within the AWIPS environment. A verification of the output has been conducted for Julian Days 230-320 Y2000 and for Julian Days 140-320 Y2001 with additional evaluation of the forecasts from Y2002 underway.

Evaluation of time-dependent values of surface roughness (based on wave height) have indicated that effects are small so it may not be worthwhile to (two-way) couple the wave model to the Eta06. To this date the coupling of a lake-circulation model still remains; however, GLERL has begun the process of running the circulation model for Lake Huron at their facility.

Internal grid evaluation procedures are currently under development at DTX to offer on demand feedback of model system performance to identify short term biases. Additionally, improved grid verification techniques designed to measure the performance of mesoscale output are being developed and incorporated into the ongoing evaluation processes. Recent work at NSSL has presented evidence that traditional grid verification cannot produce meaningful results when evaluating mesoscale forecasts, the greater detail in higher resolution simulations penalizes the forecast with greater detail when traditional measures are employed. Therefore, similar techniques are being explored with relation to both model gridded forecasts and IFPS gridded forecasts.

Furthermore, in an effort to improve the timely delivery of the model output to forecasters DTX has made significant modifications to the Workstation Eta system to utilize the RUC-20 output for initialization and boundary conditions for the first 12 hours of the modeling system, then use boundary conditions from a coarser version of the Workstation Eta, locally run at DTX, that uses the AVN for initial and boundary conditions. Also, DTX has contracted to 2X per day with an extension to 84hours on the 12Z cycle to longer range marine forecasts implemented by the NWS in May 2002- open lake forecasts now extend to day 5. Finally, the strength of the locally run modeling system is the ability to adapt to the needs of the local forecast office, without an extensive review process, which would delay implementation.




Wagenmaker, R. B., 1999: Use of a High Resolution Mesoscale Model for Marine Forecasting in the Great Lakes. 8th U.S./Canada Workshop on Great Lakes Operational Meteorology, August 18-20, 1999, Ann Arbor MI, and NWS Central Region Great Lakes Marine Conference, December 6-8, 1999, Chicago, IL

Schwab, D. J., 1999: A Status Report on the Great Lakes Forecasting System (GLFS). 8th U.S./Canada Workshop on Great Lakes Operational Meteorology, August 18-20, 1999, Ann Arbor MI, and NWS Central Region Great Lakes Marine Conference, December 6-8, 1999, Chicago, IL.

Wagenmaker, R. B., 2000: Applications of High Resolution Mesoscale Wind and Wave Models in the Great Lakes. NWS Central Region Great Lakes Marine Conference, September 26-27, 2000, Chicago, IL

Wagenmaker, R. B., 2000: Use of High Resolution Mesoscale Models for Marine and Other Forecast Applications. NWS Central Region SOO Conference, August 28-September 1, 2000, Kansas City, MO.

Schwab, D. J., 1999: The Great Lakes Forecasting System (GLFS). NWS Great Lakes Regional Marine Conference, May 15-16, 2001. Ann Arbor, MI.

Mann, G. E., and R. B. Wagenmaker, 2001: Local Area Models and Marine Forecasting Initiatives at WFO DTX. NWS Great Lakes Regional Marine Conference, May 15-16, 2001. Ann Arbor, MI.

Sousounis, P.J., G. E. Mann, D. J. Schwab, and R. B. Wagenmaker, 2001: Development of a Coupled Air-Lake Mesoscale Model For Operational Marine Forecasting in the Great Lakes Region: Some Initial Verification Results. NWS Great Lakes Regional Marine Conference, May 15-16, 2001. Ann Arbor, MI.

Mann, G. E., and R. B. Wagenmaker, 2003: Marine Forecasting Initiatives utilizing GFE at WFO DTX. NWS Great Lakes Regional Marine Conference, June 10-12, 2003. Buffalo, NY.


Sousounis, P.J., G. E. Mann, D. J. Schwab, and R. B. Wagenmaker, 2001: Development of a coupled air-lake mesoscale model for operational marine forecasting in the Great Lakes Region. Preprints, 18th Conference on Weather Analysis and Forecasting, Fort Lauderdale, FL, AMS, J31-J35.

Currently working jointly on a publication for Weather and Forecasting


University Benefits

A much better appreciation of the difficulty in forecasting wind direction. Both the operational Eta (22 km) and the DTX version of the Eta (06 km) indicate significant RMS errors (25°) in forecasting wind direction.

A much better understanding of how the Eta06 performs relative to the Eta32. To date, the results indicate that wave height forecasts as verified at the two buoy locations on Lake Huron (45003 and 45008) are slightly worse from the Eta06 than from the Eta32. These results are extremely curious and are being investigated further. The Eta06 does provide better forecasts over certain regions of Lake Huron during certain synoptic situations. For example, during cool or cold air outbreaks over the Great Lakes in early autumn, the higher resolution in the Eta06 simulates the surface heat fluxes more accurately, which allows the model to simulate more accurately the lake aggregate effect (e.g., aggregate scale cyclonic vortex) that develops over the lakes in such situations. The more accurate forecast of this phenomenon allows a more accurate forecast of wave heights.

NWS Benefits

The daily use and subjective evaluation of products from the mesoscale/hydrodynamic modeling system have increased forecaster awareness of lake-atmosphere interactions. Specifically the modeling system has reaffirmed basic principles and offers greater insight into the real time impacts of surface layer stability and wave generation, fetch length and its relation to wave growth and decay, wave age and wave train evolution, and wave growth in regions of channeling and funneling. Additionally, the greater detail and structure has resulted in a higher level of interrogation/understanding of both available model and observational datasets during the forecast processes, which enables forecasters to portray greater detail in the forecast product. In many instances, this level of detail has been supported by transient observations (e.g., ships), which further lends credence to the use of the modeling system in an operational setting. Ultimately, the additional guidance from the modeling system increases the forecasters’ ability to discern the onset, coverage, and duration of threshold events- small craft advisories, gale/storm warnings.

Unfortunately, forecast improvements cannot be measured effectively at this time with the current set of metrics (single point verification, two observations per day- 06Z and 18Z, when buoy data is available). Subjectively, forecast accuracy and certainly precision have improved with the implementation of the modeling system. This perceived improvement is likely a consequence of a combination of the high resolution output from the modeling system with the increased attention to detail by the forecasting staff. Certainly, greater precision is now being offered to the marine forecast user within the current forecast construct is increasing as graphical marine products are being offered and will become the standard in the near future. Finally, the efforts associated with this project have led to the participation of DTX in the development of a national grid based marine verification system for AWIPS/IFPS.

Additionally, the knowledge of wave forecasting techniques has expanded WFO DTX’s capabilities in producing gridded wave forecasts using the Graphical Forecast Editor (GFE) with in the AWIPS/IFPS system. Producing wave height forecasts beyond the guidance period provided by the GLCFS and the local wave model output requires use of the GFE smart tools. More importantly, whenever the forecasted winds differ from the Eta12 or Eta06 guidance, the wave height grid must also be recalculated. This invariably happens every day. In an effort in making wave height forecasts consistent with the wind forecasts, we have developed these set of smart tools that can be used to calculate wind wave heights based on the forecasters wind grids. Additionally, it offers the flexibility of using the forecasters wind grid to produce wave heights, rather than attempting to manipulate the guidance grids values. We have had great success with these tools.