Uncertain descriptions of Arctic climate processes in coupled models and their impact on the simulation of Arctic sea ice

• Main Authors: Dorn, Wolfgang, Dethloff, Klaus, Rinke, Annette, Gerdes, Rüdiger, Karcher, Michael, Kauker, Frank, Frickenhaus, Stephan
• Format: Conference Object
• Language: unknown
• Published: 2007
• Subjects: albedo; Arctic; Arctic Ocean; Global warming; Sea ice
• Online Access: https://epic.awi.de/id/eprint/18195/; https://hdl.handle.net/10013/epic.28730

The presence of an ice covered Arctic Ocean plays an important role inthe Arctic climate system by its influence on the exchange betweenatmosphere and ocean. The interactions between atmosphere, ocean, andsea ice are determined by a couple of feedback processes which are notyet completely understood. For this reason, coupled climate modelshave still difficulties in reproducing present-day sea-ice conditionsand their variability close to reality. The outcome of this is a largescatter in the simulated sea-ice cover and thickness among differentmodels, which is further amplified when applying the models to climatechange scenarios where an Arctic amplification of global warming isexpected as a result of diminishing sea-ice cover.Sensitivity experiments with the coupled regional climate modelHIRHAM-NAOSIM have shown that uncertainties in the description forArctic clouds, snow and sea-ice albedo, lateral freezing and meltingof sea ice, and a snow cover on sea ice are responsible for largedeviations in the simulation of Arctic sea ice. While moresophisticated schemes for the albedo, the treatment of lateralfreezing and melting, and the snow cover have already beensuccessfully introduced into the model, the parameterization of cloudsis still an open issue. Currently the model overestimates the cloudcover during winter associated with a warm temperature bias and toolow ice growth during the cold season. The outcome of this is athinner ice cover at the beginning of the melting period which tendsto disappear more quickly than observed. This has strong consequencesfor the model performance at large, since feedback processes mayfurther amplify biases in the coupled model system.