activity_id PMIP experiment_id have no parent?

[chrisdane]

Hi As far as I understand Otto-Bliesner et al. 2017 (p. 3988):

2.6 Setup and documentation of simulations To provide initial conditions for the [midHolocene] simulations, it is recommended that a spin-up simulation is performed departing from the CMIP6 piControl experiment.

the PMIP experiments such as midHolocene should branch from the piControl run from the DECK simulations. Is that correct?

If yes, I do not understand, why e.g. the experiment_id midHolcene (line 5911 in CMIP6_experiment_id.json) reads

        "midHolocene":{
            "activity_id":[
                "PMIP"
            ],
            "parent_activity_id":[
                "no parent"
            ],
            "parent_experiment_id":[
                "no parent"
            ],

As far as I understand the paper above, midHolocenes parent activitiy and experiment should be CMIP and piControl?

Thanks a lot for help Chris

[charliepascoe]

Hi Chris, My understanding is that this experiment is listed as having no parent experiment because the period for which data is archived follows on from the spin-up. Charlotte

[chrisdane]

But am I right that midHolocene must branch from piControl? Or may I initialize midHolocene from any other (spinup) run?

? --> spinup --> midHolocene

[charliepascoe]

The spin-up for the midHolocene experiment should be forced with "mid-Holocene (6 kyr ago) boundary conditions for orbital parameters and the atmospheric concentration of well-mixed greenhouse gases." And that spin-up should branch from the piControl.

[chrisdane]

So

piControl (quasi-equilibrium) --> midHolocene (x years spinup) --> midHolocene (min 200 years for cmor)

? Ok I understand that, still, I think it is a little counter-intuitive to say that midHolocene has no parent. For me that implies that the piControl and midHolocene experiments of one model are totally independent of each other. But, in fact, they are not?

[charliepascoe]

Yes that is how I understand it, @jypeter can you confirm. And yes, I agree, the connection between experiments could be more intuitive but it's difficult to fit that subtlety into the data model.

[durack1]

Just to throw it out there, there is the piControl-spinup experiment defined, which following the thread above may be just the experiment you're looking for

[chrisdane]

Ok and what would you say if you have a midHolocene run that was never in touch with any piControl or piControl-spinup experiment? Would you regard this as a valid midHolocene experiment? Thanks a lot for all your help.

[taylor13]

If a model was not directly branched from a simulation listed in the list of CMIP6 experiment i.d.'s it is said to have "no parent", even though it might have been branched from some simulation external to the suite of CMIP experiments (and even if some of them had a CMIP6 parent).

As I understand it the recommended procedure is what you summarized in https://github.com/WCRP-CMIP/CMIP6_CVs/issues/939#issuecomment-631387635 above. It therefore has no parent because midHolocine-spinup is not an official CMIP6 experiment.

I do not know whether following a different spin-up procedure would disqualify the simulation.

[chrisdane]

Thank you all very much for your help. I close this for now.