TY  - JOUR
A1  - Mehlmann, C.
A1  - Danilov, S.
A1  - Losch, M.
A1  - Lemieux, J. F.
A1  - Hutter, N.
A1  - Richter, T.
A1  - Blain, P.
A1  - Hunke, E. C.
A1  - Korn, P.
T1  - Simulating Linear Kinematic Features in Viscous‐Plastic Sea Ice Models on Quadrilateral and Triangular Grids With Different Variable Staggering
Y1  - 2021-10-28
VL  - 13
IS  - 11
JF  - Journal of Advances in Modeling Earth Systems
DO  - 10.1029/2021MS002523
PB  - 
N2  - Observations in polar regions show that sea ice deformations are often narrow linear features. These long bands of deformations are referred to as Linear Kinematic Features (LKFs). Viscous‐plastic sea ice models have the capability to simulate LKFs and more generally sea ice deformations. Moreover, viscous‐plastic models simulate a larger number and more refined LKFs as the spatial resolution is increased. Besides grid spacing, other aspects of a numerical implementation, such as the placement of velocities and the associated degrees of freedom, may impact the formation of simulated LKFs. To explore these effects this study compares numerical solutions of sea ice models with different velocity staggering in a benchmark problem. Discretizations based on A‐,B‐, and C‐grid systems on quadrilateral meshes have similar resolution properties as an approximation with an A‐grid staggering on triangular grids (with the same total number of vertices). CD‐grid approximations with a given grid spacing have properties, specifically the number and length of simulated LKFs, that are qualitatively similar to approximations on conventional Arakawa A‐grid, B‐grid, and C‐grid approaches with half the grid spacing or less, making the CD‐discretization more efficient with respect to grid resolution. One reason for this behavior is the fact that the CD‐grid approach has a higher number of degrees of freedom to discretize the velocity field. The higher effective resolution of the CD‐discretization makes it an attractive alternative to conventional discretizations.
N2  - Plain Language Summary: Sea ice in the Arctic and Antarctic Oceans plays an important role in the exchange of heat and freshwater between the atmosphere and the ocean and hence in the climate in general. Satellite observations of polar regions show that the ice drift sometimes produces long features that are either cracks (leads) and zones of thicker sea ice (pressure ridges). This phenomenon is called deformation. It is mathematically described by the non‐uniform way in which the ice moves. For numerical models of sea ice motion it is difficult to represent this deformation accurately. Details of the numerics may affect the way these models simulate leads and ridges, their number and length. Specifically, we find by comparing different numerical models, that the way the model variables are ordered on a computational grid to solve the mathematical equations of sea ice motion has an effect of how many deformation features can be represented on a grid with a given spacing between grid points. A new discretization (ordering of model variables) turns out to resolve more details of the approximated field than traditional methods.
N2  - Key Points: The placement of the sea ice velocity has a mayor influence on the number of simulated linear kinematic features (LKFs). The CD‐grid resolves twice as many LKFs compared to A, B, C‐grids. A, B, C‐grids on quadrilateral meshes resolve a similar number of LKFs as A‐grids on triangular meshes (with the same total number of nodes).
UR  - http://resolver.sub.uni-goettingen.de/purl?gldocs-11858/9859
ER  -