FATES in CLM5

[wvsi3w]

Dear CTSM/CLM scientists,

I was reading the CLM5 paper and it is mentioned that FATES is not active by default which made me look into these links (FATES tutorial + user guide + technical note + initialization) to see how this FATES in turned on in CLM.

As far as I have read, FATES can be used (turned on) in CTSM (or maybe CLM-FATES) as in the initialization link describes how. like in this part it says:

use_fates_inventory_init = .true.
fates_inventory_ctrl_filename = '<full path to the control file>'

and continues saying stuff about "Control File Specification" for that.

I am really new to this thing and I don't know how to use FATES in my CESM (CLM5) for studying vegetation dynamics, so sorry if my questions seem trivial.
So, let's say I want to do a spin up for my project, it needs about 1200 years of spin up (biogeochemistry compset), and after that (after reaching equilibrium) I would run onward to the future (say 2100). The AD spin up for my case is 1000 years and ND spin up is 200 years.
1- how should I turn on FATES in this process? Is that "use_fates_inventory_init = .true." part the only trick we need to use in order to turn it on?
2- Can I turn it on later using some other restart file from for example, ND spin up (or any other restart file from transient simulation from 1901 onward)??? and most importantly, how can someone turn this FATES on?
3- in this practical tutorial of FATES there is not much to understand and the link for the running fates (walk through) doesn't work. do you have something similar and thorough for this?
4- In this FAQ page for FATES it says we can not still run it globally (I guess?), so it can be only used for single-point simulations? how?

The tutorials on how to do the global run and etc for FATES mostly need access to be granted and there is another link for 2024 tutorial that the application form is expired, so here is my last chance to ask these questions.

Thanks for your help.

 

[afoster]

Hi,

This is correct that FATES is not active by default.

FATES is used by selecting a FATES compset. (e.g., I2000Clm60Fates for an I case).

The user_nl_clm variables you reference (`use_fates_inventory_init = .true.`, `fates_inventory_ctrl_filename = '<full path to the control file>'`) are for if you want to initialize FATES with inventory data, which it doesn't seem like you want to do.

You must have FATES on from the start of your spinup, and cannot turn it on after a spinup.

Here is a FATES tutorial you can check out that may help with some of your questions.

FATES can certainly be run globally, but as the FAQ says, the global parameters are not yet fully calibrated, so it depends on what you want to do and how much parameterization you are interested in doing. At a single-point, you would still need to calibrate parameters, but it is likely going to be more tractible.

You can find out how to run a single point for CTSM/FATES here.

 

[wvsi3w]

Hi,

This is correct that FATES is not active by default.

FATES is used by selecting a FATES compset. (e.g., I2000Clm60Fates for an I case).

The user_nl_clm variables you reference (`use_fates_inventory_init = .true.`, `fates_inventory_ctrl_filename = '<full path to the control file>'`) are for if you want to initialize FATES with inventory data, which it doesn't seem like you want to do.

You must have FATES on from the start of your spinup, and cannot turn it on after a spinup.

Here is a FATES tutorial you can check out that may help with some of your questions.

FATES can certainly be run globally, but as the FAQ says, the global parameters are not yet fully calibrated, so it depends on what you want to do and how much parameterization you are interested in doing. At a single-point, you would still need to calibrate parameters, but it is likely going to be more tractible.

You can find out how to run a single point for CTSM/FATES here.

Dear Adrianna,
Thank you for your response.
I was reading the forum and found this thread "CLM5 DATM_MODE for point simulation" where Keith mentions that FATES should be used as DV is not available for CLM5 and one thing came to my mind that if you know it would be possible to find out how to run with FATES/CESM2.1.3 because I can not currently switch to CTSM and I am stuck with cesm2.1.3 (due to system difficulties and error in configuring the model on them + objective issues of my project).

I think it should be a way since clm5 is part of cesm2.1.3 and when FATES is our only option for vegetation dynamics then there should be some tutorial on it, right?

 

[slevis]

I don't know if @afoster has tried running FATES from the full CESM, and I definitely have not. But two things come to mind:
1) I think you should be able to run an "I" case from the full CESM. If so, then I expect that you could make that "I" case a FATES case the way you would in the CTSM.
2) If you're thinking about running in coupled mode, then I would recommend confirming first that (1) works. If it works, then you likely need to come up with a coupled-case compset using the "long name" capability, so that you may include FATES in the case.

 

[afoster]

Yes you can run FATES in CLM with CESM the same way you would run it as a standalone module.

 

[wvsi3w]

Thank you @slevis and @afoster for your responses,

I have to mention something:
The main objective of my project is to give enough time (like 1000y) to the model to capture the surface temperature signals that propagates downward so that we then check the impacts of that on permafrost region + vegetation dynamics + atmosphere circulation as a result of that thawing (and changes that happened to the land-veggies) using my modified soil layer structure that is deep enough for permafrost and land simulations.

1- I have read that FATES is our only chance to capture the vegetation dynamics in CLM5 (or maybe I am wrong and there is still hope with other DV available in CLM5-CESM2.1.3), so I must start my case with the compsets that include FATES in them (?):
I suppose I must use only long names like this one "2000_DATM%GSWP3v1_CLM50%FATES_SICE_SOCN_MOSART_CISM2%NOEVOLVE_SWAV" to create the case, however in the short name there is also FATES written so I guess there is no need to only use long name?

2- The case I used to start the AD runs is I1850 (currently running) and the cases with FATES in them are I2000, so may I ask if there are any huge differences between them? the forcing they used is the same "GSWP3", but I guess the simulation for I2000 starts from the year 2000 and if someone wants to do transient simulation for the last 150 years or 1000 years they can't use FATES, am I right?

3- Also, as you both mentioned, what would be the compset for creating the case for fully coupled simulation which includes FATES? and if this fully coupled is again using recent data and can we use different time periods for our transient simulation? and I noticed that Sam mentioned FATES wont allow AD and cant be used for Transient data, so this makes me worried!

4- And as Adrianna also mentioned FATES is best used in regional simulations, so which one should I pick and follow, fully coupled?or doing several cases with different compsets that one of them is for FATES only? sorry I am a bit puzzled in mind there are a lot of questions regarding this.
I also asked this in this threat that Keith forwarded to paleo forum:

Running CLM5 with your atmosphere forcing

Hello everyone, I have a question that might sound trivial: I have looked into these links link 1 , link 2 , link 3 , link 4 , and I couldn't decide what variables we need to give the CLM5 to run it with our data. I was thinking of using the last millennial data from another institution, which...

bb.cgd.ucar.edu

Sorry for long questions.
I am a bit worried,
Thanks for your time and support.

 

[afoster]

1. You can use either the long name, or the short name/alias if there is one. Information about choosing a compset can bet found here.
2. The difference between 1850 and 2000 would be the years of climate run, as well as the years used for the other datm streams (population, n deposition, etc.). This is more of a scientific question of which one you would like to you. You can still use FATES with the I1850 compsets, there may not be a "short name" for them, but you can definitely use one by choosing a long name.
3. Perhaps @slevis can help you with this question, I have not tried this before.
4. I'm not sure I understand your question. FATES can definitely be run globally/fully coupled, we are just still working on calibration so you may need to modify parameters.

 

[wvsi3w]

1. You can use either the long name, or the short name/alias if there is one. Information about choosing a compset can bet found here.
2. The difference between 1850 and 2000 would be the years of climate run, as well as the years used for the other datm streams (population, n deposition, etc.). This is more of a scientific question of which one you would like to you. You can still use FATES with the I1850 compsets, there may not be a "short name" for them, but you can definitely use one by choosing a long name.
3. Perhaps @slevis can help you with this question, I have not tried this before.
4. I'm not sure I understand your question. FATES can definitely be run globally/fully coupled, we are just still working on calibration so you may need to modify parameters.

Thanks for the answers.
Good to know that we can also use FATES with long name using I1850s.

About the last question, so right now I am waiting for my I1850 case which is using another soil layer structure (modified, not the default) to finish its spinup (AD and ND) then I will do the transient run for control (unmodified soil layer) and for modified version from 1901 to 2100 (or maybe 2300 using ssp developed data).

This is the regular path everyone took and this that I explained is the first part that I am going to do. The main part is to do it from year 1000 till 2100 (or 2300). The data that is available could be NorESM data and bunch of other data that people in paleo studies used and the paloe thread I sent is related to that, in which Keith mentioned "CPLHIST" as a possible method.

As I explained, the reason I need to do that 1000y transient is to capture the effects of T and etc on land for a deep version of the land model.
This part made me wonder if there is a way to run with I1850 cases (or I2000 cases) and start the run from year 1000 (and not 1901). I think there is no way, and no matter what you set as starting date, the forcing is there and starts from 1901.

So if I am concluding correctly then I would need to get access to already spunup restart files from paleo thread (or authors who did this before) and start my 1000y transient simulation from that using their data and their method because no compset is giving me the access to last 1000y data other than those paleo CPLHIST methods. However, I doubt it that they used the soil layer I chose so I might need to do spinup for that CPLHIST method too. I am not familiar with CPLHIST and I am referring to it to convey my thoughts here.

So even with that, the FATES is not accessible I suppose (?). I hope you know a way to study the DV in CESM2.1.3 when we run a transient run from e.g. year 1300-2300. Or maybe FATES is not the only option and I am not thinking clearly and there might be another way to have vegetation dynamics in my case?

About the 3rd question, do you agree with Sam when he mentions FATES and transient not doable? if so, then the whole transient run to see the DV using FATES in any case is not achievable. Or maybe I also misunderstood that part?

 

[slevis]

Regarding question 3:

@wvsi3w in the post that you mentioned, I said that FATES could not use the transient land-use data in clm5. The model can still use changing co2 and other input timeseries. This means that you can simulate the vegetation dynamics of unmanaged ecosystems as the climate and co2 change over time.

I will not suggest a specific compset. You will come up with the compset best suited for your work. To come up with your own compset, you may find it helpful to look at the list of existing compsets by typing "./query_config --compsets" in /cime/scripts.

 

[wvsi3w]

Regarding question 3:

@wvsi3w in the post that you mentioned, I said that FATES could not use the transient land-use data in clm5. The model can still use changing co2 and other input timeseries. This means that you can simulate the vegetation dynamics of unmanaged ecosystems as the climate and co2 change over time.

I will not suggest a specific compset. You will come up with the compset best suited for your work. To come up with your own compset, you may find it helpful to look at the list of existing compsets by typing "./query_config --compsets" in /cime/scripts.

Oh, thanks a lot for the clarification.
That's a relief. I misunderstood that part.

Great, I will take a look at cime path for it.
Do you have any suggestions for this question:
"is there a way to study the DV in CESM2.1.3 when we run a transient run from e.g. 1000 years like from year 1000 onward (longer than existing compset durations that mostly start from 1850 onward)"

 

[slevis]

About performing millennium-length simulations, I recommend contacting the Paleo group (as far as I know, they have a Forum on this site), because they have done (and probably published) such simulations over the years.

 

[wvsi3w]

Yes you can run FATES in CLM with CESM the same way you would run it as a standalone module.

Dear Adrianna,

I checked the fully coupled compsets and I guess if I want to use FATES with them I should for example change this:
1850_CAM60_CLM50%BGC-CROP_CICE_POP2%ECO%ABIO-DIC_MOSART_CISM2%NOEVOLVE_WW3_BGC%BDRD

into this:
1850_CAM60_CLM50%FATES-CROP_CICE_POP2%ECO%ABIO-DIC_MOSART_CISM2%NOEVOLVE_WW3_BGC%BDRD

is that correct?

 

[slevis]

Looks reasonable to me. It does not hurt anything to give it a try. Trial and error is the name of the game in numerical modeling.

 

[wvsi3w]

Looks reasonable to me. It does not hurt anything to give it a try. Trial and error is the name of the game in numerical modeling.

Thank you for your response.
Yes, I agree. But the reason I asked is that I can not test it currently it would slow down my current AD spinup as my allocation on the system is low, and I wanted to check if that long name change is the way to do it.
Thanks again for your support.

 

[slevis]

See another thread about this here:

CAM+CLM-FATES - AMIP type run?

Dear Forum Member, I am new user with the latest version of CESM code, In the past, I have used CESM1.2 for running the CAM5 using F compset for AMIP type runs. Now, I am planning to use the CESM2.1.5 or CESM2.2.2 to run the CAM6 standalone (using F compset) by prescribing the SST and Sea-ice...

bb.cgd.ucar.edu