Summary
I will review current efforts to model the Antarctic ice pack. Sea ice models in GCMs have become more sophisticated in recent years; state-of-the-art models now include viscousplastic dynamics, second-order-accurate advection, multilayer thermodynamics, and multicategory thickness distributions. However, these models generally do not have realistic treatments of snow-ice formation, lateral melting, or sea ice biology. I will focus on one such model, CICE, which is used in the Community Climate System Model (CCSM). In ice-only runs using CICE, the broad features of the Antarctic ice pack can be reproduced well if the deep-ocean heat flux is tuned to limit wintertime ice growth. In coupled runs with CCSM, the ice pack is realistic overall, although the ice is too extensive in the South Atlantic and probably too thick in the Weddell Sea. Validation of ice thickness is difficult because of the sparseness of data. In greenhouse scenarios, Antarctic ice volume decreases more rapidly than ice extent, suggesting that thickness may be a better climate change indicator. The most important positive feedback is increased solar absorption in open water; the ocean heat flux convergence does not increase in most regions.