Update the BVOC emission model in CLM/CESM from MEGANv2.1 to MEGANv3.1.

[HuiWangWanderInGitHub]

The project we proposed will update the BVOC emission model in CLM/CESM from MEGANv2.1 to MEGANv3.1. At the first stage, we will update the following aspects in the CLM/CESM model:

1. Expand BVOC species from 150 compounds to 201 compounds that are mapped to CAM-Chem scheme.
2. Consider the effects of environmental stresses on BVOC emission.
3. Update emission factors for the new 20 BVOC categories in MEGANv3.1.

One important improvement of MEGANv3.1 is considering the impact of environmental stresses. We extended BVOC species in MEGAN from original 150 compounds to 201 compounds, and we group these 201 compounds into 20 categories for supporting the stress algorithms. The environmental stresses include drought, heatwave, cold wave, high wind and air quality (Ozone). The drought stress algorithm is coming from Jiang et al. (2018), and the air quality stress algorithm is coming from Ghirardo et al. (2016). Other stress algorithms are from limited available observations and some unpublished lab experiments by UCI Biosphere-Atmosphere Interaction lab led by Prof. Alex Guenther.
We have been collaborating with Dr. Louisa Emmons from NCAR. Currently, we have finished the first two aspects we mentioned above. We already extended the BVOC species in CLM and transplanted the stress algorithms in MEGANv3.1 to CLM. The stress algorithms except for the air quality stress one can successfully run with CLM/CESM. The air quality stress requires the surface ozone concentrations from CAM-chem model that is not delivered to CLM at present through the coupler, so the air quality stress algorithm is not working. Our current work is finalizing the parameters we used in these algorithms and the new version MEGAN model, especially the emission factors we mentioned above.
Here is a list of the modification we already did for updating MEGAN in CLM:

1. We did the extension of species by modifying the “megan_factors_file”.
2. New namelist parameters for controlling the stress algorithms are added in following files:

• ${CESM}/components/clm/src/biogeochem/MEGANFactorsMod.F90
• ${CESM}/cime/src/drivers/mct/shr/shr_megan_mod.F90
• ${CESM}/components/clm/bld/namelist_files/namelist_definition_drv_flds.xml
• ${CESM}/cime/src/drivers/mct/cime_config/namelist_definition_drv_flds.xml

3. We added the codes of stress algorithms to ${CESM}/components/clm/src/biogeochem/VOCEmissionMod.F90.

[ekluzek]

Does this change just add new capability or does it also change answers for VOC's if using the same chemical species as with MEGANv2.1? It sounds like the biggest update is adding additional species. It'll be important for us to know if we should expect answers for VOC's to change for our current test cases that run with MEGAN on.

It's OK if answers do change, and I'd expect they do with a major version update, but we need to check. If answers do change it would be good to know in what capacity? Is if just a roundoff level change? Or more substantial?

[ekluzek]

Do you have a sandbox somewhere we could look at to see these changes? Thanks.

[wwieder]

@HuiWangWanderInGitHub Thanks for letting us know about this project and creating the detailed issue! This looks like an important contribution for CAMchem and the larger CESM project. Passing ozone through the coupler looks like an issue that will help enable new science (#1224, #618), address some old issue in CLM (#270, #134, #267), and introduce some new challenges (#962). Hopefully we can work on a way to bring in your modifications efficiently.

Let us know when you have a PR ready for us to look at. Alternatively, you can submit a PR earlier as you work on this project and get input from @ekluzek and others along the way.

[billsacks]

One other question is whether this is being done in a backwards compatible way so that we can select either the old or new version. Whether this backwards compatibility is needed is partly up to the CAM-Chem group (@lkemmons and others).

[lkemmons]

I don't see a need to keep the capability to run MEGANv2.1 in the new code - it seems better to just update MEGAN. Unless Alex Guenther feels otherwise. Louisa

…

On Thu, Apr 8, 2021 at 9:34 AM Bill Sacks ***@***.***> wrote: One other question is whether this is being done in a backwards compatible way so that we can select either the old or new version. Whether this backwards compatibility is needed is partly up to the CAM-Chem group ( @lkemmons <https://github.com/lkemmons> and others). — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1323 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH5BH7LCNM6LXMPPTJRN3UTTHXEITANCNFSM42NS6TVQ> .

[HuiWangWanderInGitHub]

We add new capacities of considering stresses and also change the answers for VOCs. For the new species, we didn’t put them into the chemistry scheme of CAM-chem right now and still use the old way, but we may change it in the future. We will also update the emission factors with the new data, which could change the emission amount also chemistry in some regions. For the environmental stresses, we just put the framework in the code. Among these stresses, we may have higher confidence in the drought one than others. So, we added the namelist flags for users to decide if they will use these stress algorithms or not. We will prepare a preliminary PR for you guys to look at. Thanks for your help!

[wwieder]

Thanks @HuiWangWanderInGitHub let us know if you need help how to set up and use the namelist flags.

