Update subset-mesh capability of the subset_data tool with nco/esmf two-line approach #1884
Conversation
Rework fates test definitions and add new fates tests This tag includes a number of updates to the fates test definitions and test list to gain more coverage: (1) Reorder and update the fates test definitions so that fates satellite phenology mode can be configured for cases using a compset. (2) Add a long-term exact restart test to catch issues that may arise due to updates to fates procedure calls during end of year simulation dates. (3) Add a no-competition + fixed biogeography, non-satellite phenology test definition to provide additional mode combination configuration. (4) Update the fates externals tag to incorporate a fix a vegetation temperature exact restart issue discovered while implementing (1) above. (5) Truncate all testmods starting with "Fates" to mitigate running over the limits on the length of testnames, typically when specifying custom `test_id` using `run_sys_test`. (6) Updates the expected failures list. Resolved conflicts: python/ctsm/subset_data.py
|
This is awesome, Sam. Is there information in the attached google doc that
would be helpful as a discussion post before we document this properly?
…On Wed, Oct 26, 2022 at 6:15 PM Samuel Levis ***@***.***> wrote:
Description of changes
Starting from #1735 <#1735> I will
introduce to the subset_data tool a two-line subset-mesh approach mentioned
elsewhere, e.g. here
<#1513 (comment)>.
Until this PR is merged, one can use the "two-line approach" manually, as
shown in a couple of detailed examples here
<https://docs.google.com/document/d/1kyd0AFNi3YcuzWbcHVtxAV3DHVM7O3amRl1agKcsf3Q/edit?usp=sharing>
.
Specific notes
Contributors other than yourself, if any:
@jkshuman <https://github.com/jkshuman>
CTSM Issues Fixed (include github issue #): #1513
<#1513>
------------------------------
You can view, comment on, or merge this pull request online at:
#1884
Commit Summary
- 0ccf855
<0ccf855>
a first draft of implementing xarray mesh
- c3d981d
<c3d981d>
subset_mesh capabilities
- 66f46c3
<66f46c3>
a few changes
- 29ed6fa
<29ed6fa>
clean ups
- 9155ec8
<9155ec8>
adding write shell commands
- 18c9ca7
<18c9ca7>
clean ups
- b8c9115
<b8c9115>
adding bound checks
- 7682185
<7682185>
adding w/e s/n instead of start/end to help
- 60b232f
<60b232f>
lnd for land_domain_mesh
- 623da32
<623da32>
Merge tag 'ctsm5.1.dev112' into subset_mesh_slevis
File Changes
(3 files <https://github.com/ESCOMP/CTSM/pull/1884/files>)
- *M* python/ctsm/site_and_regional/regional_case.py
<https://github.com/ESCOMP/CTSM/pull/1884/files#diff-0553b092b0109c5e732670da06b8fb05e1dd83e954a0445d9dfd77791339c300>
(248)
- *M* python/ctsm/subset_data.py
<https://github.com/ESCOMP/CTSM/pull/1884/files#diff-547250bb217c7e1d1e4b205d2af2ec6be9edaaf1b67c3720186cfd362ed6a3c6>
(110)
- *M* tools/site_and_regional/default_data.cfg
<https://github.com/ESCOMP/CTSM/pull/1884/files#diff-4c88f30b100e209a7f24ca5218102568774ce58b0a2efb4fac23d2a9fea0ff05>
(2)
Patch Links:
- https://github.com/ESCOMP/CTSM/pull/1884.patch
- https://github.com/ESCOMP/CTSM/pull/1884.diff
—
Reply to this email directly, view it on GitHub
<#1884>, or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AB5IWJFGXQ62J733EHYM5QTWFHCRXANCNFSM6AAAAAARPQKHEM>
.
You are receiving this because you are subscribed to this thread.Message
ID: ***@***.***>
|
|
@glemieux had the same suggestion, and I think it's a good idea.
…On Wed, Oct 26, 2022 at 5:43 PM will wieder ***@***.***> wrote:
This is awesome, Sam. Is there information in the attached google doc that
would be helpful as a discussion post before we document this properly?
Message ID: ***@***.***>
|
|
In this morning's SE meeting (2022/10/27), after some back and forth, we concluded that I should wait and that
|
|
@billsacks given that this subsetting capability is to obtain regions for use with CTSM and NUOPC, can we assume that the user will have both nco and esmf? |
We can assume that the user has esmf, but not necessarily nco. However, even for esmf, the challenge is ensuring that the user's path is set up correctly to have the esmf tools in their path. I imagine for most users, they have no idea how to load the appropriate esmf module to ensure this, instead relying on CIME to do the correct module loads behind the scenes – but that mechanism wouldn't be invoked for this standalone tool. My feeling is that @slevisconsulting 's approach is an excellent stop-gap measure, but I began to feel that it doesn't seem like a great long-term solution because of these portability (or user education) issues, so it felt to me that spending a day to implement something temporary like this might not be a good use of time. (That day could be better spent working on the python-based solution.) |
|
Thanks for the explanation @billsacks I understand your concerns on this with regard to relying on CIME, and that these CIME loads are not used for this stand alone tool. Given your explanation of this, it is likely to eventually fail due to the disconnect between modules. Should @negin513 PR #1735 go back onto the NUOPC project board? |
I put it back on the upcoming tags board. The relevant issue is still on the NUOPC (dropping MCT) project board: https://github.com/orgs/ESCOMP/projects/2/views/12 |
|
@negin513 that's awesome to hear. We do want to be respectful of your time in your new position. I'm wondering if you and I could meet next week to go over where you are at in the process? Even if you finish it out, it would be good for you to explain it to me. Thanks so much... |
|
@ekluzek Thanks for your message. I would like to give a quick presentation at this CTSM SE meeting (Nov 3 2022) about how this works and what additional capabilities it has. |
|
@negin513 awesome. I'll add your presentation as one of the first things in the agenda. It'll be good to hear from you on Thursday. Take care. |
Description of changes
Starting from #1735 I will introduce to the subset_data tool a two-line subset-mesh approach mentioned elsewhere, e.g. here.
Until this PR is merged, one can use the "two-line approach" manually, as shown in a couple of detailed examples here.
Specific notes
Contributors other than yourself, if any:
@jkshuman
CTSM Issues Fixed (include github issue #): #1513