Open davidhassell opened 5 years ago
use the seeding technique as for CMIP5
Make test case to try out process ?
There is an issue with getting updates from the "parent" documentation after the seeding has occured.
The solution could simply be to advise that this is an issue, and suggest delaying seeding until the parent model is sufficiently described. This will only affect a few groups, so this is probably OK?
Seeding needs to be possible from MOHC documents, too.
I suggest a new file initialization_from_CMIP6.json
, to separate concerns. This file will have the the same format as the CMIP5 version, but will only be created by ES-DOC on a bespoke basis - i.e. when we know which model they want to be seeded.
This file will allow insititute:source_id
. If institute:
is missing, then it is assumed to be the same as the requesting institute.
A few institutes are using another institute's model: e.g. MPI-ESM1-2-HR is used by three institutes: MPI-M, DWD, DKRZ. The CMIP6 CV (https://github.com/PCMDI/cmip6-cmor-tables/blob/master/Tables/CMIP6_CV.json) describes this as follows:
This should model should only be documented once, and the other institutes should use (in some way) the existing documentation.
Given that MPI-M have documented the model, there are are two possible approaches: 1) DWD and DKRZ reference the MPI-M documents 2) DWD and DKRZ copy the MPI-M documents (i.e. the 9 realm and toplevel spreadsheets)
Whilst approach 1) is the right (i.e. CIM) way to do it, I think that there are practical considerations for CMIP6 that favour 2):