Guidance on Running with LME datasets with CESM2.1.3 (CLM5)

[wvsi3w]

Hello,
I have some questions regarding how to run with LME data and other related inquiries that I would like to know your opinion on:

For my project, I’m using CESM 2.1.3, and I would have a combination of paleo and future scenarios in my project. I would be using data from LME (850 CE onward) and SSPs for GLOBAL and REGIONAL simulations. So it would cover 850 to 2100 (or let's say 2300, using SSP development data). I have my own soil layer structure therefore I will need to do my own spinup as the restart files from other simulations and other projects won't be useful. I am facing a few technical and strategic challenges...

1. Dataset Selection and Compatibility: I plan to use the CCSM4-based LME data (Dataset: CESM1 Last Millennium Ensemble) as atmospheric forcing for the last millennium. However, I am unclear about which specific dataset version (e.g., the “003” or “007” series). As you know in the download link of the atm 6h data there are two types of datasets, for example for the Q variable:
b.e11.BLMTRC5CN.f19_g16.007.cam.h2.Q.0850010100Z-0899123118Z.nc
b.e11.BLMTRC5CN.f19_g16.850forcing.003.cam.h2.Q.0850010100Z-0899123118Z.nc
would be optimal for this purpose and why? Also if this dataset is not compatible with CESM2.1.3 (CLM5) what do you suggest to use instead that covers 850 onward?

2- Are there recommendations on using “6-Hourly Instantaneous” vs. “6-Hourly Averaged” data for climate dynamics over this period?

3- I asked a question here in this thread regarding the process and I would like to know your thoughts on it too, what would be the steps to do a simulation using LME datasets from the beginning (setting up the case, paths, names, spinup, transient, ...), any guideline available for global simulation? I found this link for PaleoResources but I don't think it would be useful since it is for the CLM4 version.

4- Is there a different process for surface dataset generation in single-point mode for using the LME dataset?

4.1- Also (a less related question) regarding the single-point process, when people talk about "using our own forcing dataset" on the forum, they seem to simulate "single-point/regional" jobs. I wondered if it is even possible to use our forcing for global simulations or different resolutions. I mean I think it is possible but why this CLM_USRDAT_NAME method is only mentioned in the single-point simulation process?

5- What kind of vegetation dynamic would be ideal for this type of simulation? Is FATES good for it? Is there any other DV than FATES? any limitations? I am asking because one important part of my project is to have vegetation dynamics analyzed, so knowing this before running the lengthy simulation would be very beneficial and save time and energy.

6- How about coupling CLM with CAM or fully coupled CESM? can you think of a way to have any of the mentioned coupled ways using LME data in a +1000y simulation? This part is also important because some of the atmospheric variables (or circulation/pattern/teleconnection) must be analyzed too, so I will need CAM.

6.1- therefore if you had some similar experience with coupling CLM&CAM or with giving the output of CLM to CAM as initial conditions or I don’t know maybe “forcing” please let me know the general process or guidelines. Or if I am not thinking clearly and maybe this question is not correct also please notify me because I don't know if doing so "giving the output of CLM to CAM (as the initial condition)" is a way to do it. I suspect I need to have a CLM-CAM coupled run (or fully coupled) to achieve the aforementioned objectives of my project, but that's me, I need your advice.

Thanks a lot.

 

[wvsi3w]

Dear @jiangzhu I could only find your ID who I think is related to Paleo forum. I would really be grateful if you let me know what you think about these questions (specially question 1 and 2). Thanks a lot

 

[wvsi3w]

Dear Scientists,
I have a question about my LME dataset which I downloaded (9TB, 7 variables: U, V, Q, T, PS, PRECT, FLDS, FSDS):

The data was 2deg and since I needed it to be 1deg I did some re-griding (cdo remapbil,r360x180). I noticed that most people are using their own atmospheric forcing to run a single-point or regional case but my question is that can we run it globally? I think we can, but is it a different process for it? because most of the threads I read are doing it for regional studies, and I need to do it globally. If so, may I know which link or documentation I should follow?

Another question is my dataset consist of 23 netcdf files like one for 850-899 another for 900-949 and so on till 1950-2005. So I read in some threads that we need to separate the data into monthly sets or we need to reduce the number of files (forcing years) in stream file; e.g. PRECT/850-01.nc, 850-02.nc, .... 2005-12.nc
Is that right? or can I keep it like that 50 years?

