Nudging in CAM6 in CESM2.0.1

[Ronald Li]

From the CAM6 user's guide, there is "nudging" under "Physics modifications via the namelist". However, I can't find the directory components/cam/tools/nudging/ which is mentioned in the user guide. I am using cesm2.0.1. The page also says "The examples below focus on some specific modifications that would be included in user_nl_cam." Does the "would be" mean "nudging" is not yet available?

I also searched the CAM namelist definitions, for different versions of CESM2. I notice that the namelist group "nudging_nl" appears in cesm2.1.2, 2.1.1 and 2.1.0 , but not in 2.0.1 or 2.0.0 . However, when I asked other users who are using 2.1.0 and 2.1.1 , they both couldn't find the aforementioned nudging directory.

I would appreciate any guidance, thank you!

 

[cacraig]

Sorry about the confusion. The nudging code itself exists in the CESM2.1 release starting with CESM2.1.0. What missed the delivery deadline was the tools/nudging directory which is described in the User's Guide and which will be in the upcoming CESM2.2 release.

The tools/nudging directory is in a tar file that can be downloaded via ftp. In any web browser go to: ftp://ftp.cgd.ucar.edu/pub/patc
Download the NudgingTools.tar file, untar the contents and put the files into components/cam/tools/nudging in your CESM2.1.0 or later checkout.

Let me know if you have any issues with this process.

 

[hhzhang]

Hi Cheryl,

I am also in need of the nudging directory (for my cesm2.1.3). I followed the instructions above and went to ftp://ftp.cgd.ucar.edu/pub/patc, but didn't find anything there. I guess the NudgingTools.tar file must have been moved somewhere else. Can you provide guidance on where to download the tools/nudging directory? thanks.

honghai

 

[cacraig]

The most reliable way to use nudging is to use the CESM2.2 release. Nudging along with the accompanying documentation is in that release. The ftp download was just a temporary fix until the CESM2.2 release and the ftp site no longer has the code.

 

[hhzhang]

Thanks for the prompt response!
I've done some experiments with CESM2.1.3, and will want to keep using the same CESM version and do nudging experiments to compare with what I already have. To this end, can I just copy the tools/nudging directory from the CESM2.2.0 release to my CESM2.1.3 directory? What kind of risks would this approach have?

 

[patc]

Everything should work fine for you. The tool/nudging directory did not make the cut off date at the time, otherwise
it would be in the CESM2.1.3 version. So go ahead and use the version from the CESM2.2.0 release. There were
no code changes. For generating the nudging data, if you have not already worked out the mods you need for the
scripts, it would be a good idea to look at the IPT tools for the MUSICA GitHub repo. You can look in GitHub and also
check the MUSICA webpage for useful information. The scripts there have been cleaned up and the process of
creating the nudging data from reanalysis is better documented.

--> Patrick

 

[hhzhang]

Hi Patrick,

Thanks for the answer and pointing to the additional GitHub docs. After going through the READMEs and GitHub docs, I still have some questions about how to prepare the required input files for my nudging experiment. The nudging experiment I want to do should be easier than the examples provided, because I want to nudge CAM6 (CESM2.1.3) towards a control run that's conducted with the same compset (not to any reanalysis products). In this case, there will be no need for regridding (as my nudging experiment and control run share the same resolution and topography). Is there an example for this kind of nudging, or suggestions on how I should make use of the examples provided for nudging towards reanalysis products?

In addition, I want my nudging run to be as close to my control run as possible (in the nudging domain). To achieve this, I think I will want my control run to output [U, V, T, Q] target fields ideally at model time step and on model levels (??). But saving outputs at this high resolution is probably not practical or even necessary. Any recommendations on the resolution of the control run [U, V, T, Q] target fields? My experiments will be 10 years long on fv09 grid.

Best,
Honghai

 

[patc]

Hi Honghai,

The nudging framework was designed to read the model history files, but unfortunately only for a single time slice per file.
So unless there are some coding changes, you will need to output your history files with only 1 time record per file.
A paper by Nick Davis, et.al. 'The impact of specified dynamics scheme on wave-mean flow dynamics, convection,
and tracer transport in CESM2(WACCN6)' carried out an experiment similar to what you described (not for 10 years though).
The results of varying the frequency of the nudging data applied showed that the time resolution of the history files needs
to be on the order of 1 hour.

--> Patrick

 

[mlague]

@Honghai, did you have any luck with the experimental setup you described? I'm starting to set up something that sounds similar - running a CESM simulation, then running another CESM simulation with the same compset with a few tweaks, nudged towards the original CESM simulation, for as many years as I can get away with given the size of the data to nudge towards (and I'm also assuming the nudged runs will be slower to run given the extra file reading).