Abstract To satisfy the needs of the Arctic sea ice seasonal prediction in summer, we developed the Arctic Seasonal Prediction System (ArcSPS). This system integrates an Arctic sea ice–ocean–atmosphere coupled model with an ensemble‐based Kalman Filter data assimilation model. The data assimilation model has a capacity of assimilating sea ice concentration, sea ice thickness, and sea surface temperature (SST) observations simultaneously into the coupled model state, providing an optimal state estimate at prediction onset. The mean September Integrated Ice Edge Error (IIEE) in the Arctic Ocean from totally 180 prediction runs, initialized on the 1st day of June, July, and August during 2013–2022, are 1.53 × 10 6 km 2 , 1.4 × 10 6 km 2 and 1.32 × 10 6 km 2 , respectively, which are lower than those of most dynamic models reported in Bushuk et al. (2024), https://doi.org/10.1175/bams‐d‐23‐0163.1 , indicating that the ArcSPS has a promising performance on the September sea ice prediction at lead times of 1–3 months. The predicted sea ice evolutions in summertime are basically consistent with the observations regarding both spatial patterns and total metrics. The predicted September sea ice edge and thickness distributions agree well with the observations. To further improve seasonal sea ice prediction, efforts should be concentrated on assimilating more observational variables and utilizing more reasonable air‒ice‒ocean interaction representations in high‐resolution models.
Tian et al. (Wed,) studied this question.