The impact of denying sea ice thickness initial information on the water mass formation in the Arctic Ocean

The importance of the Arctic sea ice initial conditions for the water mass formation in the Arctic Ocean is assessed within a perfect model experiment framework. To do so, we used model outputs from EC-Earth3 and HadGEM3 – two General Circulation Models (GMCs) – as true reference states. Long-term control runs were generated with both models based on present-day forcing. From the control runs, we first identified the period of the year in which the heat conductive flux in sea ice is maximum. Such condition takes place in early winter (December-January), when ice is still thin, snow depth is small, and the air-ocean temperature contrasts the most. Once this maximum heat exchange was identified, we selected two different Januarys in which the Arctic sea ice volume was (i) maximum and (ii) minimum. For each of the two identified dates, two counterfactual sets of experiments were performed. First, we restarted the run from the same initial conditions of ocean, sea ice, and atmosphere. Second, we restarted the runs from the same initial conditions of ocean and atmosphere but with climatological sea ice thickness. Climatological sea ice thickness was estimated from the control run outputs. To robustly separate the response to degrading the initial sea ice state from background internal variability, each of the two counterfactual experiments consisted of 50 ensembles members. These ensembles were generated by adding small random (O[10^-4K]) perturbations to the sea surface temperature (SST) field for the EC-Earth experiments, and to a wind component for the HadGEM3 experiments. Preliminary results reinforce the importance of an accurate initialization of the sea ice state for improving the representation of the water mass formation in the Arctic Ocean.