Thursday, 1 February 2024
Hall E (The Baltimore Convention Center)
Arctic sea ice is diminishing at a faster rate than anticipated, altering the air-sea exchange of mass, heat, and momentum. Variability in surface energy fluxes impacts the Arctic surface energy budget, sea ice cover, and atmospheric and oceanic circulations. Our ability to describe these physical processes is limited by challenges of Arctic in situ observations and modeling. Advancements in uncrewed surface vehicles provide new opportunities to validate ensemble forecast products. We use observations from saildrones deployed in the Bering, Chuckhi, and Beaufort Seas in June–September 2019 to evaluate European Centre for Medium-Range Weather Forecasts (ECMWF) ensemble forecasts of air-sea fluxes and related meteorological variables. ECMWF surface forecasts have a spatial resolution of 0.5° x 0.5° with 50 ensemble members and six-hourly output with lead times up to 15 days. We linearly interpolate gridded ensemble model data onto temporally coincident saildrone trajectories. We evaluate model reliability and accuracy using model-observation correlations, spread-skill relationships, rank histograms, and probability density functions. Interpolated ensemble surface fluxes are converted into binary probabilities based on extreme flux events and assessed using reliability diagrams. Our evaluation suggests flux forecasts are underdispersed, unskillful after 4-7 days, and unreliable for predicting extreme fluxes. We conduct a time series regression analysis to identify sources of error in forecasted fluxes, finding that sensible heat flux errors are largely caused by errors in forecasted air-sea temperature differences, while latent heat flux errors have similar contributions from errors in forecasted wind speeds and air-sea humidity differences.

