CESM1.2.1 run with f45_g37 grid

[hannay]

From user:

Hi Cecile,
you mentioned in this post that f47g37 "is not a supported resolution." What do you mean by "not supported"? Shall I just skip it and resort to other resolutions, or do some tuning before using it?

I am running CESM1.2.1 using this low resolution for paleoclimate simulation, and it gives some unreasonable results (e.g., a too weak AMOC in the simulation). I don't know whether I shall stick to this resolution. Any advices or suggestions will be highly appreciated.

Thank you in advance,

 

[hannay]

We are supporting compset and resolution at different levels as described: CESM2.2.0 Component Sets Definitions

Scientific validation of CESM consists of a multi-decadal model run of the given component set at the target resolution, followed by scientific review of the model output diagnostics. All scientifically supported component sets are also accompanied by diagnostic and model output data.

We don't do this for f47g37. We mainly used this resolution for testing purpose during software engineering testing. It means that if you want to use that resolution, you should check that it produces a realistic simulation.I expect that it would require tuning. It might be easier for you to move to a supported resolution.

 

[wangsz@fio_org_cn]

Hi Cecile,

thanks for your quick reply. If I am right, I can use the "supported" compsets with the "supported" resolutions out of box without further tuning? As far as I know, T31g37 is a supported resolution. But when I use this resolution to create a new case, the pop2_in file is almost the same with the one I created using the f45g37 resolution (same vetical and horizontal mixing coefficients). I guess for a G_INTERANNUAL compset, similar pop2_in files should yield similar results (e.g., the weak AMOC problem I mentioned here).

By the way, this link only shows the supported compsets and resolutions for CESM2.2.0. Where can I find the supported ones by other releases, e.g., CESM1.2.1?

Thanks again,
Shizhu

 

[cacraig]

In CESM2, we made a point of indicating levels of support as follows (from the CAM6.3 User's Guide). As far as I'm aware, this level of detail was not included in CESM1 documentation.

Scientifically supported: Specific compset/resolution pairs which have had significant, multi-year runs made and have been studied scientifically. It is important to note that resolutions which are not listed, are not scientifically supported, have not had tunings performed and should not be used for scientific studies without careful examination of the results.
Developmental support: Developmental configurations that are being evaluated. These are not fully scientifically supported in the sense of extensive tuning, testing and vetting.
Tested: One or more tests for this compset have been made using at least one resolution. Extensive scientific study has not been performed. The designation of “Tested” simply acknowledges that one or more compset/resolution pair(s) have been confirmed to run without crashing. No attempts have been made to validate the scientific quality of these runs and tunings have NOT been performed on them.
Unsupported: These compsets are setup as a “convenience” for various reasons and they are not supported for science runs. If a user decides to use one of these compsets, they must also supply the –run-unsupported flag to create_newcase. These compsets may not even compile and run successfully as they have not been tested.

As Cecile said, the f45g37 is mainly used for testing purposes in CESM2. All of this is just to say, that even though a grid is available, it may not be useful for scientific studies. Also the "Scientifically supported" is only for a compset/resolution pair. You can not change one of those two and still have the same level of "Scientific support" (i.e. being assured that the model will run fine without tuning).

Your last question about the pop2_in file is something that the ocean group will need to answer in your other posting.

 

[wangsz@fio_org_cn]

In CESM2, we made a point of indicating levels of support as follows (from the CAM6.3 User's Guide). As far as I'm aware, this level of detail was not included in CESM1 documentation.

Scientifically supported: Specific compset/resolution pairs which have had significant, multi-year runs made and have been studied scientifically. It is important to note that resolutions which are not listed, are not scientifically supported, have not had tunings performed and should not be used for scientific studies without careful examination of the results.
Developmental support: Developmental configurations that are being evaluated. These are not fully scientifically supported in the sense of extensive tuning, testing and vetting.
Tested: One or more tests for this compset have been made using at least one resolution. Extensive scientific study has not been performed. The designation of “Tested” simply acknowledges that one or more compset/resolution pair(s) have been confirmed to run without crashing. No attempts have been made to validate the scientific quality of these runs and tunings have NOT been performed on them.
Unsupported: These compsets are setup as a “convenience” for various reasons and they are not supported for science runs. If a user decides to use one of these compsets, they must also supply the –run-unsupported flag to create_newcase. These compsets may not even compile and run successfully as they have not been tested.

As Cecile said, the f45g37 is mainly used for testing purposes in CESM2. All of this is just to say, that even though a grid is available, it may not be useful for scientific studies. Also the "Scientifically supported" is only for a compset/resolution pair. You can not change one of those two and still have the same level of "Scientific support" (i.e. being assured that the model will run fine without tuning).

Your last question about the pop2_in file is something that the ocean group will need to answer in your other posting.

I went over some compset/resolution comninations. Just as you said, only "supported" pairs simulate reasonalbe results. I think I need to change my own compset/resolution pair to a supported one.

Thanks a lot :)