[MITgcm-devel] problem with exch1_rl and seaice_read_pickup

[Martin Losch]

Hi Jean-Michel, 

Your suggestions are good work arounds. (Local copies of SEAICE_SIZE.h/EXCH.h)

More generally, TICES is never exchanged except for here in seaice_read_pickup.F. I think one could
1. not exchange TICES at all (the overlap is never used)
2. exchange TICES "slabwise" as it has been done long ago (maybe only if Nr+1<MULTDIM)

Option (1) is not really safe, but should work for now as long the overlaps are not used, option (2) is ugly.

Martin

On Oct 11, 2012, at 2:28 PM, Jean-Michel Campin wrote:

> Hi Martin,
> 
> With ICE_multDim = 1, there is a quick way to go around this 
> by having a local copy of SEAICE_SIZE.h in our code dir
> (in the case Nr+1 >= ICE_multDim but Nr+1 < default MULTDIM = 7 )
> with MULTDIM=1
> Does this works ?
> 
> In the case ICE_multDim = MULTDIM = 7 > Nr+1, I think (like you) that 
> we cannot use Max in parameter statement.
> But my impression is that this problem was already there before 
> Gael made SEAICE_MULTICATEGORY changes.
> In this case, will have to make a local copy of EXCH.h with a larger
> value of MAX_NR_EXCH. An alternative would be to put MAX_NR_EXCH in 
> a (small) separated header file (EXCH1_SIZE.h) to minimize
> cutomized file.
> 
> Cheers,
> Jean-Michel
> 
> On Thu, Oct 11, 2012 at 09:36:27AM +0200, Martin Losch wrote:
>> Hi there,
>> 
>> I find that a restart with pkg/seaice, and SEAICE_MULTICATEGORY undefined (i.e. SEAICE_multDim = 1) does not work; here's the error message:
>> 
>> STOP  S/R EXCH1_RL: myNz  .GT. MAX_NR_EXCH 
>> 
>> When I check the values: MAX_NR_EXCH = 4 and myNz = 7 (because in seaice_read_pickup, there's a
>>        CALL EXCH_3D_RL   ( TICES, MULTDIM, myThid )
>> with PARAMETER ( MULTDIM = 7 ).
>> 
>> Obviously I have Nr = 3, so that MAX_NR_EXCH = 4 in conflict with MULTDIM  = 7.
>> Having an ice model with only a 3layer ocean (or a 1layer ocean) is not very realistic, but for tests (of the seaice model) very convenient.
>> 
>> What's a good solution to this? exchange TICES only with SEAICE_multDim>1 (will not work either) or change the definition of MAX_NR_EXCH to MIN(Nr+1, MULTDIM) (will not work in a parameter statement, will it?)
>> 
>> Martin
>> 
>> 
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel