Enable rough topograph and add remaining atm and lnd settings for v3b02 #6113
Conversation
Rough topography for atm is enabled by replacing 16x topography with x6t ones. This PR replaces the default topo file for ne30pg2 and ne120pg2. x6t topo for other production grids are pending. Companion settings to go with rough topo are set with pgrad_correction = 1 and hv_ref_profiles = 6. These options can also be beneficial when using smoother topography. The PR also reset dust_emis_fact and first and last year of pdepdyn stream files for 2010 and 1950 compsets to match the stream file used. [Non-BFB] [NML]
|
|
Hi @wlin7, just putting this issue on your radar to potentially include in this PR (or a follow-up PR since it is after v3b02). This land PR #5981 makes v stream_year_first_fan=2010
stream_year_last_fan=2010Tangentially, do we have a present-day finidata for icos_trigrid_top_bgc? I tried using the pre-industrial one we have available and it gave me an elaborate runtime error about weights being far apart (I was using fsurdat from 2010 with an finidat meant for 1850; I could have turned the error into a warning, but opted to wait and potentially use something I produce later). |
|
@mahf708 are you sure its a vanilla F2010? We have lots of F2010 cases in the test suite and they didn't fail after the final merge to next/master of that PR. |
Sorry, it's "vanilla" for v3, i.e., including the latest v3 settings and in particular, this: Do we have F2010 tests with the additional settings for v3? |
|
@rljacob, see repro below. MRE: script
#!/bin/bash -fe
readonly CODE_ROOT="/home/ac.ngmahfouz/misc/full-e3sm/fa40af6/e3sm"
readonly CASE_NAME="231213.MRE"
readonly CASE_ROOT="/lcrc/group/e3sm/${USER}/${CASE_NAME}"
readonly CASE_SCRIPTS_DIR="${CASE_ROOT}/case_scripts"
readonly COMPSET="F2010"
readonly RESOLUTION="ne30pg2_r05_IcoswISC30E3r5"
readonly MACHINE="chrysalis"
readonly PROJECT="e3sm"
readonly WALLTIME="2:00:00"
readonly PELAYOUT="M"
readonly CASE_BUILD_DIR=${CASE_ROOT}/build
readonly CASE_ARCHIVE_DIR=${CASE_ROOT}/archive
readonly CASE_RUN_DIR=${CASE_ROOT}/run
${CODE_ROOT}/cime/scripts/create_newcase \
--case ${CASE_NAME} \
--output-root ${CASE_ROOT} \
--script-root ${CASE_SCRIPTS_DIR} \
--handle-preexisting-dirs u \
--compset ${COMPSET} \
--res ${RESOLUTION} \
--machine ${MACHINE} \
--project ${PROJECT} \
--walltime ${WALLTIME} \
--pecount ${PELAYOUT}
pushd ${CASE_SCRIPTS_DIR}
./xmlchange EXEROOT=${CASE_BUILD_DIR}
./xmlchange RUNDIR=${CASE_RUN_DIR}
./xmlchange --file env_run.xml --id ELM_BLDNML_OPTS --val "-bgc bgc -nutrient cnp -nutrient_comp_pathway rd -soil_decomp ctc -methane"
./case.setup --reseterror
|
|
No v3 isn't in the testing yet. That must be it. |
|
I edited the original comment (v |
|
@mahf708 , thanks for bringing this to our attention. I did notice since late last week's update, an option related to FAN is added. The v3 @bishtgautam , can you please also give a check on this? @mahf708 , regarding your other comment |
|
Tagging @bbye and @peterdschwartz |
|
out of curiosity, why do this options need to be protected inside if (defined()) blocks? (most options dont): |
Hi @mt5555 , a default value is only available with |
|
@wlin7 : since these are dycore parameters independent of physics, I think it would be slightly better to adopt the alternative (remove phys='default') and make them defined for all cases. This matches how we treat most of the other dycore namelist settings. |
|
I think the fix for the fan stream file is to modify the namelist_defaults.xml to use the use_fan instead of the use_cn flag. Do you want me to made a separate PR to resolve that? |
| <stream_year_first_pdep use_cn=".true." sim_year="2010" >2000</stream_year_first_pdep> | ||
| <stream_year_last_pdep use_cn=".true." sim_year="2010" >2000</stream_year_last_pdep> | ||
| <stream_year_first_pdep use_cn=".true." sim_year="2000" >2000</stream_year_first_pdep> | ||
| <stream_year_last_pdep use_cn=".true." sim_year="2000" >2000</stream_year_last_pdep> | ||
| <stream_year_first_pdep use_cn=".true." sim_year="1950" >2000</stream_year_first_pdep> | ||
| <stream_year_last_pdep use_cn=".true." sim_year="1950" >2000</stream_year_last_pdep> |
There was a problem hiding this comment.
