You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
b4b-dev: Remove references to cpl7 and mct (the latter is done) #2294
Discussed in meeting, this is important to do. Really want this done before CESM3, and it's in our planning now for that. Right now we don't have someone to work on this.
@ekluzek As I understand it it is no longer needed in ctsm, or mosart and we have removed it from csm_share. If you do not remove it from slim prior to cesm3 then I propose that slim is not released with cesm3. What other requirement is there for mct?
I think this issue is about LILAC, not SLIM, but maybe I'm wrong.
Maybe it's worth hearing from @dlawrenncar what the plan for LILAC should be moving forward?
@jedwards4b LILAC is not just a single processor app. LILAC is about coupling CTSM to other atmospheric models rather than through CESM. So coupling to WRF for example. So it'll use multiple processors just as you would with CTSM or WRF on their own.
It looks like MCT is being used for CPL7-data-models for LILAC to fill the roll of a DATM for the testing. ESMF is widely used in LILAC, and the CTSM cap for LILAC is an ESMF component, so it shouldn't be too hard to do this, but it's not trivial either. I think when LILAC was developed CDEPS wasn't mature enough to use it yet. Now we just need to switch the use of CPL7/DATM to CDEPS/DATM.
We don't currently have experts in LILAC right now. @billsacks is on ESMF, but maybe we could get some consulting time from him? And Negin is in CISL now. So we could only ask her as a favor for us. Right now we don't have someone to work on this, but hopefully can before CESM3 release. And @wwieder is right we need to assess the importance of supporting LILAC for going forward.
It looks like MCT is being used for CPL7-data-models for LILAC to fill the roll of a DATM for the testing. ESMF is widely used in LILAC, and the CTSM cap for LILAC is an ESMF component, so it shouldn't be too hard to do this, but it's not trivial either. I think when LILAC was developed CDEPS wasn't mature enough to use it yet. Now we just need to switch the use of CPL7/DATM to CDEPS/DATM.
Thanks a lot for your investigation of this @ekluzek . I think this is mostly right, but a minor correction is that I think the use of MCT / data model functionality is to fill in fields that might be missing from the actual atmosphere. Currently this is used for prescribing aerosols. I agree that the right path forward is to change lilac_atmaero.F90 to use CDEPS streams.
(LILAC also has a fake atmosphere driver, lilac/atm_driver/atm_driver.F90, which serves a similar purpose as DATM for testing, though much simpler. But I don't think that relies on MCT: it just uses hard-coded data.)
I am hesitant to keep adding to SE burdens, but having a regional model
capability, coupled to CTSM is an important feature. That could be through
LILAC or it could be if WRF was coupled into CESM infrastructure ... or it
could be if MPAS was coupled into CESM with regional modeling capability
(not working, right?) ... or it could be possibly even be MPAS coupled to
CESM with regionally-refined grid and atmosphere nudging outside the
regionally-refined domain. My understanding is that the last option should
be working now (?), though I don't really know the relative costs and
whether or not this is really a good way to have regional modeling
capability. This might be part of a bigger CESM Project discussion on
regional modeling and what we want to and/or can support.
On Thu, Feb 15, 2024 at 2:11 PM Bill Sacks ***@***.***> wrote:
It looks like MCT is being used for CPL7-data-models for LILAC to fill the
roll of a DATM for the testing. ESMF is widely used in LILAC, and the CTSM
cap for LILAC is an ESMF component, so it shouldn't be too hard to do this,
but it's not trivial either. I think when LILAC was developed CDEPS wasn't
mature enough to use it yet. Now we just need to switch the use of
CPL7/DATM to CDEPS/DATM.
Thanks a lot for your investigation of this @ekluzek
<https://github.com/ekluzek> . I think this is mostly right, but a minor
correction is that I think the use of MCT / data model functionality is to
fill in fields that might be missing from the actual atmosphere. Currently
this is used for prescribing aerosols. I agree that the right path forward
is to change lilac_atmaero.F90 to use CDEPS streams.
(LILAC also has a fake atmosphere driver, lilac/atm_driver/atm_driver.F90,
which serves a similar purpose as DATM for testing, though much simpler.
But I don't think that relies on MCT: it just uses hard-coded data.)
—
Reply to this email directly, view it on GitHub
<#2294 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AFABYVACZOQTVMYDPVH5DFTYTZ2ZDAVCNFSM6AAAAABAW3SFJ6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNBXGM2DKNZTGE>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
@dlawrenncar I agree that regional modeling should be part of the longer term plan, but do we need it for cesm3? LILAC has a dependency on mct - a library that cseg has long slated for removal in cesm3. As far as I can tell it's the last place that is dependent on mct - if we leave mct in cesm3 we are on the hook to support it for many years to come - I think LILAC could be updated to have that dependency removed with fairly small effort, but the ctsm group is saying that LILAC may not be part of the longer term plan and maybe it should just be removed altogether. So I am looking for a response to a very specific question - update LILAC to remove the dependence on mct or remove LILAC from the cesm3 release. Doing nothing and releasing cesm3 with an mct dependence is not an option I want to consider.
@dlawrenncar can we make a commitment that removing MCT is a requirement of the CESM3 release? I think @jedwards4b is right about this in terms of future support. If we make that commitment that we will hold up the release for anything requires MCT. So for LILAC that means either updating it, or removing it. The other linchpin is SLIM, and I'd like to say that we hold up the release until it's updated to not require MCT as well.
