Atmosphere-Lake Coupling Improves Lake-Effect Snowfall Forecasts in the Great Lakes Region

The paper focuses on the coastal Great Lakes region which is often considered the ‘Third Coast’ of the United States. The formation of lake-effect snowfall downwind of the Laurentian Great Lakes has been shown to be highly sensitive to lake surface conditions. Attempting to accurately capture the spatial and temporal variations in lake temperature and ice distributions is paramount to successfully predicting the timing and intensity of the snowfall. During the winter months, cloud cover over the lakes often limits or delays the observations of these variations which restricts the use of remotely sensed products. This motivates the need for model-based lake surface information in weather prediction models.
We evaluate the forecast skill of the NOAA Unified Forecast System (UFS), and in particular its Short-Range Weather Application (UFS-SRW), in simulating lake-effect snowfall in the Great Lakes region when providing the lake surface characteristics generated from the Finite Volume Community Ocean Model (FVCOM) hydrodynamic model. Updates to the UFS-SRW modeling framework have been made to allow for the lower boundary conditions (lake surface temperature, lake ice fraction, and lake ice temperature) of the atmospheric model to be updated hourly over the forecast horizon. Hourly-updated and temporally-static lake surface conditions provided by FVCOM are then compared to remotely sensed lake surface conditions to show the difference in forecast skill across the three methods. It is shown that the chosen loose coupling technique between the UFS-SRW and FVCOM has potential to advance NOAA’s lake-effect forecasting capabilities which will be further tested in UFS’s Medium-Range Weather Application (UFS-MRW) in the future.