For sim_year=2010 or sim_year=1950, why is the pdep corresponding to 2000 being used?
| <stream_year_first_pdep phys="elm" use_cn=".true." ndepsrc="stream" >2000</stream_year_first_pdep> | ||
| <stream_year_last_pdep phys="elm" use_cn=".true." ndepsrc="stream" >2000</stream_year_last_pdep> |
There was a problem hiding this comment.
Same comment as https://github.com/E3SM-Project/E3SM/pull/6113/files#r1440056541.
| <stream_year_first_pdep phys="elm" use_cn=".true." ndepsrc="stream" >2000</stream_year_first_pdep> | ||
| <stream_year_last_pdep phys="elm" use_cn=".true." ndepsrc="stream" >2000</stream_year_last_pdep> |
There was a problem hiding this comment.
Same comment as https://github.com/E3SM-Project/E3SM/pull/6113/files#r1440056541.
There was a problem hiding this comment.
Good point, @bishtgautam . Intuitively, when first testing it, did try 2010. But look like only simyr2000 stream file is available, which is used for piControl as well, so I thought it being treated as time-invariant. If time varying for this data stream is necessary, we will need quick help to generate the files for compsets of different TIME (or forcing scenarios)
The file is : inputdata/lnd/clm2/pdepdata/fpdep_clm_hist_simyr2000_1.9x2.5_c150929.nc
|
Should go in before #6108 |
Rough topography for atm is enabled by replacing 16x topography with x6t ones. This PR replaces the default topo file for ne30pg2 and ne120pg2. x6t topo for other production grids are pending. Companion settings to go with rough topo are set with pgrad_correction = 1 and hv_ref_profiles = 6. These options can also be beneficial when using smoother topography. The PR also reset dust_emis_fact and first and last year of pdepdyn stream files for 2010 and 1950 compsets to match the stream file used. [Non-BFB] All tests with EAM, or ELM for simyr 2010 and 1950 [NML]
|
merged to next. |
|
Note: FDPSCREAM-ARM97 test returns diff on 11 of 243 eam.h0 fields (QDIFF, TDIFF, and *_RES variables). Other common fields (e.g., T, Q, U, V, P3 fields, precipitation and radiation fields) are not affected. Here mainly to document the test, the PR should go ahead nevertheless. |
|
Noting other test issues: |
|
Re: SMS_Ln9_P24x1.ne4_ne4.FDPSCREAM-ARM97, it does have the 2 new atm_in namelist vars: |
|
@wlin7 that is odd as all those variables (QDIFF, TDIFF, *RES) are computed from the state. So it's weird that one set would be b4b and the other not.... One commonality of the output variables listed that are not b4b is that they are all computed in the components/eam/src/dynamics/se/se_iop_intr_mod.F90 file. This provides a clue but looking at the PR (and the namelist changes that @rljacob pointed out), I can't explain off of the top of my head why these diffs are present. I will have to chew on this. |
|
I'm going to go ahead and merge this to master and bless all diffs. I'll open an issue about that SCREAM test to track it. |
|
Thanks for taking care of the integration, @rljacob . Yes, we will look into the DPSCREAM behavior separately. |
|
e3sm_developer_next_gnu reported later. Only one test did not run: SMS_Ld1.hcru_hcru.I1850CRUELMCN.pm-cpu_gnu for unknown reasons. Diffs blessed. |
|
ERS_Ld5.ne4_oQU240.F2010.chrysalis_intel.eam-rrtmgp ran the following day and was blessed. |
Rough topography for atm is enabled by replacing 16x topography with x6t ones. This PR replaces the default topo file for ne30pg2 and ne120pg2. x6t topo for other production grids are pending. Companion settings to go with rough topo are set with pgrad_correction = 1 and hv_ref_profiles = 6. These options can also be beneficial when using smoother topography.
The PR also reset dust_emis_fact and first and last year of pdepdyn stream files for 2010 and 1950 compsets to match the stream file used.
[Non-BFB] All tests with EAM, or ELM for simyr 2010 and 1950
[NML]