Speaker
Description
As part of a new climate model development effort (CliMa), an efficient LES model
has been developed at MIT in order to simulate turbulent flow in simple geometry
configurations.
One of the target application is to explicitly resolve turbulent boundary layers
in a wide range of oceanic conditions in order to help to improve mixing
parameterizations used in GCMs.
The model, Oceananigans, is based on similar numerics to the MIT General
Circulation Model (MITgcm), but relies on a more efficient modal pressure solver
applicable to simple domain geometries.
Written in Julia, Oceananigans runs efficiently on both CPUs and GPUs
and includes several recent LES sub-grid scale closures.
A brief overview of the model is presented and results from few standard LES
test-cases show good agreement with either analytic solution or well established
model results.
In the broader context of model test-cases, this illustrates that for idealized
configurations with simple geometry, reliable tests are available to validate
and evaluate individual model component. The need for robust tests addressing
more complex, ocean science motivated, modeling problems is briefly discussed
from the experience gained with MITgcm.
| Do you need an official invitation letter? | No |
|---|
