MPAS isobaric interpolation / mpas_convert

[deniseiras]

Dear all,

Me and my colleagues from MONAN project - INPE/CPTEC (using MPAS as new dyncore) are studying the best ways to post process MPAS files in order to intepolate the vertical levels to desired levels.

We already done the alternatives:
- modify the MPAS and Registry.xml to plot desired levels - big file size!
- Run mpas_convert over the nVertLevels default levels from MPAS output
- run isobaric_interp.py to convert to desired levels - here are my collaboration - run in parallel (todo: check outuput) - MONAN_GCC_issues/435 at main · monanadmin/MONAN_GCC_issues
- run MPASSIT - but it for regional only and in Lambert projection, as far as I know.

We are wondering if is there any advances in the mpas_convert or in python code to run the same MPAS interpolation ? In the best of our internet research, we don't see any alternatives.

We are planning maybe to bring the MPAS isobaric interpolator to mpas_convert, What do you think about?


Cheers,
Denis

 

[Ming Chen]

Denis,
mpas_convert conducts horizontal interpolation to put data from MPAS mesh to regular lat-lon grids. However, this program doesn't do vertical interpolation and the data are still located at native mpas vertical levels.
Python is a robust tool that can handle both horizontal and vertical interpolation. We are working to develop a MPAS post-process package for interpolation, visualization, and analysis, etc.
If you have any progress regarding MPAS post-process tools, please post in the Forum. I am sure many MPAS users will benefit from such information. Thanks in advance.

 

[deniseiras]

Hi Chen,

we get started working in monanadmin/convert_mpas repo in github.

For now, we just adapt code to read output in grads software, but we also pretend to avance in interpolation methods. We are also investigating other software alternatives for this task.

You also can check our MPAS model updates at monanadmin/MONAN-Model repo.

Best
Denis

 

[mgduda]

Thanks for the pointers to the MONAN project repositories -- monanadmin/MONAN-Model and monanadmin/convert_mpas!

Regarding the question of adding vertical interpolation to the convert_mpas program, I wonder whether it would be preferable to keep the horizontal and vertical interpolation separate? If we add vertical interpolation to the convert_mpas program, I suppose we would then require a time-varying pressure field to be present whenever vertical interpolation is requested.

If we were to develop a separate tool for vertical interpolation, it could be independent of the horizontal mesh (e.g., either CVT mesh or lat-lon grid; or even Lambert conformal grids) and it could also support interpolation to different surfaces: isobaric, constant height surfaces, etc.