CTSM New updates on MEGANv2.1

Description of changes

We added new features to MEGANv2.1 for simulating isoprene emissions based on three recent studies conducted at the BAI lab in the University of California, Irvine. The first one is about the drought impact on isoprene (Wang et al., 2022). The second study investigates the effect of temperature history on the emission factors of boreal broadleaf deciduous shrubs (Wang et al., 2024a). The third study explores a different temperature response curve for C3 Arctic grass (Wang et al., 2024b, under press). These changes improved the model's representation of isoprene emissions during drought and in high-latitude ecosystems. This work is related to issue #1323 and is based on the old MEGANv2.1 framework, but incorporates new scientific insights.

Specific notes

We modified the impact of drought based on the water stress function (β) in the CLM model. We adopted equation (11) from Wang et al., (2022) as: , where a1 (=-7.45) and b1(=3.26) are empirical parameters. The function “get_gamma_SM” is modified to simulate the change of substrate supply under the water stress. Second and third modifications are for the high-latitude isoprene emitters including boreal broadleaf deciduous shrubs (BBDS) and C3 Arctic grass (C3AG). These results come from the Biogenic Emission and Aerosol Response on the North Slope (BEAR-oNS) project. These modifications are implemented in “get_gamma_T” function in the code. For the BBDS, we included the adaptation of isoprene emission factors to the mean temperature of the previous day with the quation as: , where T₂₄ (K) denotes the mean air temperature of the preceding day (Wang et al., 2024a). For the C3AG, we adopted a dynamic temperature response curve adapted to the record of growing temperature as: . C_C3AG is the activation energy for the isoprene temperature response of C3AG and changes with the mean air temperature of the preceding 10 days (T₂₄₀) as: . The impact of T₂₄₀ on isoprene emission factor of C3AG is as: .

Any User Interface Changes (namelist or namelist defaults changes)? NO. Does this create a need to change or add documentation? Did you do so? I could provide the modification to the scientific document for these changes after they have been added.

Testing performed, if any: I tested the code on Derecho with the I2000Clm50BgcCrop compsets, and it ran well. The scientific impact of the code for isoprene flux estimations can be found in the corresponding references.

Reference

Wang, H., Lu, X., Seco, R., Stavrakou, T., Karl, T., Jiang, X., et al. (2022). Modeling isoprene emission response to drought and heatwaves within MEGAN using evapotranspiration data and by coupling with the community land model. Journal of Advances in Modeling Earth Systems, 14, e2022MS003174. https://doi.org/10.1029/2022MS003174 Wang, H., Welch, A. M., Nagalingam, S., Leong, C., Kittitananuvong, P., Barsanti, K. C., et al. (2024a). Arctic heatwaves could significantly influence the isoprene emissions from shrubs. Geophysical Research Letters, 51, e2023GL107599. https://doi.org/10.1029/2023GL107599 Wang, H., Welch, A. M., Nagalingam, S., Leong, C., Czimczik, C., Tang, J., et al. (2024b): High temperature sensitivity of Arctic isoprene emissions explained by sedges (accepted by Nature Communications, in press).

Jun 07 '24 00:06 HuiWangWanderInGitHub

Thanks for opening this PR @HuiWangWanderInGitHub.

We'll start reviewing this PR, but a quick look seems like this should be straightforward to implement. In the meantime, I also would like to ask that you open a PR that updates the MEGAN documentation in the CTSM technical note.

Directions for this are posted on the wiki

@olyson, can you help with this?

Jun 10 '24 15:06 wwieder

Purely samantic question, can we call this PR MEGANv2.2, or is this confusing with broader MEGAN documentation?

Also @lkemmons, can you help clarify the prioritization of this PR in relation to the science code freeze (July 31 deadline) for cam-chem?

Jun 10 '24 16:06 wwieder

There is a more condensed version of how to update the documentation here:

https://docs.google.com/document/d/1aLlu6C6DJiiO2T21qnj54WFCYuSdGw9F/edit

Once you've issued a PR for the new documentation and verified the documentation builds successfully, we can take it from there.

Jun 11 '24 14:06 olyson

For naming - Alex recommends calling the updated version 'MEGAN-CLM6' (or whatever the appropriate label is for the version of CLM/CTSM that it ends up in). Because the stand-alone MEGAN has features that will never be implemented in CLM, it is not really appropriate to try to use the v2.x and v3.x labels here. I will be happy to help verify results in CAM-chem with these updates, to work towards getting this in CESM3, and hope that will be possible.

On Mon, Jun 10, 2024 at 10:03 AM will wieder @.***> wrote:

Purely samantic question, can we call this PR MEGANv2.2, or is this confusing with broader MEGAN documentation?

Also @lkemmons https://github.com/lkemmons, can you help clarify the prioritization of this PR in relation to the science code freeze (July 31 deadline) for cam-chem?

— Reply to this email directly, view it on GitHub https://github.com/ESCOMP/CTSM/pull/2588#issuecomment-2158760624, or unsubscribe https://github.com/notifications/unsubscribe-auth/AH5BH7J5KW6YDNT4M3TTSETZGXE4RAVCNFSM6AAAAABI5WCGSOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNJYG43DANRSGQ . You are receiving this because you were mentioned.Message ID: @.***>

Jun 13 '24 17:06 lkemmons

@HuiWangWanderInGitHub I updated my local copy of this branch to the latest clm tag and tried to push to the PR, but I do not have permissions. Could you add me as collaborator? Instructions: Click on your branch CLM5_MEGAN_ISOP at the top of this screen, then Settings, then "Collaborators" and you should see a way to add me as a collaborator there. UPDATE: collaborator access granted, thank you!

./run_sys_tests -s aux_clm -c ctsm5.2.008 --skip-generate aux_clm test-suite on derecho FAIL (see next post) aux_clm on izumi FAIL (same as derecho)

Jul 06 '24 19:07 slevis-lmwg

Large number of tests differ from the baseline (see /glade/derecho/scratch/slevis/tests_0706-132231de).

Most variables with diffs have the MEG_ prefix or _voc suffix, which seems expected.
Some do not, but only in two tests:

ERS_D_Ld15.f45_f45_mg37.I2000Clm50FatesRs.derecho_intel.clm-FatesColdTwoStream
PEM_D_Ld9.ne30pg3_t232.I1850Clm60BgcCropG.derecho_intel.clm-clm60cam6LndTuningMode

The former has diffs in FATES variables (clm file) and in lndImp variables (cpl file). The latter has diffs in the cism file, which seems somewhat related to #2542. For now I will let others weigh in with an assessment of these results.

UPDATE 1: Same result on izumi with ERS_D_Ld15.f45_f45_mg37.I2000Clm50FatesRs.izumi_nag.clm-FatesColdTwoStream having diffs in FATES variables (clm file) and in lndImp variables (cpl file).

UPDATE 2: I tried the 3 failing tests in vanilla ctsm5.2.008: All pass against baseline. The ERS tests fail with COMPARE_base_rest (expected). The PEM test fails with COMPARE_base_modpes (expected).

UPDATE 3: I backed out my last commit and repeated the 3 tests: On derecho the PEM test still failed. On derecho and izumi the ERS test now passed.

Jul 07 '24 00:07 slevis-lmwg

The failing test has changed in ctsm5.2.012 to PEM_D_Ld9.ne30pg3_t232.I1850Clm60BgcCropG.derecho_intel.clm-clm60cam7LndTuningMode and it still fails, so I will proceed with troubleshooting.

Jul 16 '24 22:07 slevis-lmwg

This PR changes the calculation of gamma_sm and of gamma_t. In my troubleshooting I introduced the gamma_sm and the gamma_t changes separately. I also introduced the two new parts of gamma_t (Eopt and gamma_t_LDF) separately.

Regardless, the above PEM test gives new results in these three files. The one I'm concerned about is cism because it seems unexpected, but maybe it's fine:

PEM_D_Ld9.ne30pg3_t232.I1850Clm60BgcCropG.derecho_intel.clm-clm60cam7LndTuningMode.C.20240717_145335_n9yr9j.cism.gris.initial_hist.0001-01-01-00000.nc.base.cprnc.out
PEM_D_Ld9.ne30pg3_t232.I1850Clm60BgcCropG.derecho_intel.clm-clm60cam7LndTuningMode.C.20240717_145335_n9yr9j.clm2.h0.0001-01-10-00000.nc.cprnc.out
PEM_D_Ld9.ne30pg3_t232.I1850Clm60BgcCropG.derecho_intel.clm-clm60cam7LndTuningMode.C.20240717_145335_n9yr9j.cpl.hi.0001-01-10-00000.nc.cprnc.out

Jul 17 '24 21:07 slevis-lmwg

@slevis-lmwg there's another issue where we know CISM changes answers on processor count. So do the test without CISM which just means dropping the G at the end of the compset.

That will probably mean you have to also run the baseline as well.

Jul 17 '24 22:07 ekluzek

@slevis-lmwg there's another issue where we know CISM changes answers on processor count. So do the test without CISM which just means dropping the G at the end of the compset.

That will probably mean you have to also run the baseline as well.

