NearGlobalSimulations removed from ClimaOcean v0.2.0 and later

[vtamsitt]

This PR means GlobalOceanBioME no longer works. Is there more up-to-date ClimaOcean code we could update GlobalOceanBioME to run with instead? Or does it make sense to wait and set up to use OceanSeaIceModel?

[navidcy]

Let me force GlobalOeanBioME to use a tag before this PR until we update it to use the latest ocean-sea ice model

[jagoosw]

Yeah I think it makes sense to keep using an old tag of ClimaOcean for experimenting until the new setup is ready, and hopefully it'll be ready by the time anyones ready to do a full run.

[vtamsitt]

Makes sense, thanks

[navidcy]

There is already a compat entry for ClimaOcean

GlobalOceanBioME/Project.toml

Line 15 in e05d6e4

ClimaOcean = "0.1.2"

So until we change this line, it enforces using ClimaOcean versions [0.1.2, 0.2.0) which will have the NearGlobalSimulations functionality. Right?

[navidcy]

@vtamsitt just to confirm, did the code actually break or was your initial post here just anticipating the code breaking?

[vtamsitt]

Yes it works when starting the GlobalOceanBioME environment with ClimaOcean 0.1.2.

[vtamsitt]

Actually, there's an issue that I can't figure out what's going on with downloading the initial condition/bathymetry files from OceananigansArtifacts when calling one_degree_near_global_simulation.

I ran it on my laptop and it worked fine because the files were already saved in my ~/.julia/datadeps/ directory from previously running this. But when I tested on a remote machine without the data files already downloaded it didn't work, the url for the data files does not exist, and I can't seem to find any record of changes to the directory or these datafiles anywhere? I must be missing something...

[navidcy]

Damn. I know what’s the issue… I’ll try to fix tomorrow.

[navidcy]

cc @glwagner

[navidcy]

@glwagner, this package was depending on the branch we deleted from OceananigansArtifacts.jl
Could we push it back? Do you have that branch locally?

[glwagner]

Oh, we messed up. This is a good lesson.

To explain: we received a notice that we were going to be charged $$ by github for hosting too many large files on OceananigansArtifacts.jl. To address that we deleted branches, but I didn't realize the branches were being used...

I don't have it locally. @simone-silvestri might.

In case restoring the files does not work, what else might we try? What exactly does GlobalOceanBioME need? We are not actually so far from having an alternative way of running global models with bulk formula (and we already have a nice way to initialize a global model from ECCO data). So it might not be much work to fix this all and move into the future by using the latest tools, rather than trying to rewind to the old global setup.

[vtamsitt]

Ah ok, it did occur to me it might be related to too many large files. I have a local copy of these files, so could restore those?

I think what GlobalOceanBioME needs right now is some relatively well spun-up physics to use to test and optimise the BGC model(s), I think @jagoosw was planning to reach out to @simone-silvestri to see if we could get some other physics field outputs to use? @jagoosw did these old runs a while ago as a demo so it hasn't been actively developed for a while, and I've just started using it just to get up and running with the code and so we can do some tests and start looking at how the BGC models are looking. We definitely want to move this into the future with the latest tools, so working toward running this with bulk formula/initialised with ECCO as soon as that becomes available makes sense to me.

@jagoosw do you have thoughts?

[glwagner]

I have a local copy of these files, so could restore those?

This might work!

There's a couple options:

• Upload the files back to OceananigansArtifacts.jl
• Backport (ie develop in a new branch in ClimaOcean from the latest working version) a way for users to add the locally-available files so that the stuff in NearGlobalSimulations will succeed
• Upload the files to a new repo and make a new branch of ClimaOcean that points to it (so we don't make OceananigansArtifacts any bigger, but still allow setups to be easily ported from machine to machine, like we previously supported)

[navidcy]

@vtamsitt if you can upload the files somewhere and share with me (a repo or Dropbox or whatever) I can deal with it!