Utilizing Object-Based Identification to Verify UFS Snowfall Forecasts using Snow-to-Liquid Ratio Methods

A primary challenge of forecasting on a seasonal-to-subseasonal (S2S) timescale is extracting meaningful forecast skill at extended forecast hours. This project develops a novel object-based verification techniques to identify regions within the CONUS that are forecasted to receive snowfall above a specified threshold in a model forecast, and verify these forecasts against objects produced from an observational dataset. Object properties, such as area, intensity, orientation, and solidity, were used to compare the properties of model forecast and observation events. Data from a retrospective UFS S2S prototype were used to create snowfall objects, which were compared to snowfall objects generated using Stage IV liquid precipitation, paired with multiple Snow-to-Liquid Ratio methods. This presentation will discuss the methods used to identify snow events, combine small objects that are associated with an event, and pair model forecast and observation objects to ensure the object attributes being compared were from the same event.