TIPMIP: Experiments requested by TIPMIP

[matthew-mizielinski]

General Issue: New experiments

TIPMIP have asked to add experiments to CMIP6Plus and their experiment definition appears to include the following set of experiments;

See here for diagramatic description of the experiments:

experiment id	activity id	experiment	tier	sub experiment id	parent experiment id	required model components	additional allowed model components	start year	end year	min number of years	parent activity id	description
esm-up2p0	TIPMIP	ramp up at X Gt C/yr to Global Warming Level of 2K at 0.2 K/decade	1	none	esm-piControl	AOGCM BGC	AER CHEM			100	CMIP
esm-up2p0-swl2p0	TIPMIP	zero emissions at global warming level of 2K	1	none	esm-up2p0	AOGCM BGC	AER CHEM			50	TIPMIP
esm-up2p0-swl2p0-50y-dn2p0	TIPMIP	ramp down from global warming level of 2K at 0.2 K/decade	1	none	esm-up2p0-swl2p0	AOGCM BGC	AER CHEM			100	TIPMIP
esm-up2p0-swl4p0	TIPMIP	zero emissions at global warming level of 4K	1	none	esm-up2p0	AOGCM BGC	AER CHEM			50	TIPMIP
esm-up2p0-swl4p0-50y-dn2p0	TIPMIP	ramp down from global warming level of 2K at 0.2 K/decade	1	none	esm-up2p0-swl4p0	AOGCM BGC	AER CHEM			100	TIPMIP
esm-up2p0-swl4p0-50y-dn2p0-swl2p0	TIPMIP	zero emissions at global warming level of 2K after ramp down from global warming of 4K	1	none	esm-up2p0-swl4p0-50y-dn2p0	AOGCM BGC	AER CHEM			50	TIPMIP

The experiment names are systematic, but do get a little long (33 characters max). I don't think this will cause any problems in CMOR, but we might need to confirm that the ESGF interfaces can handle this.

As far as suitability for CMIP6Plus goes, I think the match up with CMIP7's goals is sufficient.

@durack1 could you review the information above?

[durack1]

the metagrid ESGF interface maxes out at ~12 chars before ellipses creep in, not an issue for uniquely identifying info, but does mean that it will be harder to identify the esm-up2p0-swl2p0-5.. or longer experiment_id's. Presumably the esm-piControl is the CMIP6 experiment_id?

Below is the listing of source_id's that published esm-piControl sims to CMIP6:

ACCESS-ESM1-5 (860)
AWI-ESM-1-REcoM (313)
BCC-CSM2-MR (508)
CESM2 (2226)
CNRM-ESM2-1 (1548)
CanESM5 (1010)
CanESM5-CanOE (660)
EC-Earth3-CC (705)
GFDL-ESM4 (966)
GISS-E2-1-G-CC (573)
MIROC-ES2L (904)
MPI-ESM1-2-LR (1281)
MRI-ESM2-0 (1090)
NorESM2-LM (933)
UKESM1-0-LL (1172)

[matthew-mizielinski]

Presumably the esm-piControl is the CMIP6 experiment_id?

Correct -- this is already in the CMIP6Plus CVs as part of the DECK. Groups may use that as the parent or submit a new one (expecting UKESM1-1-LL esm-piControl will need to be included).

Regarding the length of the experiment id in the search index. This is an extreme case of something we already have (some of the DCPP experiments have long names too). Personal opinion is that as long as the experiment proposers are aware of the issue then this isn't a problem. We could request that the metagrid devs make that side bar expandable and (ideally) allow the contents of the selection boxes to adapt.

[durack1]

Presumably the esm-piControl is the CMIP6 experiment_id?

Correct -- this is already in the CMIP6Plus CVs as part of the DECK.

All looks good to me, is there any point in adding parent_experiment_id CMIP6:esm-piControl (we can't validate this within the repo, as this is using a standing prior-phase initialization) or is this starting to get into a tangle that we don't want to complicate ourselves with?

[matthew-mizielinski]

Presumably the esm-piControl is the CMIP6 experiment_id?

Correct -- this is already in the CMIP6Plus CVs as part of the DECK.

All looks good to me, is there any point in adding parent_experiment_id CMIP6:esm-piControl (we can't validate this within the repo, as this is using a standing prior-phase initialization) or is this starting to get into a tangle that we don't want to complicate ourselves with?

esm-piControl could well be submitted to CMIP6Plus rather than CMIP6 depending on the model. I don't think we should worry too much about this and leave it to project documentation.

