Add global mean CO2 diagnostic

[spencerkclark]

Description

This PR adds a diagnostic for the global mean volume mixing ratio of CO2. It does so following the conventions / approximations in radlw_main.f:

SHiELD_physics/gsmphys/radlw_main.f

Lines 765 to 806 in 5e5e100

	!> -# Prepare atmospheric profile for use in rrtm.
	! the vertical index of internal array is from surface to top
	! --- ... molecular amounts are input or converted to volume mixing ratio
	! and later then converted to molecular amount (molec/cm2) by the
	! dry air column coldry (in molec/cm2) which is calculated from the
	! layer pressure thickness (in mb), based on the hydrostatic equation
	! --- ... and includes a correction to account for h2o in the layer.
	if (ivflip == 0) then ! input from toa to sfc
	tem1 = 100.0 * con_g
	tem2 = 1.0e-20 * 1.0e3 * con_avgd
	tz(0) = tlvl(iplon,nlp1)
	do k = 1, nlay
	k1 = nlp1 - k
	pavel(k)= plyr(iplon,k1)
	delp(k) = plvl(iplon,k1+1) - plvl(iplon,k1)
	tavel(k)= tlyr(iplon,k1)
	tz(k) = tlvl(iplon,k1)
	!> -# Set absorber amount for h2o, co2, and o3.
	!test use
	! h2ovmr(k)= max(f_zero,qlyr(iplon,k1)*amdw) ! input mass mixing ratio
	! h2ovmr(k)= max(f_zero,qlyr(iplon,k1)) ! input vol mixing ratio
	! o3vmr (k)= max(f_zero,olyr(iplon,k1)) ! input vol mixing ratio
	!ncep model use
	h2ovmr(k)= max(f_zero,qlyr(iplon,k1) &
	& *amdw/(f_one-qlyr(iplon,k1))) ! input specific humidity
	o3vmr (k)= max(f_zero,olyr(iplon,k1)*amdo3) ! input mass mixing ratio
	! --- ... tem0 is the molecular weight of moist air
	tem0 = (f_one - h2ovmr(k))*con_amd + h2ovmr(k)*con_amw
	coldry(k) = tem2*delp(k) / (tem1*tem0*(f_one+h2ovmr(k)))
	temcol(k) = 1.0e-12 * coldry(k)
	colamt(k,1) = max(f_zero, coldry(k)*h2ovmr(k)) ! h2o
	colamt(k,2) = max(temcol(k), coldry(k)*gasvmr(iplon,k1,1)) ! co2
	colamt(k,3) = max(temcol(k), coldry(k)*o3vmr(k)) ! o3
	enddo

It assumes the moles of CO2 per area can be computed by scaling the prescribed volume mixing ratio of CO2 by the moles of dry air per area. A global mean volume mixing ratio then can be computed by dividing the total moles of CO2 globally the by the total moles of dry air globally.

How Has This Been Tested?

This has been tested in a three-month C24 simulation outputting the instantaneous global mean CO2 concentration every 6 hours. The plot of the time series can be found below.

The simulation uses gfs_physics_nml.ico2 = 2. For globally uniform CO2 (e.g. with gfs_physics_nml.ico2 = 1 and gfs_physics_nml.ictm = 19970), the diagnostic behaves as expected, producing a value exactly the same as that prescribed.

Checklist:

Please check all whether they apply or not

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• Any dependent changes have been merged and published in downstream modules

[lharris4]

I think this is fine. Writing a 0-D variable to the history fields is a bit different (usually we output global integrals to stdout or to a text file) but this additional functionality is potentially useful too.

[spencerkclark]

Thanks @lharris4—indeed I was not aware of this functionality until reading the AM4 code. It is a little funky since the diagnostics manager has no choice but to write copies of the scalar fields out to all (sub)-tile files, but nevertheless it is nice to have the data all in one place (with clear / standardized metadata) instead of having to parse it out from another source.

[lharris4]

The metadata is an excellent point in favor of this method. Now I remember that we can also use diag_manager to output point soundings too and we will get all of the appropriate metadata and NetCDF niceties. There might be ways to output global-mean values to a single NetCDF file, since this is done for the restart files.

…

On Wed, Jun 5, 2024 at 2:24 PM Spencer Clark ***@***.***> wrote: Thanks @lharris4 <https://github.com/lharris4>—indeed I was not aware of this functionality until reading the AM4 code. It is a little funky since the diagnostics manager has no choice but to write copies of the scalar fields out to all (sub)-tile files, but nevertheless it is nice to have the data all in one place (with clear / standardized metadata) instead of having to parse it out from another source. — Reply to this email directly, view it on GitHub <#45 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMUQRVCLWGXXUUTHJ6ACD3LZF5JVFAVCNFSM6AAAAABI3DMZ3SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNJQGY4DQMBXGM> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[spencerkclark]

There might be ways to output global-mean values to a single NetCDF file,
since this is done for the restart files.

Indeed there is! Thanks for prompting me to think about it a bit more. The issue was that I set up my diagnostics table to include the scalar field in the same files as horizontally-resolved fields. If only scalar fields are specified to be output to a particular file then that file will not be split across (sub)-tiles.

[lharris4]

Hi, Spencer. Thanks for digging into that. This opens the doors to a lot of nice global-integral calculations that can be written to a NetCDF file instead of filling up stdout. Lucas

…

On Wed, Jun 5, 2024 at 2:37 PM Spencer Clark ***@***.***> wrote: There might be ways to output global-mean values to a single NetCDF file, since this is done for the restart files. Indeed there is! Thanks for prompting me to think about it a bit more. The issue was that I set up my diagnostics table to include the scalar field in the same files as horizontally-resolved fields. If only scalar fields are specified to be output to a particular file then that file will not be split across (sub)-tiles. — Reply to this email directly, view it on GitHub <#45 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMUQRVEBXX2C2TCFNQTGJZDZF5LIBAVCNFSM6AAAAABI3DMZ3SVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNJQG4YTCNJSGQ> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[spencerkclark]

@laurenchilutti and / or @bensonr does this look good to merge?

[spencerkclark]

Thanks @laurenchilutti!