Comparison of elastic-viscous-plastic and viscous-plastic dynamics models using a high resolution Arctic sea ice model

Abstract

A nonlinear viscous-plastic (VP) rheology proposed by Hibler (1979) has been demonstrated to be the most suitable of the rheologies commonly used for modeling sea ice dynamics (Kreyscher et al., 1997). However, the presence of a huge range of effective viscosities hinders numerical implementations of this model, particularly on high resolution grids or when the ice model is coupled to an ocean or atmosphere model. Hunke and Dukowicz (1997) have modified the VP model by including elastic waves as a numerical regularization in the case of zero strain rate. This modification (EVP) allows an efficient, fully explicit discretization that adapts well to parallel architectures. We present a comparison of EVP and VP dynamics model results from two 5-year simulations of Arctic sea ice, obtained with a high resolution sea ice model. The purpose of the comparison is to determine how differently the two dynamics models behave, and to decide whether the elastic-viscous-plastic model is preferable for high resolution climate simulations, considering its high efficiency in parallel computation. Results from the first year of this experiment (1990) are discussed in detail in Hunke and Zhang (1997). The model equations were solved on a 300x360 mesh with a resolution of about 18 km and a 4 hr timestep, and driven by 1990-1994 ECMWF atmospheric data above and 1992 ocean model output below. The two runs are identical except for the ice rheology; the EVP and VP dynamics models used here are documented in Hunke and Dukowicz (1997) and Zhang et al. (1997), respectively. For further details about the model specifications and experiment design, see Hunke and Zhang (1997).