[Go-essp-tech] GO-ESSP-TECH Digest, Vol 29, Issue 56

[Karl Taylor]

Dear Mark,

I'm not sure my last email was very clear:

1.  Your model name, as it appears in filenames and directory names, is 
now and should in the future be:  CSIRO-Mk3-6-0  .

2.  Currently you pass CMOR model_id="CSIRO-Mk3-6-0", and this gets 
stored as a global attribute and is used to construct the filenames and 
directory names.

3.  If you pass CMOR model_id="CSIRO-Mk3.6.0", this will get stored as a 
global attribute and CMOR will modify it to confirm with the CMIP5 
specs. (converting hyphens to periods) so that your filenames and 
directories will continue to include the string "CSIRO-Mk3-6-0".

4.  The unadulterated model_id is what we're telling folks to use in 
publications, so you would prefer to pass CMOR "CSIRO-Mk3.6.0", but what 
I am asking the list is how would this affect ESG .

If this will cause problems, we'll stick with model_id as it currently 
is, and we'll have to advertise your officially correct model name in 
some other way.

Best regards,
Karl


On 7/26/11 3:25 PM, mark collier wrote:
> Hi Karl,
> On 26/07/2011, at 10:52 AM, go-essp-tech-request at ucar.edu wrote:
>
>> Message: 5
>> Date: Mon, 25 Jul 2011 17:03:23 -0700
>> From: Karl Taylor<taylor13 at llnl.gov>
>> Subject: [Go-essp-tech] model names
>> To: "go-essp-tech at ucar.edu"<go-essp-tech at ucar.edu>,    Bryan Lawrence
>>         <bryan.lawrence at stfc.ac.uk>,    Eric Guilyardi
>>         <Eric.Guilyardi at locean-ipsl.upmc.fr>,   "charlotte.pascoe at stfc.ac.uk
>> "
>>         <charlotte.pascoe at stfc.ac.uk>
>> Message-ID:<4E2E044B.4010800 at llnl.gov>
>> Content-Type: text/plain; charset="iso-8859-1"
>>
>> Hi all,
>>
>> Bob Drach is on vacation, so I'll ask my question to this list.
>>
>> There is a model_id stored as a global attribute in the CMIP5 netCDF
>> files.  This name is (typically) passed to CMOR which simply writes it
>> to the files.  The name is also used by CMOR to construct filenames
>> and
>> directory names, but it first removes characters forbidden by the
>> CMIP5
>> specs.  For example, the model_id=BCC-CSM1.1 is changed to BCC-
>> CSM1-1 in
>> file and directory names.
>>
>> My question is: If a group published a bunch of data with
>> model_id=CSIRO-Mk3-6-0 and then published additional data with
>> CSIRO-Mk3.6.0, would the ESG publisher or related software get
>> confused?  Note that the model name appearing in filenames and
>> directory
>> structures is not changed in going from 3-6-0 to 3.6.0.
>>
>> Best regards,
>> Karl
> Hi Karl,
> yes we have consistently used CSIRO-Mk3-6-0 for the filenames, but our
> group would prefer to use CSIRO-Mk3.6.0 as the reason for the numbers
> is to provide versioning information. We are happy to change the
> filenames on our disk area where the NCI ESG picks up the data from,
> but of course don't want to create any confusion.
>
> As you say, we would be interested in the approach of using CSIRO-
> Mk3.6.0 for any new files that we add to the repository. The only
> downside to this is that we are trying to serve the same data up
> locally on a file-system and the 2 different filenames would create
> some confusion and make commands like unix  slightly more tricky to
> operate on them or even when developing tools to process them.
>
> We would prefer to keep using the convention CSIRO-Mk3-6-0 with the
> filenames but for the label of the model (ie. the text that would
> appear in publications or be seen on the ESG gateways) to be CSIRO-
> Mk3.6.0
>
> Regards,
> Mark.
>