Prescribed aerosols in GeoMIP CMIP7

[dan-visioni]

This is the start of a tread for prescribing aerosols in CMIP7 GeoMIP.
What worked, what didn't, how to best get it done.

[dan-visioni]

Two models for now (IITM-ESM, Swapna and her team; and CanESM; Jason and Victoria) have expressed interest in running the new GeoMIP simulations with prescribed aerosols.
I am including here Walker from NCAR, who ran the G6-1.5K-SAI runs and will run the G7 version, Simone from NCAR (who ran the CCMI-2022 prescribed SAI simulations) and Andrin and Timofei (who worked on the REMAP algorithm that was used to prescribe the aerosols, https://gmd.copernicus.org/articles/18/6023/2025/ ) and Alistair from Reflective, who can help with the technical side as well.

Jason has already provided the details for the wavelength they would need (included at the bottom of the email).
What we need to figure out:

• do we have all the information for the G6-1.5K-SAI CESM2 simulations that would require to run REMAP for CanESM and IITM-ESM, or do we need to use the CCMI-2022 simulations Simone did (and that are functionally the same)?
• do we have it for the HiLLA/termination runs Walker also did? For termination, I guess one could just scale down the AOD over 2 years and then set it to 0.
• what is required to run REMAP (we can provide some technical help through Reflective)?

[jc-cccma]

Thanks for working on this. For CanESM we are using prescribed aerosols for our CMIP7 historical and scenario simulations based on the dataset described here: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-4990/ with technical details here: https://docs.google.com/document/d/1blX5kv0We1BteqWzMKs0OuhazAcAonay/edit.

May I suggest that it would be worth looking at their approach to provide the optical properties? Unlike CMIP6, which computed band mean optics for each model, for CMIP7 they instead provide the optics at a set of wavelengths along with scripts to generate optics for any radiative transfer model. If the CMIP7 approach would also work for the dataset being generated for GeoMIP, then it would have some benefits, including that the optics only need to be generated once and groups doing CMIP7 DECK should already be set up to use this approach.

It would also provide longevity for the prescribed aerosol optics. For example, at this point if we changed the band structure for the radiative transfer model in CanESM, I can't easily update the CMIP6 stratospheric aerosols since they were generated by a different group who may not be available.

[WalkerRaymondLee]

Hi, Dan - both the G6-1.5K-SAI and G6-1.5K-HiLLA scenarios have 5-day-average aerosol/forcing data saved. This set of variables was set up by Simone in order to allow for WACCM aerosol forcing to be prescribed in CAM runs. I don't know if it would be sufficient for REMAP, but I can talk to Simone to see if this is the same set of data saved for the CCMI runs.

I didn't run the termination scenario; that is being run by Claire Zarakas at Carbon Plan.

[dan-visioni]

Claire used the same namelist you did for the SAI case, so that's fine. And I know where her folder is on Derecho. If you can check in with Simone, that would be great, thanks!

Hi, Dan - both the G6-1.5K-SAI and G6-1.5K-HiLLA scenarios have 5-day-average aerosol/forcing data saved. This set of variables was set up by Simone in order to allow for WACCM aerosol forcing to be prescribed in CAM runs. I don't know if it would be sufficient for REMAP, but I can talk to Simone to see if this is the same set of data saved for the CCMI runs.

I didn't run the termination scenario; that is being run by Claire Zarakas at Carbon Plan.

[andrinjoerimann]

Hi Daniele

From the REMAP side, this is what was required from WACCM to produce the forcing for the CCMI-2022 models:

aerosol mode wet diameter (dgnumwet[1-3])
aerosol number density (num_a[1-3])
relative humidity
temperature
surface pressure
The procedure does include making assumptions about the geometric standard deviation of the three modes of CESM2 (we used σ1=1.6, σ2=1.6, σ3=1.2), so if anything significant to that changed in the modal aerosol module since circa 2021, we would need to respect that. Also, last time we worked with monthly mean values, but the temporal resolution could be arbitrarily changed, if needed.

Then - as was already provided in the e-mail - the output radiation band limits of the target models are obviously required, as well.

To comment on a CMIP7-type procedure: we could also produce the optical properties on many, not too widely spaced bands, and if the CMIP7 interpolation script allows, these could be processed for each model. But in principle REMAP is publicly available and (after I check again that everything works for this use case) it can be run just by inputting the band limits, once the CESM2 base data is added to the directory as well.

[jc-cccma]

If the REMAP code is publicly available then generating the optics using the band limits would work well. There is some smallish differences interpolating from optics at a set of wavelengths to band means. Plus in my example above for the CMIP6 stratospheric aerosols, the code to generate the optics was not publicly available and the optics needed to be generated on request.

[alistairduffey]

Hi all, just checked and those required outputs are available in CESM for all three scenarios (G6-1.5K-SAI, -HiLLA, and -End) with at least monthly resolution (thanks to Walker and Claire!) so it looks like we are good to be able to provide fields for all three.

Others will know better, but as far as I'm aware (and based on table 2 in walker's recent paper those σ values are unchanged.

Could you point me towards the public version of REMAP Andrin? I'd be happy to try running over lots of bands and applying the CMIP7 intrpolation approach if helpful