Ability to pull met data from an existing run

[jsta]

For example, can we get met data from a fully coupled E3SM run and use it an offline context. A challenge here is that the fully coupled output will be too large to upload to GEE so tooling would need to be pointed at local files (e.g., on Perlmutter).
-- paraphrased from @matthewhoffman

[jonschwenk]

I think the first step here is to complete the CMIP met exporter. Presumably, E3SM variable availability will conform to CMIP (maybe this is wrong)--at least as far as ELM needs are concerned. If that's the case, the CMIP adapter (which maybe should be called Climate) would work for this, and the CMIP sampling tools should also be re-usable here.

@matthewhoffman If this is something that InteRFACE could use, we could build it. Would need some sample E3SM outputs to develop on.

[matthewhoffman]

There actually is a pretty long and involved process for converting raw E3SM output to the format required by CMIP (see https://e3sm.atlassian.net/wiki/spaces/DOC/pages/818086012/Data+Publication+Process and the many child pages if you want to see the gory details), and there are many E3SM simulations that will never be published to ESGF or there will be prolonged delay before they are. There are also could be some degradation of information if any regrdding has been performed (that's not always the case but it may be). So I was thinking more how to use 'raw' E3SM ATM output. It could be that this does not really make sense to do when running ELM via OLMT which makes use of the "coupler bypass", and that what I'm suggesting would make more sense to do through a data atmosphere component without OLMT. I discussed briefly with @rfiorella today and he may have a better idea what the best workflow would be. There are some performance and workflow limitations to running with DATM rather than coupler bypass.

[jonschwenk]

dapper has the scaffolding to implement DATM exports although I don't know how involved it would be. Funny that it's complicated to pass raw E3SM output to an E3SM submodel, but I guess that's why the coupler is magic. I suspect that this is possible to do, but unless a project outside NGEE wants to fund it, I doubt it'll get prioritized as I haven't seen an NGEE need so far.