[matthew-mizielinski]

@sciconaut, does the experiment list above look correct

[klauswyser]

Thanks for preparing the experiment descriptions and sorry for the belated reply. I have a couple of comments to table entries in the description, how do I do this in the most efficient way? Copy the table to an issue and update? Or is there a file with the table where I could suggest changes and add comments?

[klauswyser]

And another question: this list of experiments isn't complete, we already plan for tier 2 simulations with swl1.5 or swl3.0, different lengths of overshoot, faster or slower ramp-downs, etc. I assume it won't be a problem to add more experiments later as long as we don't modify any existing experiment, or?

[taylor13]

Note that one constraint on experiment names (as they appear in the CMIP data base and in file names) is that a period (i.e., ".") is not permitted. This is because in file names the "." is reserved to separate the the file type suffix (e.g., ".nc") from the filename. So if you want a name like "swl1.5", it will have to be modified for use in the CMIP database. In CMIP6, for example, the identification of an experiment consistent with "SSP5-8.5" scenario were referenced as "ssp585" in CMIP. And an experiment with CO2 halved abruptly was referenced as "abrupt-0p5xCO2" (where "p" stands for "point").

Some options you might consider in place of "swl1.5" include: "swl15", "swl1-5" or "swl1p5".

Happy to iterate.

[klauswyser]

Thanks, I'm aware of this restriction and just sloppily wrote gwl1.5 in the text. Matt's table in the description already takes this into account and replaces "." by "p", we will continue with that.

[taylor13]

Excellent! I obviously haven't been keeping up.

[matthew-mizielinski]

Hi @klauswyser,

Feel free to copy the table and update it, but if you do please highlight the changes. When it comes to extending with new experiments later I don't see a problem as long as the naming is consistent and you don't conflict with other experiments.

[klauswyser]

Thanks a lot for the draft of the experiment metadata for the CMIP6plus tables. Here are some suggested modifications, based on the information in ESM experiment protocol for TIPMIP. This information will also be included in the TIPMIP experiment protocol paper that currently is in the writing.

experiment id	activity id	experiment	tier	sub experiment id	parent experiment id	required model components	additional allowed model components	start year	end year	min number of years	parent activity id	description
esm-up2p0	TIPMIP	ramp up with X Gt C/yr to Global Warming Level of 2K at 0.2 K/decade constant CO2 emissions that yield a warming of approximately 2 K/century (see TIPMIP experiment protocol for details)	1	none	esm-piControl	AOGCM BGC	AER CHEM			100 215	CMIP
esm-up2p0-swl2p0	TIPMIP	zero CO2 emissions at global warming level of 2K, branching off from esm-up2p0 the year after it passes the 2 K global warming level	1	none	esm-up2p0	AOGCM BGC	AER CHEM			50 300	TIPMIP
esm-up2p0-swl2p0-50y-dn2p0	TIPMIP	ramp down from global warming level of 2K at 0.2 K/decade after 50 years, with CO2 emissions that are the negative of those used in esm-up2p0	1	none	esm-up2p0-swl2p0	AOGCM BGC	AER CHEM			100	TIPMIP
esm-up2p0-swl4p0	TIPMIP	zero CO2 emissions at global warming level of 4K, branching off from esm-up2p0 the year after it passes the 4 K global warming level	1	none	esm-up2p0	AOGCM BGC	AER CHEM			50300	TIPMIP
esm-up2p0-swl4p0-50y-dn2p0	TIPMIP	ramp down from global warming level of 2 4 K at 0.2 K/decade after 50 years, with CO2 emissions that are the negative of those used in esm-up2p0	1	none	esm-up2p0-swl4p0	AOGCM BGC	AER CHEM			100200	TIPMIP
esm-up2p0-swl4p0-50y-dn2p0-swl2p0	TIPMIP	zero CO2 emissions at global warming level of 2K after, branching off from esm-up2p0-swl4p0-50y-dn2p0 after it passes the 2 K global warming level ramp down from global warming of 4K	1	none	esm-up2p0-swl4p0-50y-dn2p0	AOGCM BGC	AER CHEM			50 300	TIPMIP

Notes

Line 1: ramp-up goes to GWL4 which means at least 200 yrs. Add 15 yrs to compute the 31-yr running mean. Shorten experiment attribute. Add reference to TIPMIP experiment protocol (unfortunately not published yet)

Lines 2 & 4 & 6: SWL simulations should go for at least 300 yrs. Update experiment attribute to describe more precisely when to launch the GWL experiments

Lines 3 & 5: Update experiment attribute to describe more precisely the ramp-down experiments.