I'm pretty sure we could get agreement in both CSEG and SEWG that the CESM3 release should require the removal of MCT. If we can have that commitment we don't have to stress about this. @dlawrenncar should @jedwards4b make the case to the CESM project meeting in this regard? Or it could be @briandobbins as well.
Technically it looks like there are a few references to cpl7 that we should remove and make sure the LILAC tests still work:
git grep -i cpl7 # shows the following...
bld/CLMBuildNamelist.pm: # do nothing - use value provided by config_grid.xml and clm.cpl7.template
lilac/CMakeLists.txt:add_subdirectory(${CIME_ROOT}/../components/cpl7/driver drv_share)
lilac/src/lilac_driver_pio_mod.F90: ! components/cpl7/driver/main/driver_pio_mod.F90, except (1) the module name has been
But, ALL of the important things are done here, and that's just a really minor cleanup, that isn't critical to do. It just removes some things that could cause confusion. So I made it low priority.
Activity
jedwards4b commentedon Dec 15, 2023
Also components/clm/src/unit_test_stubs/csm_share/CMakeLists.txt
components/clm/src/unit_test_stubs/csm_share/mct_mod_stub.F90
ekluzek commentedon Feb 15, 2024
Discussed in meeting, this is important to do. Really want this done before CESM3, and it's in our planning now for that. Right now we don't have someone to work on this.
jedwards4b commentedon Feb 15, 2024
@ekluzek As I understand it it is no longer needed in ctsm, or mosart and we have removed it from csm_share. If you do not remove it from slim prior to cesm3 then I propose that slim is not released with cesm3. What other requirement is there for mct?
wwieder commentedon Feb 15, 2024
I think this issue is about LILAC, not SLIM, but maybe I'm wrong.
Maybe it's worth hearing from @dlawrenncar what the plan for LILAC should be moving forward?
jedwards4b commentedon Feb 15, 2024
@wwieder Thanks for the clarification - it looks like lilac is a single processor application - am I reading that correctly?
ekluzek commentedon Feb 15, 2024
@jedwards4b LILAC is not just a single processor app. LILAC is about coupling CTSM to other atmospheric models rather than through CESM. So coupling to WRF for example. So it'll use multiple processors just as you would with CTSM or WRF on their own.
It looks like MCT is being used for CPL7-data-models for LILAC to fill the roll of a DATM for the testing. ESMF is widely used in LILAC, and the CTSM cap for LILAC is an ESMF component, so it shouldn't be too hard to do this, but it's not trivial either. I think when LILAC was developed CDEPS wasn't mature enough to use it yet. Now we just need to switch the use of CPL7/DATM to CDEPS/DATM.
We don't currently have experts in LILAC right now. @billsacks is on ESMF, but maybe we could get some consulting time from him? And Negin is in CISL now. So we could only ask her as a favor for us. Right now we don't have someone to work on this, but hopefully can before CESM3 release. And @wwieder is right we need to assess the importance of supporting LILAC for going forward.
billsacks commentedon Feb 15, 2024
Thanks a lot for your investigation of this @ekluzek . I think this is mostly right, but a minor correction is that I think the use of MCT / data model functionality is to fill in fields that might be missing from the actual atmosphere. Currently this is used for prescribing aerosols. I agree that the right path forward is to change lilac_atmaero.F90 to use CDEPS streams.
(LILAC also has a fake atmosphere driver, lilac/atm_driver/atm_driver.F90, which serves a similar purpose as DATM for testing, though much simpler. But I don't think that relies on MCT: it just uses hard-coded data.)
dlawrenncar commentedon Feb 16, 2024
jedwards4b commentedon Feb 16, 2024
@dlawrenncar I agree that regional modeling should be part of the longer term plan, but do we need it for cesm3? LILAC has a dependency on mct - a library that cseg has long slated for removal in cesm3. As far as I can tell it's the last place that is dependent on mct - if we leave mct in cesm3 we are on the hook to support it for many years to come - I think LILAC could be updated to have that dependency removed with fairly small effort, but the ctsm group is saying that LILAC may not be part of the longer term plan and maybe it should just be removed altogether. So I am looking for a response to a very specific question - update LILAC to remove the dependence on mct or remove LILAC from the cesm3 release. Doing nothing and releasing cesm3 with an mct dependence is not an option I want to consider.
ekluzek commentedon Feb 16, 2024
@dlawrenncar can we make a commitment that removing MCT is a requirement of the CESM3 release? I think @jedwards4b is right about this in terms of future support. If we make that commitment that we will hold up the release for anything requires MCT. So for LILAC that means either updating it, or removing it. The other linchpin is SLIM, and I'd like to say that we hold up the release until it's updated to not require MCT as well.
I'm pretty sure we could get agreement in both CSEG and SEWG that the CESM3 release should require the removal of MCT. If we can have that commitment we don't have to stress about this. @dlawrenncar should @jedwards4b make the case to the CESM project meeting in this regard? Or it could be @briandobbins as well.
28 remaining items
wwieder commentedon Feb 5, 2025
@ekluzek should this issue be closed?
If not, how close are we to having the code cleanup's from alpha04a merged to master?
ekluzek commentedon Feb 5, 2025
Technically it looks like there are a few references to cpl7 that we should remove and make sure the LILAC tests still work:
git grep -i cpl7 # shows the following...
But, ALL of the important things are done here, and that's just a really minor cleanup, that isn't critical to do. It just removes some things that could cause confusion. So I made it low priority.
slevis-lmwg commentedon Jan 14, 2026
Based on the most recent post, I removed "done"
[-]use of mct_mod is deprecated [/-][+]b4b-dev: Remove references to cpl7 and mct (the latter is done)[/+]