Modeling the thickness distribution of Arctic sea ice

Thesis (Ph. D.)--University of Washington, 1998. Arctic sea ice may be unusually sensitive to greenhouse warming, thanks to ice-albedo feedback. Most climate models represent sea ice crudely, with each grid cell occupied by ice of uniform thickness together with open water, when in reality the ice r...

Full description

Bibliographic Details
Main Author: Lipscomb, William H
Format: Thesis
Language:English
Published: 1998
Subjects:
Online Access:http://hdl.handle.net/1773/10081
Description
Summary:Thesis (Ph. D.)--University of Washington, 1998. Arctic sea ice may be unusually sensitive to greenhouse warming, thanks to ice-albedo feedback. Most climate models represent sea ice crudely, with each grid cell occupied by ice of uniform thickness together with open water, when in reality the ice ranges in thickness from a few centimeters to tens of meters. This study aims toward an improved understanding of how thermodynamic and dynamic processes interact to determine the thickness distribution of perennial sea ice in the central Arctic.Two sea ice models are developed and tested. The first is a heuristic model with simple parameterizations of growth, melting, ridging, and export. The model resolves a large number of ice categories, each spanning only a few centimeters in thickness. With linear forcings, the model can be formulated as a Markov process, whose steady-state solution and transient modes are computed using matrix algebra techniques. The steady-state solution reproduces the main features of observed thickness distributions. In its linear form the model is not very sensitive to warming, because increased ridging offsets the effects of enhanced melting. However, if the melting rate is assumed proportional to the open water area--which makes the model nonlinear--the sensitivity to warming increases sharply. Critical to this large sensitivity is a melting rate that increases as the ice thins.The second model is thermodynamic. It accounts for brine pockets and corrects an error in the way previous models have computed surface melting. Simple parameterizations of export and ridging are included. The ice pack is divided into unridged multiyear ice, ridged ice, open water, and an adjustable number of first-year ice categories, with mass and energy conserved when ice is transferred among categories. Three to five categories are sufficient to obtain converging results. With standard parameters and five categories, the equilibrium mean ice thickness is 3.4 m. An increase of 8 W m$\sp{-2}$ in the advected ...