[klauswyser]

One open question is still if we should use "swl" or "gwl" in the experiment_id's of the different experiments. The ESM experiment protocol for TIPMIP uses "gwl", also in this overview "gwl" is used.

[matthew-mizielinski]

One open question is still if we should use "swl" or "gwl" in the experiment_id's of the different experiments. The ESM experiment protocol for TIPMIP uses "gwl", also in this overview "gwl" is used.

Looks like you are talking yourself into using "gwl", if it is used in papers and websites it would be good to be consistent, plus the phrase "global warming level" is used in the experiment descriptions. I'd recommend updating the table above with this change.

[jprb-walton]

One open question is still if we should use "swl" or "gwl" in the experiment_id's of the different experiments. The ESM experiment protocol for TIPMIP uses "gwl", also in this overview "gwl" is used.

Looks like you are talking yourself into using "gwl", if it is used in papers and websites it would be good to be consistent, plus the phrase "global warming level" is used in the experiment descriptions. I'd recommend updating the table above with this change.

I'd agree with using "gwl" instead of "swl" (partly because I've forgotten what swl originally stood for, and "everyone" knows about gwl).

[jprb-walton]

Thanks a lot for the draft of the experiment metadata for the CMIP6plus tables. Here are some suggested modifications, based on the information in ESM experiment protocol for TIPMIP. This information will also be included in the TIPMIP experiment protocol paper that currently is in the writing.

These mods look fine to me, Klaus: well done. Perhaps we could tweak the descriptions slightly to ensure they're consistent, and concise? The virtue of having a modular naming system (esm-up2p0, esm-up2p0-gwl2p0, esm-up2p0-gwl2p0-50y-dn2p0, etc) is that, for a given experiment, we can refer to preceding steps using their names without having to describe the steps - e.g. esm-up2p0-gwl4p0-50y-dn2p0-gwl2p0 = zero CO2 emissions at global warming level of 2K, branching off from esm-up2p0-gwl4p0-50y-dn2p0.

[klauswyser]

Thanks for the helpful comments, appreciate!

Here is a new attempt, "swl" replaced by "gwl" and experiment descriptions shortened as @jprb-walton suggested. What do you think?

experiment id	activity id	experiment	tier	sub experiment id	parent experiment id	required model components	additional allowed model components	start year	end year	min number of years	parent activity id	description
esm-up2p0	TIPMIP	ramp up simulation with constant CO2 emissions yielding a warming of approximately 2 K/century (see TIPMIP experiment protocol for details)	1	none	esm-piControl	AOGCM BGC	AER CHEM			215	CMIP	ramp up with constant CO2 emissions
esm-up2p0-gwl2p0	TIPMIP	zero CO2 emissions simulation starting at global warming level of 2K, branching off from esm-up2p0	1	none	esm-up2p0	AOGCM BGC	AER CHEM			300	TIPMIP	stabilisation at GWL2
esm-up2p0-gwl2p0-50y-dn2p0	TIPMIP	ramp down simulation with CO2 emissions that are the negative of those used in esm-up2p0, branching off from esm-up2p0-gwl2p0 after 50 years	1	none	esm-up2p0-gwl2p0	AOGCM BGC	AER CHEM			100	TIPMIP	ramp down after temporary overshoot to GWL2
esm-up2p0-gwl4p0	TIPMIP	zero CO2 emissions simulation starting at global warming level of 4K, branching off from esm-up2p0	1	none	esm-up2p0	AOGCM BGC	AER CHEM			300	TIPMIP	stabilisation at GWL4
esm-up2p0-gwl4p0-50y-dn2p0	TIPMIP	ramp down simulation with CO2 emissions that are the negative of those used in esm-up2p0, branching off from esm-up2p0-gwl4p0 after 50 years	1	none	esm-up2p0-gwl4p0	AOGCM BGC	AER CHEM			200	TIPMIP	ramp down after temporary overshoot to GWL4
esm-up2p0-gwl4p0-50y-dn2p0-gwl2p0	TIPMIP	zero CO2 emissions simulation starting at global warming level of 2K, branching off from esm-up2p0-swl4p0-50y-dn2p0	1	none	esm-up2p0-gwl4p0-50y-dn2p0	AOGCM BGC	AER CHEM			300	TIPMIP	stabilisation at GWL2 after temporary overshoot to GWL4

[klauswyser]

Just updated the last column "description" in the table above.

