[MITgcm-devel] Re: Hi-res cube

[Dimitris Menemenlis]

> Are we using partial steps?

Do you mean StaggerTimeStep ?
If so, yes: "StaggerTimeStep=.true.".

> What happens when you change visc4Leith? How small can visc4Leith get 
> before model is unstable?

With ERA40 forcing, the model went unstable in 1997 with viscC4Leith=0.1
somewhere in ACC region.  To continue integration time step was halved to 
DeltaT=600 for a year and a half, before resuming with DeltaT=1200.

viscC4Leith=0.2 also cleared the problem and ran past the unstable period.

> In mom_vecinv.F there's two commented lines of the form:
>   c      CALL MOM_VI_U_CORIOLIS_C4(bi,bj,K,vFld,vort3,r_hFacZ,uCf,myThid)
> which follows the second order calculation of the Coriolis terms. It 
> would be nice to see this fourth order vorticity advection but only 
> after we've established how little viscosity we can get away with. I 
> just looked at the latest version of mom_vecinv.F and notice that 
> someone has added some run-time logic ("useAbsVorticity") which you'll 
> need to check is consistent when you uncomment the above lines.

What do you expect to happen if 4-th order vorticity advection is turned on?

> PS> It would be useful to see the time-mean position of the 15oC contour.

OK, will do.

> One other thing: do you have GM/Redi turned on?

No, it's turned off.  I tried turning it on some time ago and model became 
very unstable.  It's probably worthwhile to have another go at some point
in time, after JM and Chris finish tidying up horizontal viscosity.

D.

-- 
Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
Jet Propulsion Lab, California Institute of Technology
MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
tel: 818-354-1656;  fax: 818-393-6720