Martin.Losch at awi.de
Tue Mar 15 05:59:00 EDT 2016
your viscosity needs to depend on the time step to ensure stability. Even if you set it manually (via viscAh), you’ll need to take stability criteria (and deltaT) into account. Having said that, at that resolution you can probably get away with constant viscosity, i.e. viscAh = some value that you can estimate based on cfl and boundary current resolution requirements, see e.g., <http://mitgcm.org/public/r2_manual/latest/online_documents/node115.html>, section 126.96.36.199 Numerical Stability Criteria for details. Alternatively you can compute what vischAhGrid=0.1 gives you and use the average for your grid. There will always be some tuning required.
> On 15 Mar 2016, at 09:47, Ada Gjermundsen <ada.gjermundsen at geo.uio.no> wrote:
> I'm running MITgcm in an ocean only configuration using only the GM package. The horizontal grid resolution is 1 x 1cos(lat). For the viscosity parameters I have used viscAr=1.E-4, viscAhGrid=0.1,. This gives a horizontal viscosity which depends on the time step. I would prefer to have a viscosity which does not depend on the time step. Do some of you have some recommendations for what kind of viscosity scheme preferable for a 1 deg resolution on a Lon-Lat grid?
> Ada Gjermundsen
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
More information about the MITgcm-support