Incorporating Satellite Ice Data for Enhanced Operational Forecasting : Bridging Ice Floe Trends and Dynamic Parameterization

Ice conditions, integral to accurate weather prediction, maritime operations, and oil spill response, are challenging to forecast, especially in dynamic environments such as the Great Lakes. Existing operational forecast models of the Great Lakes tend to overestimate ice thickness in late spring due to shortcomings in physical parameterizations. Here, we probe the efficacy of incorporating satellite-derived ice conditions from the National Ice Center (NIC) into coastal operational systems and evaluate the impact on forecast skill. NOAA’s Great Lakes Operational Forecast System (GLOFS) leverages the coupled Finite Volume Community Ocean Model (FVCOM) and Los Alamos Sea Ice Model (CICE) for real-time prediction. Our study tests improvement to this system using three simulation scenarios: baseline operations (constant ice floe size), a climatologically informed model (spatially-varying ice floe size climatology), and a dynamic model (spatio-temporal ice floe size variability) that uses the daily NIC products. Notably, this dynamic approach seeks to address discrepancies like the under-representation of lateral melting and the assumption of a static 300 meter ice floe diameter. Our findings underscore the utility of satellite-based ice data integration to improve forecast accuracy. While the existing operational setup shows reasonable prediction of ice concentration through most of the season, it falters during spring, yielding lags compared to observed ice conditions. In contrast, the dynamic ice floe parameterization shows the promise of improvement in forecast skill. By shedding light on the pivotal role of near real-time satellite data in enhancing model accuracy, our research proposes an improved ice parameterization for operational ice prediction.