Get -chem none option (prescribed bulk aerosols) functional with CAM6

[andrewgettelman]

A bulletin board user has discovered that the -chem none option does not work with CAM6: https://bb.cgd.ucar.edu/cesm/threads/how-to-close-atmospheric-chemistry-in-fhist_bgc-case.5678/

This was reported by @cecilehannay.

The error indicates the appearance of pointing to some mam files (_m3 seems to be a mode), which might be the unused 'prescribed modal aerosols' which we DO NOT want to use.

A discussion thread developed on this, including @brian-eaton, @nusbaume and @fvitt

It would be nice to figure out how to get the prescribed bulk aerosols to work with -chem none for CAM6 as they do for CAM5. This might not be that bad:

As reported by @cacraigucar We have two CAM regression tests for -chem none which are run with every CAM tag we make. One is with ne5 aquaplanet CAM5 and the other is f10 aquaplanet CAM5. They only run for 9 timesteps, but they run.

It would be great if CAM6 could get to the same level.

[brian-eaton]

The error reported in the BB thread was:

01: ERROR: ADDFLD: VOLCANIC_m3 already on list

Grepping the code for where this addfld call is coming from I find:

./src/physics/cam/ndrop_bam.F90 call addfld(trim(aername(iaer))//'_m3', (/ 'lev' /), 'A', 'm-3', 'aerosol number concentration')

So it is coming from code associated with bulk aerosols, not modal. build-namelist is responsible for setting the default prescribed aerosol mode when the chemistry package is 'none', and it sets that mode to 'bulk'.

My guess is that the cam6 physics package includes some prescribed volcanic aerosols that are independent of the chemistry package and that's what's causing this conflict. @fvitt, is that a possibility?

[nusbaume]

@andrewgettelman @cecilehannay @fvitt @cacraigucar @gold2718

Hi All,

There was another CESM forum post recently (https://bb.cgd.ucar.edu/cesm/threads/hybrid-run-in-cesm2-1-3-model-run-error.5870/) on using -chem none with CAM6 physics, which apparently still doesn't work for F-compsets (I just tested it on the development branch), and which throws the same VOLCANIC_m3 error as listed in this issue above.

Is there any motivation to get -chem none working with at least a basic F-compset? If so then I'll re-open this issue. Otherwise I'll just let the poster know that -chem none is incompatible with CAM6 outside of an aquaplanet configuration.

Thanks!

[andrewgettelman]

We should have this working. It just takes a bit of work to map some of the mode 4 aerosols back to mode 3 I guess.

I thought we had this fixed, but I guess not. It's probably something Francis would be most efficient at.

On Wed, Feb 10, 2021 at 1:35 PM Jesse Nusbaumer notifications@github.com wrote:

@andrewgettelman https://github.com/andrewgettelman @cecilehannay https://github.com/cecilehannay @fvitt https://github.com/fvitt @cacraigucar https://github.com/cacraigucar @gold2718 https://github.com/gold2718

Hi All,

There was another CESM forum post recently ( https://bb.cgd.ucar.edu/cesm/threads/hybrid-run-in-cesm2-1-3-model-run-error.5870/) on using -chem none with CAM6 physics, which apparently still doesn't work for F-compsets (I just tested it on the development branch), and which throws the same VOLCANIC_m3 error as listed in this issue above.

Is there any motivation to get -chem none working with at least a basic F-compset? If so then I'll re-open this issue. Otherwise I'll just let the poster know that -chem none is incompatible with CAM6 outside of an aquaplanet configuration.

Thanks!

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ESCOMP/CAM/issues/261#issuecomment-777011550, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACJL32QXU4F5ZRF4ADDWRLTS6LUXVANCNFSM4TFOGARA .

-- Dr. Andrew Gettelman andrew@ucar.edu http://www.cgd.ucar.edu/staff/andrew/ +1.303.497.1887 National Center For Atmospheric Research 3090 Center Green Drive, Boulder, Colorado 80301, USA

[fvitt]

Please provide a compset name and other details where this is failing. Or point me to a case directory.

[nusbaume]

Hi @fvitt,

You can find my test case on Cheyenne here:

/glade/scratch/nusbaume/model_runs/cam_dev_chem_none_test

Basically I just ran an F1850 compset with the f09_f09_mg17 grid out of the box using the head of the cam_development branch, with the one change being the addition of -chem none to CAM_CONFIG_OPTS. The error is the same as above, e.g.:

ERROR: ADDFLD: VOLCANIC_m3 already on list

I can run with the debugging flags on if you would like a traceback or something similar. Just let me know!

[fvitt]

@andrewgettelman, This will run if I remove the volcanic forcings from the rad_climate setting. Should we simply turn off the volcanic forcings when the model (cam6) does not have prognostic modal aerosols?

[andrewgettelman]

If that is the easiest solution, we can do that. It would be good to see if we can somehow add volcanic aerosols to the bulk aerosols, but I'm fine with doing as you suggest: easy and it will work.

Good suggestion.

On Wed, Feb 10, 2021 at 3:49 PM fvitt notifications@github.com wrote:

@andrewgettelman https://github.com/andrewgettelman, This will run if I remove the volcanic forcings from the rad_climate setting. Should we simply turn off the volcanic forcings when the model (cam6) does not have prognostic modal aerosols?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ESCOMP/CAM/issues/261#issuecomment-777087568, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACJL32R6ASWC6PC7Q2J7PZ3S6MEPLANCNFSM4TFOGARA .

-- Dr. Andrew Gettelman andrew@ucar.edu http://www.cgd.ucar.edu/staff/andrew/ +1.303.497.1887 National Center For Atmospheric Research 3090 Center Green Drive, Boulder, Colorado 80301, USA

[fvitt]

@andrewgettelman, Are suggest that we try to use the prescribed volcanic modal aerosols as we normally do in cam6? Or try to use prescribed bulk volcanic model aerosols as we did in cam4? The easiest solution is to have no volcanic forcings.

[fvitt]

The root of the problem is that these phys props files have the aerosol name "VOLCANIC" : volc_camRRTMG_byradius_sigma1.6_mode1_c170214.nc volc_camRRTMG_byradius_sigma1.6_mode2_c170214.nc volc_camRRTMG_byradius_sigma1.2_mode3_c170214.nc

Module ndrop_bam tries to construct history field names base on the names from the phys prop files and thus tries to add the same field name 3 times to the history.

[nusbaume]

Since it sounds like a fix is coming in one form or another, I will go ahead and re-open the issue so we can more easily keep track of it.