[MITgcm-devel] obcs/exf and yearlyfields

[Martin Losch]

Hi Dimitris,

I can use useObcsYearlyFields, but where does this go? Into data.obcs/ 
data.exf? Default to false?

I was not planning to fix this bug (which I always circumvene by  
copying the first year, say 1992 to 1991 to have a fake record at the  
beginning of the intergration), looks to me like one would have to  
look for the file and if it is not available use genfile 
(1:il),'_',year1 instead (in exf_set_gen)?

Martin

On 24 Jan 2008, at 16:43, Dimitris Menemenlis wrote:

> Martin, if you do decide to add this capability, I have two comments:
>
> 1. Could you use a different flag, for example,  
> useObcsYearlyFields.  This is because there are already several  
> folks (Alan, An, Michael) who are using a mixed yearly and non- 
> yearly input files.
>
> 2. Keep your eyes open for a bug that fails to properly wrap around  
> between two years during initialization.  For example, if you  
> attempt to start from hour zero 1992, while the yearly files start  
> from hour three, the code will give an error message and exit  
> rather than use the last record of 1991 and wrap around.  This  
> problem only occurs during initialization.  It works OK if you are  
> integrating from say, December 31, 1991.
>
> D.
>
>> I noticed that exf_set_obcs_* do not use the convenient  
>> useExfYearlyFields
>> flag. Is there a principle problem with this flag and obcs, or is  
>> it just a
>> historical problem that is on someone list to be fixed? (I may  
>> want to fix it
>> for me now, can I try?)
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel