Work with MPAS

[mbranson]

Does ADF work with CESM history file that use the MPAS dycore?

[brianpm]

Hi Mark. Right now everything in ADF assumes the data is on a regular lat/lon grid. So for SE or MPAS experiments, we either write the output to a lat/lon grid or do a regridding step before running through ADF.

Longer term we'd like to handle SE & MPAS better. There's already a tiny bit of code that knows how to handle and ncol dimension, but it's going to be a while before we can really handle irregular grids end-to-end.

[mbranson]

Thanks, Brian. So should it work if I first remap the MPAS history file to the FV grid? I tried that as well but then it crashed attempting to do the vertical interpolation to the constant pressure levels.

[brianpm]

Hmm. I thought we had dealt with this by checking for the variable PMID in the history files, and using that as the pressure field. Maybe @nusbaume can chime in.

[mbranson]

I thought the problem might be in adf_diag.py before it checks for PMID it already adds hyam,hybm,hyai,hybi to ncrcat_var_list and then later on obviously it can't find those so it crashes. I tried hacking it with this:

                        if 'PMID' in hist_file_var_list:
                            ### mdb hack ###
                            ###ncrcat_var_list = ncrcat_var_list + ",PMID"
                            ncrcat_var_list = f"{var}"+",PS,PMID"
                            print("Adding PMID to file")

but I still get errors.

[nusbaume]

Yeah, it looks like before checking the vertical coordinate type it automatically tries to add the hybrid coordinate variables, as can be seen here:

https://github.com/NCAR/ADF/blob/main/lib/adf_diag.py#L548

I think if you either just comment-out that line, or put it inside a vert_coord_type check to skip it if you have a height vertical coordinate, then it should hopefully get you past your current error. Of course if that doesn't work, or if you run into a new set of errors, then just let us know. Thanks!