Improved Predictability of Lake-Effect Snowfall in the UFS Short-Range Weather Application Through the Use of Modeled Lake Surface Conditions

The prediction of lake-effect snowfall in numerical weather models requires accurate representations of the atmosphere (temperature and moisture profiles, wind direction, etc.) and lake surface (surface temperature, ice cover) in the atmospheric model. Lake surface conditions have historically been derived from satellite-based measurements where the realization of changes in these conditions can be delayed in the cold season due to extended periods of cloud cover. This time-lag can introduce biases in the atmospheric model that results in errors in snow band placement and/or intensity. To better represent the lake surface in the atmospheric model, output from the Great Lakes Coastal Forecasting System (GLCFS), a coupled hydrodynamic-ice model, is applied to the Unified Forecasting System’s Short-Range Weather Application (UFS-SRW) for a lake-effect snowfall case study from November 2022. The GLCFS is applied to the UFS-SRW using two techniques to test the need for changing lake surface conditions over a short forecast horizon: temporally static and hourly updating. Both updating techniques showed relatively similar skill in the placement of snowfall bands and showed improvements over other lake surface conditions. Use of the hourly updating lower boundary conditions shows an improvement in the intensity of the snowfall downwind of the lakes compared to temporally static conditions. All UFS-based simulations displayed a consistent error which resulted in excessive snowfall melt downwind of Lake Michigan. Additional sensitivity studies, including modifications to surface roughness lengths, were conducted to better understand the drivers behind snowfall melt.