Rapid summertime sea ice melt in a coupled numerical weather prediction system

Coupled Numerical Weather Prediction (NWP) models have only recently been implemented for short-term environmental prediction and both challenges and benefits are evident in polar regions. Their simulation of surface exchange over sea ice depends on the model's sea-ice characteristics, however these are hard to constrain due to a lack of in situ and accurate remotely sensed observations. We focus on the Fram Strait region during peak melt conditions and during the passage of an Arctic cyclone: very challenging conditions for coupled NWP. We use in situ aircraft observations from the Arctic Summertime Cyclones field campaign in July-August 2022, plus satellite products, to evaluate a set of 5-day forecasts from the Met Office Unified Model. Our model set ups are based on operational GC4 (Global Coupled 4) and developmental GC5 (Global Coupled 5) configurations, which use the CICE5.1 and SI3 sea-ice models respectively. We find a combination of deficiencies in the simulated sea-ice field, due to initialization and modeling problems. An initially low concentration of sea ice results in excessive absorption of shortwave radiation by the ocean, leading to excessive basal melting of the sea ice, and further sea-ice loss; leading to relatively poorly simulated sea-ice fields in general. In contrast, the passage of an Arctic cyclone and its impact on sea-ice velocities are captured well. Although we demonstrate several deficiencies in the short-term forecasts of two state-of-the-art coupled NWP models, we also find promising aspects of model performance and some clear benefits from a fully coupled atmosphere-ice-ocean system.