fco2nat variable (IFS grid) requires fgco2 (NEMO grid)

[etiennesky]

The variable fco2nat is defined as "This is what the atmosphere sees (on its own grid). This field should be equivalent to the combined natural fluxes of carbon that account for natural exchanges between the atmosphere and land (nep) or ocean (fgco2) reservoirs."

Right now, fco2nat is output by LPJG in the IFS grid, and we have Omon:fgco2 which are the fluxes from PISCES in the NEMO grid.

fco2nat needs to be computed from fco2nat from LPJG and fgco2 from NEMO.

Is it possible to horizontally interpolate data from the NEMO grid to the IFS grid, and to add it to an existing variable?

[goord]

Yes this should be possible, but it will require some work because right now ifs2cmor assumes all input data is in spectral and gridpoint files in the same leg directory etc.

Could you explain where the variables are supposed to be found and, maybe you have an example run output we can test with?

[etiennesky]

Hi @goord

we have 2 variables which are already cmorized

1. fco2nat is output by LPJG and cmorized by lpjg2cmor.py in the Amon table with the IFS grid

./output/lpjg/001/fco2nat_monthly.out.gz
./cmor_outputs/cmor_1/CMIP/EC-Earth-Consortium/EC-Earth3-CC/piControl/r1i1p1f1/Amon/fco2nat/gr/v20200326/fco2nat_Amon_EC-Earth3-CC_piControl_r1i1p1f1_gr_185001-185012.nc

2. fgco2 is output by NEMO (*pisces_grid_T_2D.nc file) and cmorized in the Omon table

./nemo/001/a2qd_1m_18500101_18501231_pisces_grid_T_2D.nc
./cmor_outputs/cmor_1/CMIP/EC-Earth-Consortium/EC-Earth3-CC/piControl/r1i1p1f1/Omon/fgco2/gn/v20200326/fgco2_Omon_EC-Earth3-CC_piControl_r1i1p1f1_gn_185001-185012.nc

3. final result of fco2nat in Amon should be the sum of fco2nat (cmorized by LPJG) and fgco2 (cmorized by NEMO, regridded to IFS grid).

I would assume it might be easier to do this as a post-processing job, once the nemo and lpjg cmorization have been done.

You can find the raw and cmorized output here

ftp://bscesftp.bsc.es/etourign/ec-earth/cmor_issue_613/

[etiennesky]

on second thought, it might be better to change lpjg2cmor.py to add fgco2 from NEMO to the result of fco2nat, instead of adding a post-processing step.

Is there a way that ece2cmor3 knows if the experiment was run with nemo ? Because if lpjg2cmor.py runs runs before nemo2cmor.py has run it should fail, unless there is no nemo output in that experiment.

[etiennesky]

Hi @goord I am looking into a solution for this, which involves modifying the function create_lpjg_netcdf to add the values from a remapped Omon/fgco2 file. Could you point me how to find the appropriate path to the cmorized file Omon/fgco2 file?

[goord]

Hi @etiennesky you should search recursively in the output directory no? That directory you get via

cmor.get_cur_dataset_attribute("outpath")

[etiennesky]

thanks @goord , that will give the root of the output -

in order to get the complete path (e.g. $outpath/CMIP/EC-Earth-Consortium/EC-Earth3-CC/piControl/r3i1p1f1/Omon/fgco2/gn/v20200323/) I need to reconstruct it from the other attributes as well, any hints on getting that?

[nierad]

Wouldn't it actually be useful if it would also be written out by IFS?

[etiennesky]

it is written out by LPJG, IFS/TM5 do not know the difference between the natural and anthropogenic unfortunately, unless we make the distinction

[nierad]

That was my thinking. Is it costly to add extra fields to be exchanged via OASIS such that TM5/IFS have all the info needed to appropriately write it out?

[etiennesky]

I am stuck in the remapping process, all cdo conservative remapping operators fail when generating the remap fields. Will try to generate the remap fields offline using oasis.

[goord]

Etienne is ifs going to output it online on the reduced grid, so lpjg sends it to ifs and ifs adds it to pextra?

[etiennesky]

IFS does't have that flux... TM5 does, and then it sends daily co2 mixing ratio (not fluxes) to IFS, @nierad wrote that by mistake I think...

and we want this in the concentration-based runs where there is no co2 exchange between any components.

[etiennesky]

Trying with Thomas Arsouze to generate the remap file with oasis, that is also not working now, seems that cdo does not recognize the remap file we generate. They have to have the same source grid AND mask.

[goord]

So one can discuss whether those fluxes are 'what the atmosphere sees', since tm5 uses them to prescribe concentrations for IFS? Could we get away with outputting them on the tm5 grid?

[etiennesky]

implemented and will soon submit a MR for this, the output is now on the ifs grid and the nemo output is remapped wth cdo then added to the lpjg output.

[etiennesky]

code is here: etiennesky@e58901b

etiennesky@08163fb

[etiennesky]

see the MR

[etiennesky]

I have uploaded test data in

ftp://bscesftp.bsc.es/etourign/ec-earth/cmor_issue_613/

[etiennesky]

MR updated and ready for review

[treerink]

The fco2nat branch is merged into the master branch.

[etiennesky]

Hi @treerink @goord @pabretonniere @aeramos

Note I added a comment here about this issue #635 (review)

I am having a minor issue with this feature, when nemo/pisces is not activated (this happens in AMIP runs or offline land surface runs) the code fails because the nemo output variable fgco2 is not found. I added this as a safeguard to be sure that we always get the correct output, I think due to someone's suggestions last year.

However, I would like to be able to override this when I am certain that no nemo output exists.

I there any way this can be detected inside ece2cmor3 or passing a flag to lpjg2cmor.py ?

[goord]

Hi @etiennesky this should be implemented with an extra flag to lpjg2cmor, because once the separate components are running, they have no way of knowing whether the run had nemo output...