Jinlun Zhang and Drew Rothrock
Polar Science Center, APL/UW, Seattle, WA 98105
July 1998
1. Efficient Numerical Model for Sea Ice Dynamics. Commonly used numerical models for sea ice dynamics, such as those of Hibler (1979) and Zhang and Hibler (1997), require an iterative procedure to solve the sea ice momentum equations, which is not desirable for parallel computing. As an approximation of viscous plastic rheology, the elastic viscous plastic (EVP) model of Hunke and Dukowicz (1997) is suitable for parallel computing. But this model artificially introduces unphysical elastic waves into the modeled sea ice system so that the simulated ice motion has a large departure from a viscous plastic solution. This is why the EVP model has been found by the Sea Ice Model Intercomparison Project (see http://www.ifm.uni-kiel.de/me/SIMIP/simip.html for details) to be undesirable for sufficiently accurate ice modeling. Here we present a numerical model of Zhang and Rothrock (1998) for solving sea ice momentum equations with a nonlinear plastic rheology. This model is based on an alternating direction implicit technique that involves a direct solution of the momentum equations without an iterative procedure. It is much more computationally efficient than those models that require an iterative procedure. Tables 1 and 2 show the CPU time consumed solving ice momentum equations using different numerical dynamics models. In addition, with no iterative computations, the model can be easily adapted to parallel computing for higher efficiency. Therefore, it is particularly suitable for high-resolution modeling of sea ice. Another important feature is that the model approaches a true rigid plastic solution while previous numerical models may or may not reach a true viscous plastic solution. It was found that a viscous plastic solution tends to predict excessive ice stoppage in the Arctic, whereas the rigid plastic solution does not.
2. Modeling Sea Ice with Oriented Leads. One of the challenging goals of PIPS 3.0 is an implementation of "a lead resolving ice model." Hibler and Schulson (1997) recently developed a model with isotropic realizations of dynamical treatment of oriented leads. Their "strain hardening" model, in which only oriented leads having the potential to open rapidly are allowed, appears to be particularly attractive. As an initial step toward a more complicated anisotropic model, this "strain hardening" model may be implemented into PIPS 3.0 unless there is a more practical model available that includes more physics of anisotropic flow of sea ice. One of the features of the "strain hardening" model is that it follows a teardrop-like yield curve but may or may not obey the normal flow rule. We have done some work to develop plastic sea ice rheologies with a variety of teardrop-like yield curves with or without obeying the normal flow rule. These rheologies may be useful for implementing the "strain hardening" model.