Variable(s) to assess N limitation

[aswann]

Is there an appropriate variable or multiple variables to look at to find a change in N limitation between two experiments? I see a boost in LAI and NBP after I increase soil NO fluxes to the atmosphere in CAM6-CHEM w/BGC, and I'm trying to figure out if there is a real reason why this should happen. Thanks!

[wwieder]

Sorry, I'm not really sure what experiments you're looking at, but for a quick look I typically use FPI, which is metric of the soil biogeochemical N limitation (and easier to ncview; 1 = no limitation). Otherwise I think @rosiealice recommends looking at the ratio between NPP_NUPTAKE/(NPP_NUPTAKE + NPP), alternatively using GPP may also be appropriate? This gives a proxy N limitation by normalizing FUN autotrophic respiration fluxes as a fraction of primary production.

[lkemmons]

The input files are in molecules/cm2/s and are output in kg(NO)/m2/s, using a molecular weight of 30 g/mole.
The input files have dates (I think) of the 15th of each month, and the model interpolates between the dates in the file at each timestep in the model. So, the monthly average in the output will be shifted. I can't think of any other reasons for the difference between input and output.

[aswann]

I found an error and they are matching more closely now without an offset, but still with slight differences probably due to averaging on the different time steps.

[lkemmons]

OK. Good. I hadn't seen this correction and just checked one of my runs that indeed input and output are very similar:
Input: date, anthro, bb, soil, total
20100116 92.2 13.2 7.4 112.8
20100216 90.4 7.9 6.6 104.9
20100316 90.0 10.0 8.3 108.3
20100416 89.2 5.5 10.4 105.0
20100516 88.4 6.4 13.3 108.1
20100616 87.9 13.3 14.1 115.3
Output:
20100201 112.120
20100301 106.322
20100401 107.458
20100501 105.802
20100601 108.696
20100701 115.513
(remember that in monthly average files the date is at the end of the averaging period, so '20100701' is the June 2010 average.)

[aswann]

Although the discrepancy is smaller my input and output still aren't equivalent, and the error is ~20% of the total because the number is low. The SFNO is always larger which makes me think that averaging in time still can't explain this.

[lkemmons]

Further discussion on the mismatch of input emissions and SFNO will be continued in https://bb.cgd.ucar.edu/cesm/threads/how-is-surface-flux-output-variable-sfno-related-to-input-files.8075/.

[mvalmartin]

Hi, N deposition is coupled in the model, NHx and NOy (wet and dry deposition) can be passed from CAM to CLM, but you need to activate that in user_nl_cam as ndep_list = 'noy', 'nhx', otherwise CLM uses prescribed N deposition. @lkemmons should be able to confirm that, as N dep may be active now by default in newer released versions!

[lkemmons]

CAM-chem compsets in CESM2 should by default include nitrogen deposition from CAM to CLM.

[aswann]

I can also confirm that it looks like there is a response in FPI when I modify soil flux of NO, so I'm assuming this extra NO is causing extra deposition which is alleviating N limitation.

[rgknox]

@aswann , I'm not so up to date with the plant uptake diagnostics on the big-leaf CLM side (since I work with FATES), but, I also use fraction of potential immobilzation (FPI) as one of my metrics to get a sense of N limitations.

[mvertens]

@mvalmartin - once the CAM PR ESCOMP/CAM#735 is merged (it has been approved but I'm not sure why it has not been merged yet) - then CAM will always be passing NDEP to CLM. However, CLM in turn will need to be modified to always listen to the NDEP being passed in.

[mvalmartin]

Thanks for the clarification. So, this means that CLM is not using N deposition from CAM, as it is. Am I correct? I have passed N dep from CAM to CLM in some of my runs, but i always had to tell CLM to use it. Sorry if my note is confusing!

[mvertens]

@mvalmartin - sorry - I think my message was confusing. Currently CAM sends NDEP to CAM if it computes it prognostically. This is done via the settings in the drv_flds_in file that CAM produces and that CAM reads. If CAM does not produce the NDEP fields prognostically - then CLM is able to recognize this from the drv_flds_in file and read in NDEP itself. Once CAM always sends NDEP - then CLM can ignore the drv_flds_in namelist and always get NDEP data from CAM or DATM (which always sends NDEP right now).
The key point is that once NDEP is always sent to CLM (particularly if the atm component is CAM) then CLM needs to turn off the logic where it decides if it will read in NDEP and just use what is passed. I hope that clarifies things.