[MITgcm-devel] [MITgcm-cvs] MITgcm/model/src CVS Commit

[Martin Losch]

Hi guys,

not sure why I am thought to be qualified to confirm this, but I think you are right and by comparison with the matlab sw-library, this is a bug. I am wondering if the “compatibility” with the matlab package could be taken further to actually call sw_ptmp from sw_temp with swapped arguments p,pr. That would save some code, reduce error sources and a decent compiler can optimise this with proper inlining anyway. What do you think?

BTW: the adiabatic temperature gradient (sw_adtg) is computed by the 1983 unesco formula; maybe it is inconsistent to have it like this, because we have at least two EOS that are not based on this formula (MDJWF and TEOS10). I am not so familiar with this EOS stuff: Would it be sufficient to add code only to sw_adtg that is used for MDJWF and TEOS10?

Martin


> On 05 Feb 2015, at 05:50, Matthew Mazloff <mmazloff at ucsd.edu> wrote:
> 
> Hi Jean-Michel
> 
> Yes -- please revert as you like. 
> 
> Thanks for dealing with this. Much appreciated!
> Matt
> 
> 
> On Feb 4, 2015, at 6:00 PM, Jean-Michel Campin <jmc at ocean.mit.edu> wrote:
> 
>> Hi Matt and Martin,
>> 
>> Sorry, I went out and only came back just now.
>> 
>> I can confirm that it does change isomip (fwd) results, but
>> also global_with_exf.yearly which uses SW_TEMP in pkg/frazil.
>> 
>> I don't know much about this seawater.F thing,
>> but it seems to me that you are right and you fixed a bug.
>> Martin, could you confirm ?
>> 
>> I would be OK for keeping the fix and update few reference output*.txt files, 
>> but it's not very early here and to run TAF & OpenAD experiments take some time.
>> 
>> Matt: Is it OK if I revert to the previous (buggy) version until 
>> tomorrow (to get a chance for Martin to confirm or not) ?
>> Then I can put back the fixed version and update the output*.txt files
>> that need to change.
>> 
>> Cheers,
>> Jean-Michel
>> 
>> On Wed, Feb 04, 2015 at 03:32:33PM -0800, Matthew Mazloff wrote:
>>> Hi Jean-Michel
>>> 
>>> I am sorry, I did not run a test report. I didn't realize shelfice package used SW_TEMP. Sorry for my ignorance -- if I had known I would have contacted the devel list first. 
>>> 
>>> We are assimilating tomography and found this bug. It doesn't change things much but I felt it important to fix it. You can revert it for now, but at some point we should fix this and update the test reports
>>> 
>>> Matt
>>> 
>>> 
>>> 
>>> 
>>> On Feb 4, 2015, at 3:15 PM, Jean-Michel Campin <jmc at ocean.mit.edu> wrote:
>>> 
>>>> Hi Matt,
>>>> 
>>>> I am currently re-running testreport (forward), because in an earlier
>>>> attempt, I was getting the 3 isomip test experiments that fail with
>>>> only 2,2 & 4 digits of agreement.
>>>> 
>>>> Since I know that pkg/shelfice is using some seawater conversion,
>>>> before checking-in the modifications in seawater.F
>>>> did you check by running a forward testreport ?
>>>> 
>>>> Cheers,
>>>> Jean-Michel
>>>> 
>>>> On Wed, Feb 04, 2015 at 01:54:35PM -0500, Matt Mazloff wrote:
>>>>> Update of /u/gcmpack/MITgcm/model/src
>>>>> In directory forge:/tmp/cvs-serv32054
>>>>> 
>>>>> Modified Files:
>>>>> 	seawater.F 
>>>>> Log Message:
>>>>> Fix bug to make sw_temp compatible with matlab library 
>>>>> 
>>>>> 
>>>>> _______________________________________________
>>>>> MITgcm-cvs mailing list
>>>>> MITgcm-cvs at mitgcm.org
>>>>> http://mitgcm.org/mailman/listinfo/mitgcm-cvs
>>>> 
>>>> _______________________________________________
>>>> 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
>> 
>> _______________________________________________
>> 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