The sensitivity of CSIM5 sea-ice simulations to atmospheric state variables

• Main Author: Lemcke, Mary D.
• Format: Thesis
• Language: English
• Published: University of Delaware 2009
• Subjects: Physical oceanography|Atmospheric sciences; Antarctic; Ross Sea; Southern Ocean; The Antarctic; Antarc*; ice pack; Sea ice
• Online Access: http://pqdtopen.proquest.com/#viewpdf?dispub=3373054

The response of a simulated sea-ice distribution to different numerical approximations of atmospheric state variables is investigated with respect to the improvement of sea-ice model physics within coupled climate system models (CSM). Model representation of the Antarctic sea-ice environment within CSM's has been limited by poor model physics and a lack of continental-scale, sea-ice thickness observations for model evaluation. The availability of continental-scale thickness observations was addressed using sea-ice thickness estimates derived from weekly, National Ice Center (NIC) operation sea-ice charts for the Southern Ocean from 1995 to 1998. Previous studies reported a good agreement between Antarctic seaice thickness estimates derived from NIC observations and in situ sea-ice thickness observations within the Ross Sea. Prediction errors determined for sea-ice thickness estimates derived from the NIC dataset indicate a mean absolute error (MAE) less than ±10 cm, an index of agreement (d1) of 0.94, and resolution errors < 20 cm for commonly used model resolutions. The response of the sea-ice distribution produced by the National Center for Atmospheric Research (NCAR) Community Sea Ice Model, Version 5 (CSIM5) to atmospheric energy fluxes is evaluated using NIC observations. The simulated sea-ice environment responded strongly to the value of atmospheric specific humidity due to the parameterized numerical approximation of the latent heat flux used by the CSIM5. The CSIM5 produced a thinner ice pack with a lesser extent in response high atmospheric moisture, as indicated by extensive areas of large (> 50 cm), negative model prediction errors. The model produced a thicker, more extensive ice pack in response to low atmospheric moisture as indicated by even larger (> 200 cm), positive model prediction errors.