[MITgcm-devel] pickups: mdsio and mnc with unorthodox fields

[Martin Losch]

Hi Dimitris, et al.
I think I now understand how this works in seaice_pickup. A related  
question: is it really necessary to store all 3 time levels of UICE/ 
VICE/HEFF/AREA? Wouldn't the first (the current or last) timelevel be  
sufficient, because the seaice model uses some "modified" Eulerian  
time stepping, which only requires the previous time step. That would  
make the pickup smaller and easier to understand. In general, it is  
my opinion that this 3-timelevel business in the prognostic variables  
is a drag, difficult to read and not really always necessary.

Martin

On Mar 20, 2006, at 3:17 PM, Dimitris Menemenlis wrote:

> Hi Martin, I have never used mnc pickups (see Jean Michel's e-mail  
> for that) but
> I'm pretty sure that current seaice_pickup.F is OK.  In terms of 2- 
> D fields,
> this is the way it goes:
>
> #ifdef SEAICE_MULTILEVEL
> TICES    field 1-7
> YNEG     field 8
> HSNOW    field 9
> UICE     field 10-12
> ...
> #else /* SEAICE_MULTILEVEL */
> TICE     field 1
> YNEG     field 2
> HSNOW    field 3
> UICE     field 4-6
> ...
> #endif /* SEAICE_MULTILEVEL */
>
>> My understanding is, that the 7th argument of MDSWRITEFIELD give  
>> the position
>> the file, so 4 (as for UICE) means the 4th record, but where are  
>> record 2
>> and 3?
>
> Records 1-3 (in terms of nNz=3 records) for SEAICE_MULTILEVEL are  
> filled in by
> TICES, YNEG, and HSNOW.
>
>> Doesn't that mean, that UICE (second record) overwrites YNEG  
>> (second record
>> also), equivalently VICE and HSNOW?
>
> TICE, YNEG, and HSNOW have nNz=1.  So first three records fill the  
> first three 2-D levels.  UICE has nNz=3, so second record fills in  
> 2-D levels 4-6.  So nothing is overwritten.
>
>>  Is this correct? I want to add more fields, so I need to know.
>
> Yes.  I have tested that it is OK using a
> 2timesteps = 1timestep + 1timestep.
>
> Dimitris
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel