Scripting assistance for multiple runs from a reference case

[jzweifel]

Hello CESM Team!

My name is Jack Zweifel, and I am reaching out for some guidance on scripting multiple CESM2 simulations on Derecho.

In the course of my project, I've successfully implemented a large plume of sulfate aerosols off the East Coast of the United States. Following a 50-year run of my base F2000 compset, I modified the surface emissions specifier for sulfates in the CAM user namelist variable to achieve this.

Now, my goal is to run 20 three-month simulations, each branching off from my implemented aerosol plume. I've noticed options like branch or hybrid runs, as well as creating clones of my run and I believe some of these tools could be useful in scripting my runs, but I don't know which!.

What I am kindly asking is for your assistance in crafting a script to streamline this process. My scripting background is pretty limited, so I would greatly appreciate any help in creating a script suitable for execution on Derecho.

Thanks for considering my request. I'm open to any suggestions or guidance you might have, and I'm available for a discussion or meeting at your convenience!

Thanks,

Jack Zweifel

 

[nusbaume]

Hi Jack,

That's great to hear that you have successfully completed your CAM simulation! Sadly we probably won't be able to dedicate the time to help you create a script, but we can point you to some materials that may help, and can always try and debug a specific CESM-related problem with your script if you run into one.

To start with, I would recommend checking out the CESM tutorial materials, which can be found online here:

Welcome to the CESM Tutorial — CESM Tutorial

Specifically, check out the "Simple XML Modifications" section, which describes both the xmlchange tool, which you can use inside a script to change XML variables in your model case, and the differences between branch and hybrid runs, which can help you decide which type of run you'll need for your particular model experiments. Also please note that all namelist modifications can be made via the user_nl_XXX files in your case directory, so having your script add whatever namelist changes you need to those files (which can be treated as regular text files) can help you automate namelist changes.

Finally, in terms of scripting itself, Derecho supports common unix shells (e.g. BASH and TCSH), as well as other languages like Python which can also be used for scripting, so feel free to choose whichever language you would like to use.

Anyways, apologies that I can't be of more help at this stage, but hopefully that will help you get started? Again, if you run into any specific problems or issues using CESM commands like create_newcase in your scripts then feel free to let us know.

Thanks, and have a great day!

Jesse