ePIC

Amatos: Adaptive Mesh Generator for Atmospheric and Oceanic Simulation

Edit Item Edit Item

General Information:

Citation:
Rakowsky, N. , Behrens, J. and Hiller, W. (2003): Amatos: Adaptive Mesh Generator for Atmospheric and Oceanic Simulation , The 2nd International Workshop on Unstructured Grid Numerical Modelling of Coastal, Shelf and Ocean Flows, 23-25 Sept, Delft, Netherlands. .
Cite this page as:
Contact Email:
Download:

Supplementary Information:

Abstract:

The grid generator amatos(http://www-m3.mathematik.tu-muenchen.de/m3/software/amatos/index.html)was developped to make adaptive mesh generation available for dynamictime dependent flow problems as arising in climate simulations.The philosophy of amatos is to hide away all nontrivial tasksconcerning mesh generation and adaption from the applicationprogrammer. The complete mesh generation process can be controlled byapprox. 20 Fortran 90 subroutines of the application programminginterface (GRID API), which also provides additional variables, datastructures and constants.The generic field of application for amatos is atmosphere and oceancirculation modelling with semi-Lagrangian convectionschemes. The time stepping scheme, however, is not part of amatos. Thesemi-Lagrangian background is merely reflected by the choice ofservice routines (e.g., for interpolation) implemented in the GRID API,which may be extended for further needs.At the time being, amatos provides*) a F90-interface based on modular software techniques,*) mesh adaption (i.e., refinement and coarsening corresponding to a givenerror criterion) of planar, spherical or volume grids with triangular ortetrahedral elements,*) adaptivity by hierarchical data structures (refinement by bisectionmethod),*) support of arbitrary FEM basis functions (to be defined by the user,linearand quadratic functions are preinstalled),*) fast mesh partitioning by a space filling curve (SFC) approach (forparallelization by domain decomposition and as cache efficient numberingfor local calculations),*) service routines e.g., for interpolation including a mass conservingscheme,*) shared memory parallelisation with OpenMP.To demonstrate the application of amatos, we want to give a shortinsight into two projects.1. The simulation of tracer transport and mixing in the arcticstratosphere shows the power of adaptivity allowing a high localresolution of small-scale phenomena, without exhausting limitedcomputing resources. The grid is refined locally and dynamically atthose regions of the computational domain, where small scale phenomenaare active. (Collaboration with AWI Potsdam)2. The modelling of dynamics of large-scale atmospheric circulationstructures with a spherical shallow water model is a project inprogress. In this presentation, we want to focus on the technicalaspect of how to parallelize a model with an adaptive grid, whosecomputational domain has to be redistributed among the processorsafter each adaption. Here, the SFC shows to be a very elegant andefficient approach. (Collaboration with AWI Potsdam, funded in therange of the german DEKLIM research program).

Further Details:

Imprint
AWI
Policies:
read more
OAI 2.0:
http://epic.awi.de/cgi/oai2
ePIC is powered by:
EPrints 3