CTSM5.3 SSP370 16pft landuse timeseries not available for 2deg (needed for PPE)

[olyson]

Brief summary of bug

Using ctsm5.3.055:

./create_newcase --case Clm60Bgc_ctsm53055_2deg_CRUJRA2024_SSP370 --compset SSP370_DATM%CRUJRA2024_CLM60%BGC_SICE_SOCN_SROF_SGLC_SWAV --res f19_g17 --project P93300041 --run-unsupported

Then ./case.setup and ./preview_namelists yields:

ERROR : CLM build-namelist::CLMBuildNamelist::add_default() : No default value found for flanduse_timeseries.
            Are defaults provided for this resolution and land mask?

The PPE historical (/glade/u/home/linnia/ctsm6_ppe/gen_ensembles/jobscripts/user_mods/user_nl_clm_transient_1850) uses:

fsurdat = '/glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/surfdata_esmf/ctsm5.3.0/surfdata_1.9x2.5_hist_1850_16pfts_c240926.nc' flanduse_timeseries = '/glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/surfdata_esmf/ctsm5.3.0/landuse.timeseries_1.9x2.5_hist_1850-2023_16pfts_c240926.nc'

The only available SSP370 landuse timeseries datasets at /glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/surfdata_esmf/ctsm5.3.0 are 1deg (and are 78pft, not 16pft).

General bug information

CTSM version you are using: ctsm5.3.055

Does this bug cause significantly incorrect results in the model's science? No

Configurations affected: ISSP370

Definition of Done:

• [x] Get a dataset for ctsm5.3 for PPE
• [ ] Make sure this is right for ctsm5.4

[ekluzek]

@olyson this is a task that we'll do as part of ctsm5.4 datasets.

So I've added it to the grids spreadsheet.

https://docs.google.com/spreadsheets/d/1Osq56e423CF107zhoNQ0VS7-iH_JXLF9AtCvBdXyfJ4

One question though -- does this NEED to be SSP3-7.0 or can it be SSP2-4.5? We are using SSP2-4.5 for other cases where we want simulations that run through present day (I've put it in as SSP2-4.5 for now in the spreadsheet, until I hear back). Also what is the reason for running SSP's here? Mainly I'm wondering if we need other SSP's for f19 with 16-pfts?

[ekluzek]

Oh, the other question is if you just need a ctsm5.3 dataset that we create now so you can do work in current tags? We could do that. But, I also want to make sure we get what's needed into ctsm5.4 for longer term.

[olyson]

Right, PPE needs a ctsm5.3 dataset that works in current tags (ctsm5.3.055).

[wwieder]

What resolution is needed for PPE activities @linniahawkins, can you advise?

[wwieder]

@lawrencepj1 would it be realistic or appropriate to create SSP extensions from the CMIP7 historical data (5.4 datasets) that we can use until the scenarios for CMIP7 are available?

[samsrabin]

@linniahawkins Do you just need ssp370, or others too? Or is there a list of priorities here?

[linniahawkins]

Hi @samsrabin Yes, just ssp370 at 2-degree with 5.3 surface dataset. We are running with tag ctsm5.3.012. Thank you!

[ekluzek]

@linniahawkins the next question is will you 2-degree 16-pft support for 5.4 datasets? Actually, I wonder if at 5.4 we could transition to just using the 78-pft datasets. I think the main thing there is that we'll need to make new sparse grid datasets. But, if we could avoid having to have both 16-pft and 78-pft datasets in 5.4 that would be really great.

[lawrencepj1]

Hi @linniahawkins @ekluzek @slevis-lmwg @wwieder @samsrabin @olyson .

We don't currently have raw or surface datasets for the SSPs generated for CTSM 5.3/5.4 on glade. This could be done very quickly if needed. Let me know and Sam Levis and I can coordinate.

Cheers Peter

[slevis-lmwg]

To clarify... We have SSP data for 5.3 here: /glade/campaign/cesm/cesmdata/cseg/inputdata/lnd/clm2/rawdata/CTSM53RawData so I can generate (in the next day or so) the file that @linniahawkins needs right now.

@lawrencepj1 from your comment do I infer correctly that the 5.4 SSP data has become available now? If so, then we would like to have raw data as soon as possible.

[wwieder]

@slevis-lmwg, can you help with this. Linnia needs a f19 land use time series with the 5.3 datasets for SSP3-7.0.

Down the road, she may need additional SSPs & resolutions but we can figure out later.

[slevis-lmwg]

I will create two versions of this file:

• ctsm53
• with fix to ag peak fire month (coming in with ctsm54)

[lawrencepj1]

Hi @slevis-lmwg Sam

Yes I meant that the ctsm54 is available but not on glade. Thanks for the correction.

Cheers Peter

[wwieder]

Oh great. When do you think your can pre-process the future scenarios can be ready for us to make all of 5.4 datasets?

[wwieder]

@lawrencepj1 I suggest we move discusson of CMIP7 datasets over to #3246. This issue is really is really about the f19, 5.3 SSP3.70 dataset needed for Linnia's PPE project.

[linniahawkins]

the next question is will you 2-degree 16-pft support for 5.4 datasets? Actually, I wonder if at 5.4 we could transition to just using the 78-pft datasets. I think the main thing there is that we'll need to make new sparse grid datasets. But, if we could avoid having to have both 16-pft and 78-pft datasets in 5.4 that would be really great.

We haven't formalized our experimental design for phase 2 of this project so I'm not sure. But I expect that I will want: 1-degree 5.4 datasets with CMIP7 "SSPs" and 78-pft datasets would work. We will revisit this down the road (Aug/Sep). Thanks!

[wwieder]

Sounds good, @linniahawkins that should be doable for phase 2, especially since the CMIP7 future scenarios are now available.

[slevis-lmwg]

I will create two versions of this file:

• ctsm53
• with fix to ag peak fire month (coming in with ctsm54)

@linniahawkins the file you asked for (the first of the two listed here) should be ready in a few hours here: /glade/work/slevis/git/b4b-dev/tools/mksurfdata_esmf/landuse.timeseries_1.9x2.5_SSP3-7.0_2015-2100_16pfts_c250612.nc

I will copy the file to the permanent /inputdata location tomorrow. Let me know if you are also interested in the second file (with the correction to ag peak fire month).

[linniahawkins]

Great thank you! @olyson is helping me set up anomaly forcing SSP cases for the PPE work. Keith let me know if I can help test this dataset; or if I should just wait for an example case from you. Thanks!

[olyson]

Thanks @slevis-lmwg . @linniahawkins , I was planning on incorporating this new dataset into my example case setup. I'll be back in the (virtual) office on Tuesday, if you can wait until then. I think we will also need the surface data for 2015 that matches the landuse timeseries file, since the years of this SSP370 landuse timeseries file that @slevis-lmwg produced are 2015-2100:

/glade/work/slevis/git/b4b-dev/tools/mksurfdata_esmf/surfdata_1.9x2.5_SSP3-7.0_2015_16pfts_c250612.nc

However, one thing I am now wondering about is what the provenance of the raw data is that went into the historical run you ran, in particular the data for 2015-2023 in the landuse timeseries file:

/glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/surfdata_esmf/ctsm5.3.0/landuse.timeseries_1.9x2.5_hist_1850-2023_16pfts_c240926.nc

An ncdump of this file yields blank strings for the input_pftdata_filename field so I'm not sure what was used as raw input data for these years (2015-2023). We would want to make sure that those years were from SSP370 for years 2015-2023.

[slevis-lmwg]

@olyson I looked at the surfdata_1.9x2.5_SSP3-7.0_2015_16pfts_c250612.log file in the same directory that I mentioned above, and you were right to be concerned. The tool defaults to using historical data from 2015 to 2023. I will change the landuse_timeseries_SSP3-7.0_2015-2100_16pfts.txt and surfdata_1.9x2.5_SSP3-7.0_2015_16pfts_c250612.namelist files and rerun.

[linniahawkins]

Great, thanks Keith! There's no rush, just whenever you have time.

I'm not sure how to track down the origin of the landuse time series we are using in the PPE, but if there is any information I can provide that would help just let me know.

[slevis-lmwg]

Other than this file that I'm generating for the PPE work, @olyson would you make this the new default? I.e. for any landuse file with SSP data from now on, should we start with the SSP data from 2015? While when we run historical, then we use the historical data out to 2023? I will not change the defaults until we discuss this further.

[wwieder]

For cmip6 data I guess it makes sense to start in 2015, but for 5.4 datasets I'm assuming that for the cmip7 scenarios will start in 2023

[slevis-lmwg]

@linniahawkins @olyson the new fsurdat/landuse files are here: /glade/work/slevis/git/b4b-dev/tools/mksurfdata_esmf/. I also copied them to the permanent /inputdata location.

[ekluzek]

For cmip6 data I guess it makes sense to start in 2015, but for 5.4 datasets I'm assuming that for the cmip7 scenarios will start in 2023

This is a good point. I'm not sure how baked into the system the 2015 date is. But I think there are likely several places that will need to be switched including CIME and CDEPS possibly. As well as in CTSM.

[wwieder]

For cmip6 data I guess it makes sense to start in 2015, but for 5.4 datasets I'm assuming that for the cmip7 scenarios will start in 2023

This is a good point. I'm not sure how baked into the system the 2015 date is. But I think there are likely several places that will need to be switched including CIME and CDEPS possibly. As well as in CTSM.

+1 @ekluzek , see also #3137.

@linniahawkins I also wonder what's going to be best for you here with the sparse grid LHC ensemble you've already run? I'm assuming the CLM6 PPE was using CRUJRA and went through 2023? Did these inherit the correct datm streams through the end of the HIST period (co2, ndep, population density, etc)?

I also wonder how we harmonize the land use time series here most appropriately for what you're wanting to do with the first PPE paper (extending the simulations through 2100 with SSP370 anomaly forcing? Simplest would be to continue your HIST case from 2024-2100 using the land use timeseries that @slevis-lmwg already made (2015-2100). This may give a discontinuity at the SSP transition, but maybe that's OK?

What did @djk2120 do for the CLM5 PPE work? I'm assuming this a GSWP3 forced run where the historical case ended in 2015, which would have been more consistent with the CMIP6 scenarios and land use timeseries?

Sorry for the rambling note before I leave town 🛩

[olyson]

@linniahawkins , @slevis-lmwg , maybe we should meet to discuss @wwieder 's questions. I looked at what I think your historical case was (/glade/work/linnia/BNF_v2.n01_ctsm5.3.012/cime/scripts/transient/basecases/BNF_v2.n01_ctsm5.3.012_transient) and it seems that CO2, ndep, etc would end up being extended or cycled for 2015-2023. I also looked at the log file for the historical surface/landuse file you used and it used TRENDY for 2015-2023, which presumably is real year data and not SSP data. So it seems like there would be a bit of a discontinuity starting your simulation at 2024 with SSP data.

[slevis-lmwg]

I'm happy to meet.

[linniahawkins]

Thanks Keith, I'm free after 2pm today. @djk2120 would you also be available this afternoon?