IMIS | Flanders Marine Institute

Flanders Marine Institute

Platform for marine research

In:

IMIS

Publications | Institutes | Persons | Datasets | Projects | Maps
[ report an error in this record ]basket (0): add | show Printer-friendly version

Modeling ocean circulation on unstructured meshes: comparison of two horizontal discretizations
Danilov, S.; Wang, Q.; Losch, M.; Sidorenko, D.; Schröter, J. (2008). Modeling ocean circulation on unstructured meshes: comparison of two horizontal discretizations. Ocean Dynamics 58(5-6): 365-374. dx.doi.org/10.1007/s10236-008-0138-5
In: Ocean Dynamics. Springer-Verlag: Berlin. ISSN 1616-7341, more
Peer reviewed article

Available in Authors 

Keywords
    Baroclinic motion; Finite element method; Ocean models; Oceanic circulation; Marine

Authors  Top 
  • Danilov, S.
  • Wang, Q.
  • Losch, M.
  • Sidorenko, D.
  • Schröter, J.

Abstract
    Finite-element models on unstructured meshes are frequently formulated in terms of continuous linear elements, which suffer from pressure modes and require stabilization. Alternatively, horizontal velocities may be represented with linear nonconforming elements. While the latter formulation uses three times more degrees of freedom for the velocity, it does not support pressure modes. The effects of stabilization are estimated by comparing the performance of continuous linear and nonconforming versions of the finite-element ocean circulation model (FEOM) in two simple configurations: a Munk gyre and baroclinic turbulence in a zonally reentrant channel. It is shown that, outside the free slip boundary layers, the presence of stabilization does not lead to noticeable effects if its strength is kept within certain limits. In order to evaluate the performance of FEOM, the baroclinic turbulence test is repeated with the MIT general circulation model (MITgcm), which serves as a benchmark, and reasonable agreement between different model codes is found. The two versions of FEOM have a similar computational cost, but both are significantly slower (per node) than the regular-mesh MITgcm. The paper also provides a brief description of the implementation of the nonconforming version of FEOM.

 Top | Authors