It would be great to know your opinion (@oleson)

 

[wvsi3w]

Dear Scientists,
I have a question about my LME dataset which I downloaded (9TB, 7 variables: U, V, Q, T, PS, PRECT, FLDS, FSDS):

The data was 2deg and since I needed it to be 1deg I did some re-griding (cdo remapbil,r360x180). I noticed that most people are using their own atmospheric forcing to run a single-point or regional case but my question is that can we run it globally? I think we can, but is it a different process for it? because most of the threads I read are doing it for regional studies, and I need to do it globally. If so, may I know which link or documentation I should follow?

Another question is my dataset consist of 23 netcdf files like one for 850-899 another for 900-949 and so on till 1950-2005. So I read in some threads that we need to separate the data into monthly sets or we need to reduce the number of files (forcing years) in stream file; e.g. PRECT/850-01.nc, 850-02.nc, .... 2005-12.nc
Is that right? or can I keep it like that 50 years?

It would be great to know your opinion (@oleson)

Dear @slevis and dear @oleson I would be happy and greatful to know your opinion about the question I asked above. Thanks.

The data that I re-grided is more than 50TB (LME 6hourly atmosphere data for the 7 variables that are required to initialize the land model) and I have to prepare it for the model on another system (because of storage limit), then send data (100 years at a time) to the system I will run the model and run it (100 years at a time = 10 continuous 100y = 1000y). Based on the two threads I mentioned in my previous question "to separate the files" and "reduce the number of forcing years" I think I must have the data in the monthly format. I started separating the whole 50TB now. Still, I need to make sure if this is the right way to do it because the storage on my archive system is reaching its limit. After I separate the files into monthly I have to delete the initial 50y files to have enough space to move on to the next step so your opinion here will be extremely influential as it will save a lot of time for me.

I have some more questions too:

1- The Temperature data is available for this LME 6hourly (Dataset: ucar.cgd.ccsm4.cesmLME.atm.proc.6hourly_inst.T), is this the TBOT? because the TS is not TBOT and I think this T should be used (its lowest level=0) (?)

2- All the 6hourly LME data has two types,

b.e11.BLMTRC5CN.f19_g16.007.cam.h2.T.0850010100Z-0899123118Z.nc
and
b.e11.BLMTRC5CN.f19_g16.850forcing.003.cam.h2.T.0850010100Z-0899123118Z.nc

I think the difference is that the second one used the year 850 for spinup (?). I downloaded all 7 variables from the first type of data. I didn't know which type (003 or 007) to choose so I went with the first one. Did I choose right? if not I have to immediately download the other one. I actually asked it in my first question in this thread.

3- When I start the spinup and my transient run for my case using this LME data (850 onwards) what would I set for the CO2 content? because besides those 7 variables that are required to run the clm5, there are CO2 also which I guess I have to make another file for it using the constant value from 850 until 1850 (?)

4- Could you confirm that the procedure to run globally is similar to what is discussed for regional single-point simulation process that people have used with their own atmospheric forcing? Because I don't know why I couldn't find a thread or a guideline for it and seems like everyone have used "running with own atm forcing" for their regional studies !

 

[slevis]

I will respond to things that I think I can help with:
3. If I were running 850 to 1850, then i would set the co2 to the known record. For an 850 spin-up, I would use the co2 value for that year.
4. Yes, should be similar since global simulations are the most general case of regional.

 

[wvsi3w]

I will respond to things that I think I can help with:
3. If I were running 850 to 1850, then i would set the co2 to the known record. For an 850 spin-up, I would use the co2 value for that year.
4. Yes, should be similar since global simulations are the most general case of regional.

Thank you Sam for your kind response.
While I was preparing the 50TB of data (FSDS, PRECT, TPHLW) I thought of another question:
Since I can not put all of these 1000 years of data in the initial path (due to low storage on my system that I run the model) I thought I could split it into 100y or 50y runs. So after the spin-up of 850 is done, I should start the transient run, then what is the xmlchange option that I must consider? because when the AD is done we do ND with "RUN_TYPE=startup" with pointing to the AD restart file. So, for the transient I should then point to the final spunup file and go with RUN_TYPE=startup?? and when the first 100y is finished I again put the restart of the first 100y transient run in "finidat" and again go with startup type?

I am asking these because I have read somewhere that I probably need the branch type run for this. I am not sure what is correct.