Wednesday, 9 January 2013: 8:45 AM
Ballroom B (Austin Convention Center)
Arctic sea ice has become one of the most rapidly-changing components of the global climate system. Due to its nature, sea ice is much more sensitive to alteration of external forcing than are many other climate-system components. It plays significant roles in modulating global climate energy and water balance, and integrates interdependent processes of global energy and water cycles. Accordingly, sea ice is considered to be an outstanding indicator of global climate-system changes. However, large uncertainties emerged in the prior CMIP3 models, which generally underestimated observed Arctic sea ice declining trends and did not capture the record low summer sea ice in 2007. The large spreads also occurred across the models and model ensembles. Understanding present sea-ice conditions, improving their predictability, and reducing uncertainty in projecting their future states have therefore become one of the hotly debated topics in scientific discourse. Here we synthesize and extend our previous approaches and conducted analyses on the multi-model and multi-model ensembles in the CMIP5 experiments. The statistical evaluation against observational data suggests that the CMIP5 models have improved sea ice simulations during the 20th century on a number of aspects. The simulated summer Arctic sea ice areas from different models and model ensembles become narrowed compared with those in CMIP3, which would benefit from the extreme summer sea ice loss in 2007. However, the winter sea ice areas still exhibit large diversity. The sensitivity analysis demonstrated a reduction of the overall feedback strength across the models and model ensembles. Comparison of sensitivity terms between the model ensemble members and observations provides a physically based constraint to reduce model uncertainties in future sea ice change projections.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner