[MITgcm-support] Re: High-resolution CS grids

chris hill cnh at mit.edu
Sun Feb 6 13:51:24 EST 2005


Hi Yuan,

  The scirpts under high_res_cube do generate high-resolution grids.
  However, it helps to have a 64-512 processor system to run these 
calculations on and the runs generate very large output datasets.

  What sort of resolutions and grid point counts do you think you need 
for your problem?

Chris
Ed Hill wrote:
> On Sun, 2005-02-06 at 01:33 -0700, lian at email.arizona.edu wrote:
> 
>>Hi, Ed,
>>
>>I am Yuan Lian, a graduate student from Lunar and Planetary Lab, the University
>>of Arizona. I heard from Dr. Campin that you are working on High resolution
>>cube-sphere grids. I am working on this as well since I prefer to use
>>cube-sphere to some simulations on atmospheric circulation on gas giant
>>planets.
>>
>>I have posted my question on MITgcm-support mailing list. My question is if I
>>can generate high resolution grids by using the matlab codes under folder
>>"/MITgcm_contrib/high_res_cube" (I tried but the generated data files for
>>cube-sphere grids wouldn't work), or is there an alternative way to generate
>>cube-sphere grids? Thanks a lot for your help!
> 
> 
> 
> Hi Yuan,
> 
> Yes, I'm working on grid generation but only started a few weeks ago.  I
> do not (yet) have programs that will create a complete set of input
> files as needed by MITgcm.  I have only a few unfinished scripts that I
> hope to have in usable shape in the coming weeks.  So if you can wait
> (perhaps a month?) and would like to be a "beta" tester, then I'd be
> willing to work with you to produce some high-res grids appropriate for
> your domain.
> 
> But if you need something immediately then I suggest that you contact
> Chris Hill and Dimitris Menemenlis.  Chris and Dimitris have generated
> and used high-res MITgcm cubes (some with 510x510 cells per face) for
> the ECCO project:
> 
>   http://ecco.jpl.nasa.gov/cube_sphere/
> 
> and you can probably adapt one of their existing MITgcm setups with a
> minimum of effort.
> 
> Ed
> 




More information about the MITgcm-support mailing list