All cases use modern atmospheric CH4 even for pre-industrial or transient cases

[ekluzek]

Brief summary of bug

For I cases atmospheric concentration of CH4 comes from the paramfile which has a modern value in it. So hence it's wrong for pre-industrial and transient cases (or for that matter paleo cases).

General bug information

CTSM version you are using: applies to all versions of ctsm

Does this bug cause significantly incorrect results in the model's science? Yes

At least for methane, it needs to be evaluated how big of a problem this is.

Configurations affected:

This value is only used when NOT coupled to CAM. When coupled to CAM it uses the values that CAM uses.

Important output or errors that show the problem

bld/namelist_files> ncdump -v atmch4 $CSMDATA/lnd/clm2/paramdata/clm_params.c170913.nc | tail -2 atmch4 = 1.7e-06 ; } bld/namelist_files> ncdump -v atmch4 $CSMDATA/lnd/clm2/paramdata/clm5_params.c171117.nc | tail -2 atmch4 = 1.7e-06 ; }

This problem was pointed out by Esther Brady...

Erik and Keith,

Thanks for all your help so far with getting our paleo CMIP runs going.

Bob and I have noticed that in all the current CESM2 runs, the atmospheric methane concentration in the land is set to 1700.e-9, which seems to be over twice the value prescribed in CAM for the preindustrial simulation and more like a 1990 value.  

This seems to be prescribed in the paramfile used with variable atmch4. 

We were thinking that we should change this for the paleo runs, but if all other cases have it constant at 1700e-9, then we will leave it as is.  

Is this setting for the offline methane model not impactful in any way? It comes out in the history h0 file. 

Thanks for any insight!
Esther

[ekluzek]

Some B cases must not pass methane as seen by this comment by Esther:

Ok, but I've looked at some different cases together with Keith Lindsay and the value of pCH4/pbot in the h0 land history file is 1700e-9 in all of them. The BHIST case I'm looking at right now also shows this: /gpfs/fs1/scratch/cmip6/archive/b.e21.BHIST.f09_g17.CMIP6-historical.006/lnd/hist

[ekluzek]

OK, looking at the code, another issue is that the namelist ch4offline needs to be set to .False. in order to use the value of atmospheric CH4 sent from the coupler.

[ekluzek]

In also looking in cime and CAM I didn't see where the setting of index_x2l_Sa_methane was being done. So this may also need work in CAM and cime as well.

[ekluzek]

@ckoven can you comment on how big of an issue this is for simulations? Thanks.

[ekluzek]

Charlie Koven had this to say about impacts on Jan 4th..

It seems to me like the only impact should be on the rates of methanotrophy. By assuming a given CH4 concentration, the CH4 diffusion into/out of the soil will have the wrong specified upper boundary condition, which will then control the CH4 concentrations in the soil and set the rate of methanotrophy. But that shouldn’t have any other downstream impacts to the biogeochemistry. Otherwise I don’t think this should impact any answers, since the CH4 has no radiative effect for I-compset cases.

Charlie

And he followed up regarding B compsets with..

In those cases, the effect will still just be that the gas concentrations in the soil column are inconsistent with the radiative forcing of CH4. So not ideal, but should only affect the CH4 fluxes and not propagate to anything else in the model. As far as I know, there has never been a fully prognostic CH4 cycle in CESM such that the CLM CH4 fluxes actually fed into the atmospheric model or anything else. We had talked about doing this ages ago, but if I remember correctly there are real science issues that needed to be tackled: separating out the wetland, fire, termite, ruminant, etc CH4 sources in the standard surface CH4 flux files, and then replacing just the wetland fluxes with those predicted by CLM, and then demonstrating that the model gives atmospheric CH4 gradients that are consistent with what’s observed.

Charlie

And Dave Lawerence agreed with...

I agree with Charlie, this only affects the diagnostic land methane fluxes and will not propagate to anything else in the model. In most CAM runs, CH4 is prescribed in the atmosphere from an input file rather than being simulated prognostically so there are no feedbacks from land. Like Charlie says, there are a lot of things that would need to be done for a full interactive CH4 simulation. It’s not currently a priority of anyone in CESM as far as I know.