search

NOAA-EMC / GEFS

4 stars 12 forks source link

Make atmos_prep job work using GFSv17 and new COM structure #119

Closed XianwuXue-NOAA closed 1 year ago

XianwuXue-NOAA commented 1 year ago

Since the rsfcanl.nc is not variable in early cycle enkf, so I copy the c00 data to all ensemble members, I will do the change after rsfcanl.nc is available.

For ensemble member path, I still use c00, p01, ..., p30 for this time, I will open another issue to change the member path from c00, p01, .. to mem000, mem001, ... after all the atm workflow is ready.

The job works on both WCOSS2 and HARA.

XianwuXue-NOAA commented 1 year ago

This probably should've been a few different PRs. (at the very least, changes to use os.path.join() should've been separate). A bit hypocritical of me given the COM refactor PR, but I do usually try to keep PRs to one thing at a time.

Outside the scope of this PR, but we definitely have too much machine-dependent code all over the place.

Thanks Walter. Yes, I agree, at least os.path.join() should be in one PR. I may create a new PR for this. Yes, there is a lot of machine=dependent code, and you may see that I am starting to remove such codes and also some legacy code. I will continue doing this in the following PR. Maybe some of the changes will not close to that PR, however, when I found some codes can be deleted or changed (minor changes), then I will do it in the PR

XianwuXue-NOAA commented 1 year ago

@WalterKolczynski-NOAA I have modified and could you please review my changes? Thanks.