[durack1]

One open question is still if we should use "swl" or "gwl" in the experiment_id's of the different experiments. The ESM experiment protocol for TIPMIP uses "gwl", also in this overview "gwl" is used.

@klauswyser can you open this google doc up for world-viewable/commentable please? It's locked down at this moment

[klauswyser]

Hi @durack1 - the Google doc mentioned above is a milestone report from the TipESM project and as such not public, I'm afraid. Many of those invited to discuss the experiment names here are member of OptimESM, TipESM or TIPMIP and have access to the MS1 report which is why I did mention it here.

[matthew-mizielinski]

I've drafted the additional experiments in the experiment_id CV in #85, needs review and then some discussion with @wolfiex to confirm the process.

[klauswyser]

Thanks @matthew-mizielinski for the great work. I checked #85 and couldn't see anything deviating from the discussion here. But it would be good if @jprb-walton also had a look, 4 eyes see more than 2.

[antworteffekt]

Thanks @matthew-mizielinski ! I've checked #85 as well, the only thing I could find is in line 348: branching off from esm-up2p0-swl4p0-50y-dn2p0, but should be branching off from esm-up2p0-gwl4p0-50y-dn2p0 for consistency with the rest of experiment descriptions.

[matthew-mizielinski]

Thanks @matthew-mizielinski ! I've checked #85 as well, the only thing I could find is in line 348: branching off from esm-up2p0-swl4p0-50y-dn2p0, but should be branching off from esm-up2p0-gwl4p0-50y-dn2p0 for consistency with the rest of experiment descriptions.

good catch -- corrected in 9fc1048

[klauswyser]

How shall we do with tier 2 experiments? In the OptimESM project there already are other experiments than those listed in #85 (e.g. GWL3) that also wait for cmorisation. The TIPMIP description paper will list these experiments as tier 2.

Shall we add these experiments in #85, or would it be better to first execute !85 and then open a new pull request for tier 2 experiments?

[klauswyser]

Here are the Tier 2 experiments for TIPMIP, a couple of additional GWL simulations that we are running as part of OptimESM and/or TipESM. @jprb-walton, do you think these are all extra runs, or do you have any other experiments?

experiment id	activity id	experiment	tier	sub experiment id	parent experiment id	required model components	additional allowed model components	start year	end year	min number of years	parent activity id	description
esm-up2p0-gwl1p5	TIPMIP	zero CO2 emissions simulation starting at global warming level of 1.5K, branching off from esm-up2p0	2	none	esm-up2p0	AOGCM BGC	AER CHEM			300	TIPMIP	stabilisation at GWL1.5
esm-up2p0-gwl3p0	TIPMIP	zero CO2 emissions simulation starting at global warming level of 3K, branching off from esm-up2p0	2	none	esm-up2p0	AOGCM BGC	AER CHEM			300	TIPMIP	stabilisation at GWL3
esm-up2p0-gwl5p0	TIPMIP	zero CO2 emissions simulation starting at global warming level of 5K, branching off from esm-up2p0	2	none	esm-up2p0	AOGCM BGC	AER CHEM			300	TIPMIP	stabilisation at GWL5

Colin confirms that we can add these extra GWL exps in the protocol paper:

Yes, I think we are within our rights to class

1. Zero emission runs at GWL=0K (PI), 2K and 4K as TIPMIP Stream/Tier 1.
2. Zero emission runs at GWL = 1.5, 3 and 5K as TIPMIP Stream/Tier 2.

Tier 1 will be “required”, Tier 2 will be “encouraged”, but all can be official TIPMIP runs, so valid on the ESGF.
We can word it this way in the protocol paper

 
Repating my practical question: shall we add these Tier-2 experiments to #85, or open a separate PR later? @matthew-mizielinski , what do you think?  

[matthew-mizielinski]

Hi @klauswyser. @wolfiex has made some updates to the JSON-LD files that ultimately create the experiment_id.json file which account for the Tier 1 nodes. Could you open a separate issue for the Tier-2 and we'll pick them up independently.

[klauswyser]

Hi @klauswyser. @wolfiex has made some updates to the JSON-LD files that ultimately create the experiment_id.json file which account for the Tier 1 nodes. Could you open a separate issue for the Tier-2 and we'll pick them up independently.

Done, opened #87.

[matthew-mizielinski]

Updates in v6.5.2.0 release