Requested data vs. recommended diagnostics

The CMIP6 data request specifies the data which the endorsed MIPs have requested in order to complete their scientific objectives. Responding to pressure from modelling centres, the CMIP panel has specified that data should only be requested when there are specific plan to analyse that data. For any experiment, the data request allows modelling centres to determine not only the data requested but also, for each variable, to see who has requested and what scientific objective it is intended to support.

An email discussion between myself and Gokhan Danabasoglu at the start of February appeared to confirm that OMIP is not expecting to conduct and data analysis, and hence is not requesting data in the above sense. For this reason, the data request does not show OMIP requested data. The paper from Stephen Griffies at https://dl.dropboxusercontent.com/u/38722087/cmip6_ocean/WIP/OMIP_concerns_with_WIP_diagnostics.pdf shows that their is some misunderstanding here.

OMIP has defined two experiments (named OmipA and OmipB in the current version of the data request). Currently, C4MIP is the only endorsed MIP requesting data from these experiments.

Comments

  • @martinjuckes @Karl we might be getting to the bottom of this problem - the file that @stevegriffies is referring to is a "dataRequest" file named "160203_durack1_OMIPCMIP6DataRequestCompilationTemplate.xlsx" (which indicates we want data for Omon, Oyr, Oday and Ofx tables from the DECK and OMIP, in addition to other MIPs) and not the "160216_durack1_CMIP6_OMIP_physics_standard_output.xls" file that contains the diagnostic information (list of variables and their description), but importantly not the data request.

    I acknowledge that in the "160203" version of the data request file, we have requested data for all tables Omon, Oyr, Oday and Ofx and for many of the MIPs - these will have to be revisited to ensure that we are only requesting a reasonable amount of data that will in fact be analysed.
  • edited April 2016
    @PaulJDurack ; @stevegriffies: sorry for the confusion, I misread your Feb 16th email as a superseding the Feb 9th email with the above file. There are some questions about that file:
    (1) The experiments sheet lists 3 OCMIP6 experiments, while the previous version I have from you lists 2 OMIP experiments (OmipA and OmipB) ... which of these is valid?
    (2) In the covering email you refer to it as advocacy: let's try to see what this means in the case of PMIP. The complete Omon table contains, I believe 226 variables. PMIP have elected to request 54. If you wish to advocate that PMIP should use more OMIP diagnostics, this should be done by communicating with PMIP. The entry you have made in the "Request Scoping" sheet implies that OMIP is expecting to analyse the 226 Omon variables from all the PMIP experiments. If this is not the case, the entry should be removed and you should either leave the choice of ocean diagnostics to PMIP or persuade them to commit to analysing more data. This element of commitment is a critical part of the implied contract between the modelling groups and the MIPs: the modelling groups commit to providing the requesting data if the MIPs commit to analysing the data which they have requested. If you believe PMIP and other MIPs are wrong to have specified a smaller set of variables, you need to convince them.
    (2) The same question applies to DECK, CMIP6 historical and OMIP experiments: how much of this is advocacy and how much is describing data that OMIP can commit to analysing?
    (4) You have added a "CMIP6 RCP85" column: I've not seen any other reference to "RCP" runs in CMIP6 (though the experiment names are under review), instead we have a set of SSP runs defined by ScenarioMIP.
This discussion has been closed.