Using a Coupled FV3GFS–FVCOM Modeling System to Improve Lake-Effect Snowfall Forecasts

Lake-effect snowfall can be challenging to capture in numerical weather predictions (NWPs) due to the fine-scale nature of the phenomena and its significant sensitivity to the lake and atmospheric environments. Successfully depicting lake-effect snowfall placement and intensity in NWPs requires that the lake surface characteristics, including lake surface temperature and ice cover, be accurately represented in the atmospheric model. Currently, most NWPs use temporally static lake surface characteristics assigned at model initialization which can lead to poor forecasts for longer forecast horizons or in cases of rapidly evolving lake characteristics. To mitigate some of these errors, an asynchronous iterative coupling of NOAA’s latest atmospheric model (FV3GFS) with a lake hydrodynamic and ice model (FVCOM) is used to provide a more realistic lower boundary condition, both spatially and temporally, for the atmospheric model. FV3GFS, a global atmospheric model, is run at a variable resolution to achieve a horizontal resolution of approximately 3km over the Great Lakes region in order to begin to explicitly represent the snowfall bands, while FVCOM is run on an unstructured grid with horizontal grid spacing near the coasts of approximately 3km. The alterations in NWPs using the new lake surface conditions will be shown through a series of case studies, including the polar vortex of early 2019. This coupling approach is also used to compare to similar couplings of atmospheric models and FVCOM and as an intermediate step to show the scale of improvements which can be made. The ultimate goal is to create a dynamically coupled modeling system for investigating the interactions between the lake and atmospheric environments across the region.