From master: ./create_test PEM_D_Ld9.ne30pg3_t232.I1850Clm60BgcCrop.derecho_intel.clm-clm60cam7LndTuningMode -g /glade/campaign/cgd/tss/ctsm_baselines/ctsm5.2.012 and from the branch: ./create_test PEM_D_Ld9.ne30pg3_t232.I1850Clm60BgcCrop.derecho_intel.clm-clm60cam7LndTuningMode -c /glade/campaign/cgd/tss/ctsm_baselines/ctsm5.2.012

Now, the diffs from baseline are limited to MEG_ and _voc variables, as expected.

Jul 17 '24 22:07 slevis-lmwg

I now realize my mistake in thinking that the PEM failure was unexpected: The cism cprnc file (such as the one I posted above) ends in "base.cprnc.out" which is what causes the expected COMPARE_base_modpes failure here and in issue #2542.

This means that, as written, the code passes the aux_clm tests.

Jul 18 '24 15:07 slevis-lmwg

@HuiWangWanderInGitHub I'm happy to make the revisions in response to my review; however,

[x] could you provide me with names for the various constants/coefficients that currently appear without names?

Jul 18 '24 16:07 slevis-lmwg

@HuiWangWanderInGitHub I'm happy to make the revisions in response to my review; however,

[x] could you provide me with names for the various constants/coefficients that currently appear without names?

Jul 22 '24 23:07 HuiWangWanderInGitHub

@HuiWangWanderInGitHub I'm happy to make the revisions in response to my review; however,

[x] could you provide me with names for the various constants/coefficients that currently appear without names?

Hi @slevis-lmwg ,

Thank you so much for your help and sorry for accidentally closing the issue. Here are the names of the unnamed variables: bet_arc_c3: activity factor for temperature of Arctic C3 grass bet_arc_c3_max: Maximum value for the activity factor for temperature of Arctic C3 grass. Please let me know if other variables need to be named.

Thank you.

Jul 22 '24 23:07 HuiWangWanderInGitHub

@HuiWangWanderInGitHub I will post a new code review with the constants/coefficients that I would prefer ideally to be named.

Jul 22 '24 23:07 slevis-lmwg

I will submit aux_clm once more to consider this PR ready: ./run_sys_tests -s aux_clm -c ctsm5.2.017 --skip-generate

@HuiWangWanderInGitHub it would be great if you could look at my changes to your PR and let me know if you approve or have concerns. Thanks!

Aug 01 '24 22:08 slevis-lmwg

aux_clm results izumi OK derecho FAIL but I'm inclined to consider OK unless I get pushback:

Expected failures: the two FATES tests with strange diffs from the baseline (variable FATES_TRANSITION_MATRIX_LULU) due to issue #2656
Two other tests show diffs from the baseline that I think are NOT expected; however, they are the two matrixcn tests that change answers with threading. I'm inclined to ignore these as not worthy of investigation unless I get pushback. These are the diffs I see:

ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS LIT_CELN_Cap_vr                  1.2211E-03            NORMALIZED  3.8576E-04
ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS SOM_ACTN_Cap_vr                  1.4807E-18            NORMALIZED  1.3685E-19
ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS SOM_PASN_Cap_vr                  9.1732E-01            NORMALIZED  3.6759E-03
ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS SOM_SLON_Cap_vr                  4.6046E-17            NORMALIZED  2.2842E-19
ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS LIT_LIGC_Cap_vr                  1.3700E-01            NORMALIZED  1.6987E-03
ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS LIT_METC_Cap_vr                  1.1999E-01            NORMALIZED  4.2125E-03
ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS SOM_ACTC_Cap_vr                  1.1375E+01            NORMALIZED  1.0767E-01
ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/ERP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS SOM_SLOC_Cap_vr                  2.2510E-17            NORMALIZED  8.2173E-21
REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS CWDC_Cap_vr                      9.3301E-03            NORMALIZED  6.7713E-06
REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS LIT_CELC_Cap_vr                  5.6365E-14            NORMALIZED  3.2327E-16
REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS SOM_ACTC_Cap_vr                  9.7547E-20            NORMALIZED  9.2266E-22
REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int/REP_P64x2_Lm13.f10_f10_mg37.IHistClm60Bgc.derecho_intel.clm-monthly--clm-matrixcnOn_ignore_warnings.C.0801-165723de_int.clm2.h0.1851-02-01-00000.nc.cprnc.out: RMS SOM_SLOC_Cap_vr                  3.1215E-18            NORMALIZED  1.1395E-21

Aug 02 '24 18:08 slevis-lmwg

@slevis-lmwg this looks to me to be the same as what I saw in my earlier tag here:

https://github.com/ESCOMP/CTSM/pull/2545#issuecomment-2249546181

So yes this is good from a testing perspective.

Aug 02 '24 18:08 ekluzek

In preparation for the merge, I have updated to ctsm5.2.019 and submitted aux_clm: OK (same behavior reported earlier in this PR)

Aug 12 '24 17:08 slevis-lmwg