[MITgcm-support] viscAh & viscAz!

[Van Thinh Nguyen]

Hi Guttorm,

Sorry for this delay, because I was out of my work last week!

Following your advice, I set the parameter tempAdvScheme=33 in order to
use the 3rd order DST Flux limiter, then I got a better result, it reduced
significant noises. However, I still haven't got really good result as I
have been expecting. For the option with tempAdvScheme=77, the 2nd order
flux limiter, I got a worse result than the above 3rd. I suppose I have to
refine the grid. However, it took too long for a calculation, for the grid
resolution=1600*2*80 (domain:4000m*5m*40m) I used 16 cpus on irix64, the
calculation took about 90h for a simulation time 2 days due to the time
step is only 0.5s. If you would have any idea to improve the computational
time, please give me a help.

Thanks so much,

Van Thinh
--------------

 On Wed, 29 Jun 2005, Guttorm Alendal wrote:

> Van Thinh,
>
> Before adding any "spurios" viscosity. Try use a different advection
> scheme with flux-limiters.
>
> In 'data' set saltAdvScheme and/or tempAdvScheme (which ever you use to
> create the stratification) to 77 or 33 in namelist PARM01. I have done
> lock-release studies, on very high resolution, with viscosities down to
> 10^-8 with nice results.
>
> See http://mitgcm.org/pelican/online_documents/node76.html for
> documentation.
>
> Good luck
>
> guttorm
>
>
> --
> Senior scientist/Group leader
> Computational Mathematics Unit
> Bergen Center for Computational Science
> UNIFOB/University of Bergen
> Thormøhlensgt. 55
> N-5008 Bergen, Norway
> guttorm.alendal at bccs.uib.no
> Phone:+47 55 58 40 44
> Fax:  +47 55 58 42 95
>
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support
>