Read and use zonally/monthly varying CO2 concentration

[gold2718]

NorESM/CAM reads in lower boundary conditions for CO2 and other GHG from a file. These files often contain seasonally, yearly, and zonally varying concentrations, however, CAM removes the zonal variation by passing a global mean value to both the radiation code and the surface components.

It is desired for NorESM3 to preserve any zonal variations and send those values to the radiation and surface.

Some questions:

• Does this feature need to be configured or should the global mean code always be skipped?
• Does this apply only to CO2 or to all GHG (or does it need to be configured per-species)?
• For concentration-driven runs, the global mean is computed on the surface values (i.e., those read from the LBC file) and that mean value is used to fill a 3-D CO2 (or other GHG) pbuf field. If we maintain the spatial variance of the LBC value, do we just fill each column with the surface value?

Thoughts? @tjiputra, @JorgSchwinger, @oyvindseland, @DirkOlivie

[gold2718]

See NorESMhub/NorESM#830

[tjiputra]

@gold2718
For 1st Q: I suggest the global mean code always be skipped. If users want to use global mean CO2 for radiation, they should supply corresponding global mean LBC files.

For 3rd Q: I would think so for now, not sure what other practical options do we have.

[oyvindseland]

Have you verified that the radiation in CAM actually use one global values for CH4 and N2O. I thought that it calculated a vertical profile for those gases (but no horizontal variation)

[gold2718]

Have you verified that the radiation in CAM actually use one global values for CH4 and N2O. I thought that it calculated a vertical profile for those gases (but no horizontal variation)

Sorry, sloppy writing again.

• Radiation in CAM computes vertical profiles for: CH4, 'N2O, 'CFC11, and CFC12.
• Radiation in CAM uses a single value for: CO2 and O2.

[DirkOlivie]

@oyvindseland I have attached a figure of the CO2 seen by the radiation (m_CO2_c variable). This is from one year (2000) in a standard nhist simulation (concentration-driven and no explicit tracers for the GHGs).

Legend for the figure :

• every color is a different month (of the year 2000)
• regions : NH (solid line), tropoics (dotted line) and SH (dashed line)

I would conclude that :

• the radiation code sees different CO2 concentrations for different regions below 15 km. However, above 15 km the concentration is globally uniform.
• the variation over time is taken into account

[oyvindseland]

So it actually do look like it takes into account horizontal and intra-annual variation And these values are the same as in the ghg concentration files?

If so it is certainly good if it is fixed that the ocean and land sea the CO2 value used for radiation

[gold2718]

@DirkOlivie, Your plot is labeled as a mixing ratio (mol mol-1). However, in rad_constituents.F90, this field is documented as kg/m^2 and in the outfld call, the formula, mmr(:ncol,:) * state%pdeldry(:ncol,:) * rga is used which seems consistent. Would that explain these curves? They do look a bit like the CAM vertical profile.

[DirkOlivie]

@gold2718 for plotting I have indeed converted the field (which is originall indeed in kg/m2) into vmr (volume mixing ration) by using the molar mass of air and CO2, and by using the mass of air per layer (AIRMASSL or AIRMASS).

[DirkOlivie]

@gold2718 I'll try to redo the plotting taking better into account dry and wet air mixing ratios.

[DirkOlivie]

@gold2718 @oyvindseland I have now corrected for the wet/dry airmass (see figure below). The CO2 profile seems now to be constant as a function of location (no difference between NH, tropical and SH profile), and constant as a function of height.

Steve, thanks for pointing this out.