CatchCN Runs at NAS: Desirable?

[mathomp4]

This is a question for @gmao-rreichle and @gmao-jkolassa: Do you want users to be able to run CatchCN at NAS?

At the moment this is impossible due to a hardcoded discover path:

GEOSgcm_App/gcm_run.j

Line 401 in fe30cc4

>>>GCMRUN_CATCHCN<<</bin/ln -s /discover/nobackup/projects/gmao/bcs_shared/make_bcs_inputs/land/CO2/v1/CO2_MonthlyMean_DiurnalCycle.nc4

To make this "general" this file would need to be put in a path that is agnostic to center.

Now, currently this file seems to be in the inputs for make_bcs (at least from the directory name) and make_bcs is decidedly not portable and there might never be a desire for it to be as it isn't a script that is routinely run.

I'm not sure of the solution. We could copy over the make_bcs_inputs but that is 1 TB of files when running CatchCN only needs 4 MB. That seems only prudent if we want to run make_bcs at NAS.

We could put the file in the same area as other non-chem component files: ExtData/g5gcm which has files for Moist, GWD, and Solar in it.

[gmao-jkolassa]

Hi @mathomp4 ,

Thanks for starting this discussion! I am not sure I have a straight answer, but rather a few follow-up questions. My understanding was that so far neither the forcing data needed for land-only runs nor the boundary condition data were available on NAS and that this was impeding running any version of the land model offline on NAS. Has that changed?

Regarding a need for running Catchment-CN on NAS, as far as I am aware none of the current Catchment-CN users have a NAS account so there is no immediate need. However, given the recent long wait times on Discover there may be an interest. Do you have any sense of how much of an effort it would be to port Catchment-CN (and everything that it needs) to NAS?

Thanks!

[mathomp4]

@gmao-jkolassa If you want to run CatchCN 4.0 and 4.5 (at least like how I know how to run it), you only need that file. I copied it over and fixed up gcm_run.j to point to the one I copied, and it looks like I'm getting a good zero-diff run compared to a run on discover.

[gmao-rreichle]

@mathomp4, thanks for keeping us honest. At least I can say that we're generally aware of the issue. The "CO2" file in question had wrongly been placed in the "make_bcs_inputs" directory many moons ago. It is supposed to be part of the bcs (or, going forward, ExtData). Not sure where things stand exactly in terms of the reorganization. A quick check of the "bcs_shared" project directory suggests that @biljanaorescanin is still working on this. The challenge will be to update NAS once we got this sorted out on Discover.
cc: @gmao-jkolassa

[mathomp4]

@gmao-rreichle Well, if it goes into ExtData, then we'd be safe. We definitely sync that over. Then I think it'd be as simple as:

 >>>GCMRUN_CATCHCN<<</bin/ln -s ExtData/g5gcm/land/CO2/v1/CO2_MonthlyMean_DiurnalCycle.nc4 

or whatever since the ExtData link is made by that time. (Obviously, you'd know the right directory structure.)