About past, current, and future runs with coupled CESM for dust forcing

[ll859]

My goal is to assess dust radiative effect under conditions in 1850s, current (2010s or 2000s) and 2 future (2090s) scenarios (SSP2-4.5 and SSP5-8.5). I want to assess dust optic changes at some points. I would prefer not to have to run a full simulation, but just 10-years slices for each climate to calculate radiative forcing. So, I think would need the component sets, restart files, and base code that matches the restart files.

I found this case name: b.e21.BSSP245cmip6.f09_g17.CMIP6-SSP2-4.5.101: and it seems to have SSP2-4.5 restart files for 2085 (which is fine). But it has prognostic CO2. I don’t know neither where the model version (base code) is for this simulation, though I can do hybrid run with those restart files and the compset of BSSP245cmip6.

We would prefer to use a case with diagnostic CO2. I just can’t find one that works. There is a case for present day, b.e21.BHIST.f09_g17.CMIP6-historical.001, which is similar to the one I am looking for. But it turned out that I have to set the following in the land namelist of my historical run to do interpolation, just because of different land unit numbers.

use_init_interp = .true.

It seems that the two codes (my and the one used in the simulation that generated the restart files) are different somehow. But we don’t know which version was used for that CMIP6 case.

I did find a case, b.e21.BHIST_BPRP.f09_g17.CMIP6-esm-hist.001, with prognostic CO2, and I can create a fully coupled case and run it using CESM2.1.2 also with prognostic CO2 and those restart files.

Anyone any recommendation or comments? Many thanks!

 

[slevis]

My comment pertains to the need for use_init_interp = .true.
I may have misunderstood what you meant but wanted to make sure this was clear:
It doesn't imply that the codes are different. It means that the surface data are different.
And I think you will find that the surface data differ for each time period that you run and that you will need use_init_interp = .true. each time.

 

[ll859]

Many thanks for the comments!

I wonder then if it is safe for us to do hybrid simulations using different surface data as restart files. (and with the use_init_interp = .true.).

Or, we have to find and use the same surface data. For this, I just cannot find them on Casper. So, if there is any recommendation for whom I can contact to get the restart files for any year in or close to 2000s (CESM2.1.2; compset: BHISTcmip6), that would be very helpful!

If those restart files did not exist, I would appreciate clues on where I can find the case (b.e21.BHIST.f09_g17.CMIP6-historical.001) used to get the restart files under cmip/CMIP6/restarts/ on Casper.

 

[slevis]

To be absolutely clear, when I say surface data, I mean the fsurdat file, which cannot replace a restart (finidat) file. Due to the confusion of terms, I do not think I follow your question. Could you ask again with the fsurdat/finidat distinction in mind?

 

[slevis]

If you're looking for restart (finidat) files to avoid a continuous simulation from 1850 to 2100 and not finding all the files that you need, then it may be most consistent (and, therefore, scientifically more sound, but also quicker) to complete the continuous simulation. Otherwise you need to track down restart files that may or may not exist in someone's directory, and this someone may or may not see your post and then respond to it.

 

[ll859]

Hello! Many thanks again, as well as for your clarity!

The continuous run will work for us as well. But, to do that run, we would still need inputs from a case that used the same surface data (fsurdat). And, we also need to know where the case and the source code someone used in the reference simulation are. Right?

Since we don't need the model to restart in a bit-for-bit fashion, a hybrid run remains a good option for us. Per your clarity, the primary question becomes if it is safe for us to do a hybrid simulation using different finidat files. Hope this is clearer. Many thanks!

 

[ll859]

Hi! I found the restart files that work well with the source code and component set I am using. Thank you very much for all your comments!