The Challenge of Arctic Sea Ice Thickness Prediction by ECMWF on Subseasonal Time Scales

Abstract A recent study has provided the first thorough assessment of subseasonal Arctic sea ice edge prediction in operational forecast systems. However, the corresponding assessment of the Arctic sea ice thickness (SIT) is still lacking. Here, the Arctic SIT reforecasts from the European Centre for Medium‐Range Weather Forecasts (ECMWF) on subseasonal time scales are evaluated for the first time using a well‐developed SIT reanalysis data set. The results show that ECMWF forecasts of Arctic SIT are more skillful than persistence forecast (PFs) during the transition seasons for lead times longer than 30 days, indicating the advantages of dynamical prediction. However, ECMWF suffers from large initial SIT errors and has lower skill than a PF, especially from March to June. Thus, subseasonal Arctic SIT predictions still face many challenges, and improving the mean SIT in the model and assimilating satellite‐estimated SIT in the ocean analysis would significantly improve SIT forecasts.