[MITgcm-support] High diffusion offline runs blow up

[Martin Losch]

Hi Emma,

your values violate the stability criteria for diffusion, see here:
http://mitgcm.org/public/r2_manual/latest/online_documents/node131.html
in 3.12.2.1, formula (3.44) gives S = 4*4e4*360/(1/20*pi/180*6371000)^2 = 1.86 > 1
with Kh=1e4 the simulation is marginally stable (0.47 < 1), which is enough for tracers (but probably not for momentum the recommenation is to have S<0.3)

In order to use Kh=4e4 you'll have reduce the time step, I suggest by a factor of 4.

Martin

On Jun 6, 2013, at 4:03 PM, Emma Boland <emmomp at bas.ac.uk> wrote:

> Hi all,
> 
> I'm running 2D offline runs of MITgcm 62m at 1/20 degree, using advection scheme 2 (2nd order central) to advect a domain filling tracer field (360 by 40 degrees domain). Running for on the order of a year, with a time step of 360 secs.
> 
> I need to do some very high diffusion runs, and Kh = 2000m2/s works fine, but Kh=4000m2/s and anything higher blows up - after one week there is a large patch of NaN/inf values and the whole tracer field becomes NaN by week 4.
> 
> Colleagues have carried out similar calculations at Kh = 1e4m2/s on previous MITgcm versions - any hints as to what might be going wrong? The runs that don't blow up behave as expected.
> 
> Thanks,
> Emma
> -- 
> Emma Boland
> Physical Oceanographer
> 
> The British Antarctic Survey
> High Cross, Madingley Road
> Cambridge
> CB3 0ET
> 01223221276
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support