Abstract
•FVCOM-based Arctic Ocean forecast system is being developed for the first time.•Sea ice concentration, thickness, and drift forecasts assessed against climatology.•Identified key forecast errors from thermodynamics, dynamics, and ice initialization.•Competitive performance compared to existing Arctic forecast models.
A developing Finite Volume Community Ocean Model-based Arctic Ocean forecast system (FVCOM-AOFS) is presented. At the current stage, daily free-run sea ice forecasts with a 24-hour lead time over 2019–2020 are evaluated against climatology on the monthly timescale. Process-oriented experiments explore the impacts of thermodynamic and dynamic factors, as well as sea ice initialization, on forecast errors. The forecast model reasonably reproduces the major spatiotemporal patterns of sea ice concentration (SIC), thickness (SIT), and drift (SID) without data assimilation, although some shortcomings remain. The forecast generally outperforms climatology in SIC and associated sea ice extent. SIC errors mainly occur near the sea ice edge, where inaccurate forecasts of surface air temperature and net heat flux lead to overestimations. For SIT, the model does not surpass climatology, with errors primarily due to inaccuracies in the initial SIT. However, correcting SIT initialization can improve the forecast over a seasonal period. The assessment of SID shows that the forecast model exhibits better skill than climatology, as expected. However, errors in SID speed and direction are positively and negatively correlated with wind speed, respectively, making simultaneous improvements in these areas challenging. A comparison between FVCOM-AOFS and other existing forecast models is also conducted. Overall, these comprehensive assessments demonstrate notable potential for further enhancement, with future efforts toward several key aspects including improvements in atmospheric forcing forecast and sea ice initialization.