[MITgcm-support] Time stepping for momentum and tracers

[Martin Losch]

Hi Uchida,

in the simplest case you just set deltaT to your timestep in seconds and this value is copied to all other deltaT’s

The other parameters (deltaTmom, deltaTtracer, deltaTclock, deltaTfreesurf) can be set separately if you want to use the tracer acceleration method of Bryan (1984 not sure about the year). In this method, which strictly should only be used for spinups, the tracer time step is made longer than the momentum time step (because stability allows a longer timestep for tracer than momentum), but this has the cos of distorted physics, slowed down waves, reduced seasonal cycle (refer to the literature, I found Danabasoglu 2004, Ocean Modelling interesting). The larger the ratio deltaTtracer/deltaTmom, the worse the situation. In the verification directory you’ll find a few examples how to do it most efficiently, e.g. global_ocean.cs32x15:

 deltaTMom   =1200.,
 deltaTtracer=86400.,
 deltaTFreeSurf=86400.,
 deltaTClock =86400.,

This means that only the momentum equations are stepped with 20min, the rest with one day, and also the model time (deltaTclock) follows the “fast” tracers. The problem is easly visible, when you consider that in the equations the momentum tendency is Gu*deltaTmom, which is 72 time smaller than it should be.

Martin

> On 19. Dec 2018, at 23:54, Uchida Takaya <tu2140 at columbia.edu> wrote:
> 
> Dear MITgcm support,
> 
> In the data namelist file, there is an option to set the time stepping for momentum (deltaTmom) and tracers (deltaTtracer) separately. 
> Ideally, they would have the same values but due to computational constraints, I would like to loosen the tracer time stepping if possible.
> Is there a best practice for this, e.g. ratio between deltaTmom and deltaTtracer etc.,  or should they not have different values at all?
> 
> Thank you,
> Takaya
> ———————
> PhD Candidate
> Physical Oceanography
> Columbia University in the City of New York
> https://roxyboy.github.io/
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support