[MITgcm-support] Question about generating unified .meta and .data files in Cubed-Sphere simulations

[Martin Losch]

Hi Fisher,

I recommend using (or at least trying) useSingleCpuIO = .TRUE. instead of globalFiles=.True. In my experience, this is faster and more robust if the data fits on one CPU/node.

For merging tiled mds-data, you could try to read the data into python with rdmds and write them with wrmds (both in MITgcmutils/mds.py

Martin


> On 7. Jul 2025, at 08:12, ht <ht.fisher at protonmail.com> wrote:
> 
> 
> Dear Community,
> 
> I am currently conducting simulations using the MITgcm with a Cubed-Sphere grid configuration.
> 
> In the past, I have read output data consisting of multiple .meta files (e.g., XC.001.001.meta, XC.002.001.meta, etc.) using MITgcmutils.rdmds, followed by applying my own regridding functions during post-processing.
> 
> Recently, I came across a Python package called cubedsphere(https://github.com/AaronDavidSchneider/cubedsphere), which seems to offer a more streamlined workflow for handling Cubed-Sphere grids. However, it appears to rely on the existence of unified .meta files such as XC.meta, which are not generated by default in my MITgcm outputs. Similarly, the open_mdsdataset function in xmitgcm also expects grid meta files in a unified format.
> 
> I have since learned that setting globalFiles = .TRUE. in input/data allows MITgcm to generate single .meta and .data files per timestep, even with a Cubed-Sphere grid. This seems like a promising approach for future simulations.
> 
> However, for simulations that I have already completed, the output files are split by tile (e.g., S.0000000000.001.001.meta, S.0000000000.002.001.meta, etc.).
> 
> Therefore, I would like to ask:
> 
> Is there any recommended method or tool to combine these previously generated tiled output files into unified .meta and .data files (e.g., S.0000000000.meta) after the simulation has finished?
> 
> Any advice, suggested tools, or workflows for this post-processing task would be greatly appreciated.
> 
> Thank you very much for your time and support.
> 
> Best regards,
> Fisher
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20250707/a00d5953/attachment-0001.html>