[danicalombardozzi]

@HuiWangWanderInGitHub Great to see this new work to update MEGAN! As @wwieder mentioned, it connects to the older issue #270 that determines ozone damage to plant productivity, which I've worked on. One aspect that still needs consideration for this issue is that data are needed when CAM-chem isn't running interactively, such as for CLM-only simulations. Have you also thought about this, and/or is it relevant to your work?

Given the episodic nature of ozone concentrations, it seems best (at least from a plant damage perspective) to use hourly ozone concentration data rather than daily or monthly. However, this high frequency is challenging -- both in terms of the file size as well as the computational time required to read the file.

It seems as though we should work to connect ozone through the coupler, as mentioned here and in #270, and then address the data needs for simulations without CAM-chem.

[HuiWangWanderInGitHub]

@HuiWangWanderInGitHub Great to see this new work to update MEGAN! As @wwieder mentioned, it connects to the older issue #270 that determines ozone damage to plant productivity, which I've worked on. One aspect that still needs consideration for this issue is that data are needed when CAM-chem isn't running interactively, such as for CLM-only simulations. Have you also thought about this, and/or is it relevant to your work?

Given the episodic nature of ozone concentrations, it seems best (at least from a plant damage perspective) to use hourly ozone concentration data rather than daily or monthly. However, this high frequency is challenging -- both in terms of the file size as well as the computational time required to read the file.

It seems as though we should work to connect ozone through the coupler, as mentioned here and in #270, and then address the data needs for simulations without CAM-chem.

Hi @danicalombardozzi. It is good to know we're working on the same issue of ozone. About the offline ozone inputs, I think it is relevant to my work. I haven't tried to solve this, could we use a low frequency inputs but not that low likes 3h to ensure the scientific part and the efficiency?

[rosiealice]

Just found this issue in the context of an EU project that will compare CLM BVOC emissions with a bunch of other LSMs. Are there any updates on the status of this?

[danicalombardozzi]

The ozone connections in the code are still not finished, but Marianna has plans to do so when she returns from vacation and after the CESM workshop. I'm not sure where the VOC work from @HuiWangWanderInGitHub is at.

[HuiWangWanderInGitHub]

I already have a stable version CESMv2.1.3-MEGAN3.1 before I posted the issue. I'm now trying to fit the code into the latest CESM/CLM. I did some modification recently and I'm testing it right now.

[rosiealice]

Thanks! This is useful for planning. We aren't planning to do this any time soon, just wanting to think about what might be interesting in the medium term.

[wwieder]

Are there any updates on this issue to consider?

[HuiWangWanderInGitHub]

Are there any updates on this issue to consider?

Hi Will. We already finish merging the code, but I feel we may need more time to do more tests and comparisons. Btw, our group has a very new publication on JAMES talking about how we simulate drought stress in MEGAN and CLM, and you can check it here: https://doi.org/10.1029/2022MS003174. Thank you!

[wwieder]

Oh, how did I miss this? I see @fvitt made some changes that came in with ctsm5.1.dev108: Connect ozone from atmosphere. Does that mean we can close this issue, @billsacks and @ekluzek

[lkemmons]

@wwieder You have mixed up issues -- this is for MEGAN emissions, not passing ozone to CTSM.

[wwieder]

Thanks @lkemmons, I didn't see any other MEGAN's updates in response to @HuiWangWanderInGitHub's comment that this was done. Does this MEGAN3.1 work still need to be integrated, or have I missed another PR?

[wwieder]

Just wanting to follow up here, @lkemmons and @HuiWangWanderInGitHub. Does this MEGAN3.1 work still need to be integrated onto the main development branch of CTSM, or have I missed another PR?

[lkemmons]

@HuiWangWanderInGitHub Hui: what is the status of MEGANv3.1? Do you want us to get this on to the main trunk soon, or do you want to do more tests? Thanks!

[HuiWangWanderInGitHub]

@lkemmons @wwieder Hi Louisa and Will, I think the code combination is already done as I mentioned before. If the CAM-chem ozone has been introduced to CLM, I can test the model with the current ozone stress algorithm. So far, we have more confident in drought stress with one publication came out recently, for other aspects like other environmental stresses and emission factors, we may need more time to make it more scientifically solid but still good to have them in the model as the first step. I'm now also involving other projects happening in my group, but I'm still thinking to wrap up this MEGAN3-CESM project and have a manuscript before June this year. So I'm wondering the schedule of the CTSM and CESM group.

[wwieder]

Sorry, this is not my area of expertise.

You can see the MEGAN and VOCE code we currently have on the main development branch of CTSM. I'm not sure if changes to these modules are needed for CAM-chem in CESM3?

[lkemmons]

@HuiWangWanderInGitHub Could you please clarify what the state of the MEGAN3.1 code in CTSM is? Have you started a pull request? Or would you like us to look at your sandbox and check the code changes first? If you are willing to let others start using it, it would be good to start getting it on the main development branch. If you do not want to do this until you have submitted another paper, please let us know.

[rosiealice]

Just pinging this issue. I have a PhD student starting who is interested in BVOCs, and so if there is any status update this would be great! Or indeed we could find an opportunity to discuss potential paths forward in a in a call with MEGAN folks?

