Abstract: As the climate warms in the Arctic region, the sea ice extent (SIE) in the Arctic Ocean is rapidly decreasing and retreating northward, making Arctic sea ice forecasting increasingly important for scientific expeditions and commercial navigability assessments. This study utilizes hindcast data from the First Institute of Oceanography-Climate Prediction System (FIO-CPS) v1.0. After applying bias correction to the Arctic sea ice predictions, the system’s predictive capability for summer (June–September) Arctic sea ice extent and sea ice concentration (SIC) was evaluated. The evaluation results indicate that: (1) There is a systematic bias in the FIO-CPS v1.0 predictions of summer Arctic SIE. After bias correction, the prediction performance improves significantly, with the most notable improvement observed in July; (2) Following bias correction, FIO-CPS v1.0 demonstrates good performance in predicting Arctic SIE, with root mean square errors for both monthly and daily predictions below 0.5×10⁶ km², and anomaly correlation coefficients exceeding 0.7; (3) The prediction skill of FIO-CPS v1.0 is significantly higher than that of linear trend climatology predictions and, to some extent, superior to damped anomaly persistence predictions. However, this advantage gradually diminishes as the lead time increases; (4) During years with extremely low Arctic SIE, the prediction performance of FIO-CPS v1.0 is relatively weaker, but it still holds a significant advantage over linear trend climatology predictions; (5) After bias correction, FIO-CPS v1.0 also exhibits high predictive capability for SIC, with the root mean square error of predicted sea ice concentration generally below 30%.

Key words: short-term climate prediction system, FIO-CPS v1.0, assessment of sea ice prediction skills, sea ice concentration, Arctic