[rosiealice]

Noting also that getting MEGAN to run with FATES (#1834) is thus moving higher up our priority list, and so it would be interesting to scope whether these modifications could be coordinated...

[lkemmons]

I just had a chat with @HuiWangWanderInGitHub and Alex Guenther - Hui will be working on implementing MEGAN3 over the next couple of months.
Also, Alex pointed out that the CO2-inhibition algorithm is probably incorrect for future climates based on current understanding, so we will update that as well.

[ekluzek]

@lkemmons thanks for that update. We appreciate it!

[wwieder]

agreed, thanks for the update @lkemmons

[adrifoster]

hi @lkemmons and @HuiWangWanderInGitHub do you have any movement on moving to MEGAN3 for CSTM?

Additionally, I was looking at this paper, and was wondering if this new pollen module is in MEGAN3 and if this could also be implemented in CTSM? I am happy to help with moving this project forward.

[lkemmons]

Additionally, I was looking at this paper, and was wondering if this new pollen module is in MEGAN3 and if this could also be implemented in CTSM? I am happy to help with moving this project forward.

Alex Guenther replies that the pollen module you refer to was developed as an ACCESS database code and there is no FORTRAN version. Also it is only for the US.
So, it is not available to incorporate into CTSM-MEGAN.

[wwieder]

Thanks @lkemmons. A few more clarifying questions:

• Is MEGAN3 all developed as an ACCESS database, or just the pollen module? Also,
• I'm still working under the impression that MEGANv3.1 will not be available in CESM3 or do you have other plans?

[lkemmons]

Only the pollen module was created in ACCESS.

[rosiealice]

I am also interested in MEGAN3. I just wrote a link to map from FATES into the MEGAN PFT space, and we have a project that will use the BVOC coupling in NorESM. So any updates on the status of that would be e cool. (There is a nonzero chance we could help out...)

[HuiWangWanderInGitHub]

Hi everyone, sorry for the late reply. Here are some updates from our side. I have implemented several features in MEGAN 3, including expanding species and stress algorithms, as I mentioned when I started this issue. However, after discussing with Alex, we realized that some features in MEGAN may not be directly applicable to CLM due to differences in the canopy models. Specifically, MEGAN 3 uses a multi-layer canopy model, whereas CLM currently employs a big leaf canopy model. There is a multi-layer CLM, but it requires more effort and could be considered for the next step.

Additionally, the stress algorithms for conditions like ozone and high winds are quite immature and require further evaluation. Considering that, we decided to contribute only the work that has been well investigated and published (or is under review). The following scientific advancements have been included for the finalized version this time:

1. Drought impact on isoprene emission (Wang et al. 2022);
2. Temperature response of the Arctic shrubs (Wang et al., 2024);
3. Temperature response of the Arctic tundra sedges (Wang et al., under review).

I can send the finalized code before the end of this month (May 2024).

[lkemmons]

One further clarification:
Since the stand-alone MEGAN3 cannot be fully implemented in CLM, we will not refer to this implementation in CLM as "MEGAN3", but will call it MEGAN-CLM5 (or whatever the version is for CLM/CTSM when it is finalized and incorporated on the trunk.

[rosiealice]

Hi all. Thanks for the update. Just for background, what aspect of the canopy does MEGAN3 want to be multi layered? In FATES we have a quite complex canopy in terms of light interception and vertical variation of leaf properties (and, optionally, leaf age) and water stress (when the hydro option is run) but not in terms of temperature and atmospheric conditions (RH, CO2, wind speed). Would it be interesting to see if MEGAN3 might better link to FATES than CLM5? Perhaps we can pass this info as an optional boundary condition?

[wwieder]

@HuiWangWanderInGitHub, thanks for your update. Opening a work in progress PR would be helpful so we can see how involved these changes are going to be to integrate into CTSM.

@lkemmons I'm assuming this is a high priority for the CAM-CHEM group. Is it required for CESM3 (i.e. we hold up the release of CESM3 until MEGAN-CLM6 comes in), or does it fall in the "would be nice to have, but not required" category. I ask because I'm trying to figure out how to prioritize SE resources in the LMWG.

@rosiealice you mentioned that NorESM may have SE resources to help with this? Would this be for the features Hui mentioned, above, or more on the fully MEGAN3 implementation that it sounds like you're askig about.

[wwieder]

Hi All, After further discussing this PR at our SE meeting today I think it's clear that the LMWG will not be able to get this work onto main by the June 30 code freeze. I'm happy to outline what needs to happen to bring a forthcoming PR and make it available to the community, but it's a larger discussion about deadlines and priorities for the default configuration of CESM3. I think everyone else should be notified, but I also wanted to bring in @dlawrenncar to this conversation.

[lkemmons]

Just to make everyone on this discussion aware:
Hui Wang @HuiWangWanderInGitHub has submitted a pull request #2588 with updates to MEGAN.
This is not implementing MEGAN3, but is a few key updates to MEGANv2.1.

[wwieder]

For the purposes of CESM3 this issue is superseeded by PR #2588, which is a narrower modification to CLM-MEGAN. I'll close this issue for now with a won't fix. If other's feel differently, please reopen this issue.